diff --git "a/train_vdr.log" "b/train_vdr.log" new file mode 100644--- /dev/null +++ "b/train_vdr.log" @@ -0,0 +1,14147 @@ +[2023-10-07 19:00:47,201][root][INFO] - args.local_rank 1 +[2023-10-07 19:00:47,202][root][INFO] - WORLD_SIZE 4 +[2023-10-07 19:00:47,223][root][INFO] - args.local_rank 0 +[2023-10-07 19:00:47,223][root][INFO] - WORLD_SIZE 4 +[2023-10-07 19:00:47,246][root][INFO] - args.local_rank 3 +[2023-10-07 19:00:47,246][root][INFO] - WORLD_SIZE 4 +[2023-10-07 19:00:47,449][root][INFO] - args.local_rank 2 +[2023-10-07 19:00:47,449][root][INFO] - WORLD_SIZE 4 +[2023-10-07 19:00:47,950][torch.distributed.distributed_c10d][INFO] - Added key: store_based_barrier_key:1 to store for rank: 3 +[2023-10-07 19:00:47,950][torch.distributed.distributed_c10d][INFO] - Added key: store_based_barrier_key:1 to store for rank: 1 +[2023-10-07 19:00:47,956][torch.distributed.distributed_c10d][INFO] - Added key: store_based_barrier_key:1 to store for rank: 2 +[2023-10-07 19:00:47,980][torch.distributed.distributed_c10d][INFO] - Added key: store_based_barrier_key:1 to store for rank: 0 +[2023-10-07 19:00:47,981][torch.distributed.distributed_c10d][INFO] - Rank 1: Completed store-based barrier for key:store_based_barrier_key:1 with 4 nodes. +[2023-10-07 19:00:47,981][torch.distributed.distributed_c10d][INFO] - Rank 3: Completed store-based barrier for key:store_based_barrier_key:1 with 4 nodes. +[2023-10-07 19:00:47,981][torch.distributed.distributed_c10d][INFO] - Rank 0: Completed store-based barrier for key:store_based_barrier_key:1 with 4 nodes. +[2023-10-07 19:00:47,982][root][INFO] - Initialized host lccpu22.cse.ust.hk as d.rank 1 on device=cuda:1, n_gpu=1, world size=4 +[2023-10-07 19:00:47,982][root][INFO] - Initialized host lccpu22.cse.ust.hk as d.rank 3 on device=cuda:3, n_gpu=1, world size=4 +[2023-10-07 19:00:47,982][root][INFO] - 16-bits training: True +[2023-10-07 19:00:47,982][root][INFO] - 16-bits training: True +[2023-10-07 19:00:47,982][root][INFO] - Initialized host lccpu22.cse.ust.hk as d.rank 0 on device=cuda:0, n_gpu=1, world size=4 +[2023-10-07 19:00:47,983][root][INFO] - 16-bits training: True +[2023-10-07 19:00:47,987][torch.distributed.distributed_c10d][INFO] - Rank 2: Completed store-based barrier for key:store_based_barrier_key:1 with 4 nodes. +[2023-10-07 19:00:47,988][root][INFO] - Initialized host lccpu22.cse.ust.hk as d.rank 2 on device=cuda:2, n_gpu=1, world size=4 +[2023-10-07 19:00:47,988][root][INFO] - 16-bits training: True +[2023-10-07 19:00:47,994][root][INFO] - train_datasets: ['marco_train_bm25'] +[2023-10-07 19:00:47,996][root][INFO] - CFG (after gpu configuration): +[2023-10-07 19:00:47,996][root][INFO] - train_datasets: ['marco_train_bm25'] +[2023-10-07 19:00:48,000][root][INFO] - dev_datasets: ['marco_dev_30neg'] +[2023-10-07 19:00:48,000][root][INFO] - train_datasets: ['marco_train_bm25'] +[2023-10-07 19:00:48,001][root][INFO] - dev_datasets: ['marco_dev_30neg'] +[2023-10-07 19:00:48,002][root][INFO] - ***** Initializing components for training ***** +[2023-10-07 19:00:48,003][root][INFO] - ***** Initializing components for training ***** +[2023-10-07 19:00:48,005][root][INFO] - dev_datasets: ['marco_dev_30neg'] +[2023-10-07 19:00:48,008][root][INFO] - ***** Initializing components for training ***** +[2023-10-07 19:00:48,013][root][INFO] - biencoder: + alias: vdr + sequence_length: 512 + dropout: 0.1 + shared_encoder: false + semiparametric: true + device: null + encoder_q: + model_id: bert-base-uncased + max_seq_len: 256 + pretrained: true + norm: false + shift_vocab_num: 1000 + prefix: encoder_q. + encoder_p: + model_id: bert-base-uncased + max_seq_len: 256 + pretrained: true + norm: false + shift_vocab_num: 1000 + prefix: encoder_p. +datasets: + nq_train: + _target_: src.utils.data.BiEncoderDataset + file: ${HOME}/data/train/nq-train.jsonl + trivia_train: + _target_: src.utils.data.BiEncoderDataset + file: ${HOME}/data/train/trivia-train.jsonl + webq_train: + _target_: src.utils.data.BiEncoderDataset + file: ${HOME}/data/train/webq-train.jsonl + dl: + _target_: src.utils.data.BiEncoderDataset + file: ${HOME}/data/train/dl_10m.jsonl + cm: + _target_: src.utils.data.BiEncoderDataset + file: ${HOME}/data/train/cm_10m.jsonl + marco_dev: + _target_: src.utils.data.BiEncoderDataset + file: ${HOME}/data/train/msmarco-dev.jsonl + marco_dev_30neg: + _target_: src.utils.data.BiEncoderDataset + file: ${HOME}/data/train/msmarco-dev-30neg.jsonl + marco_train_bm25: + _target_: src.utils.data.BiEncoderDataset + file: ${HOME}/data/train/marco_bm25_20230119.jsonl + marco_train_1000: + _target_: src.utils.data.BiEncoderDataset + file: ${HOME}/data/train/marco_train_1000.jsonl +HOME: /export/data/jzhoubu/workspace/VDR-dense +train_datasets: +- marco_train_bm25 +dev_datasets: +- marco_dev_30neg +datastore: null +output_dir: null +train_sampling_rates: null +batch_size: 64 +num_train_epochs: 20 +num_warmup_epochs: 1 +num_save_epochs: 1 +num_eval_epochs: 1 +hard_negatives: 1 +other_negatives: 0 +ret_negatives: 0 +train_insert_title: false +valid_insert_title: false +adam_eps: 1.0e-08 +adam_betas: (0.9, 0.999) +learning_rate: 2.0e-05 +max_grad_norm: 2.0 +log_batch_step: 100 +train_rolling_loss_step: 100 +weight_decay: 0.0 +sym_loss: true +do_lower_case: true +seed: 12345 +checkpoint_file_name: vdr +save_every_epoch: true +model_file: null +local_rank: 0 +local-rank: 0 +device: cuda:0 +distributed_world_size: 4 +distributed_port: null +no_cuda: false +n_gpu: 1 +fp16: true + +[2023-10-07 19:00:48,015][root][INFO] - train_datasets: ['marco_train_bm25'] +[2023-10-07 19:00:48,022][root][INFO] - dev_datasets: ['marco_dev_30neg'] +[2023-10-07 19:00:48,025][root][INFO] - ***** Initializing components for training ***** +[2023-10-07 19:00:52,021][root][INFO] - model.embedding_q.sum: -656919.625 +[2023-10-07 19:00:52,032][root][INFO] - model.embedding_p.sum: -656919.625 +[2023-10-07 19:00:53,441][root][INFO] - Initializing task/set data ['marco_train_bm25'] +[2023-10-07 19:00:53,446][root][INFO] - Initializing task/set data ['marco_train_bm25'] +[2023-10-07 19:00:53,447][root][INFO] - Initializing task/set data ['marco_train_bm25'] +[2023-10-07 19:00:53,454][root][INFO] - Initializing task/set data ['marco_train_bm25'] +[2023-10-07 19:03:20,716][root][INFO] - Load data size: 398792 +[2023-10-07 19:03:24,125][root][INFO] - Load data size: 398792 +[2023-10-07 19:03:24,128][root][INFO] - Load data size: 398792 +[2023-10-07 19:03:25,210][root][INFO] - samples_per_shard=99698, shard_start_idx=0, shard_end_idx=99698, max_iterations=1557 +[2023-10-07 19:03:25,227][root][INFO] - rank=0; Multi set data sizes [398792] +[2023-10-07 19:03:25,244][root][INFO] - rank=0; Multi set total data 398792 +[2023-10-07 19:03:25,261][root][INFO] - rank=0; Multi set sampling_rates None +[2023-10-07 19:03:25,278][root][INFO] - rank=0; Multi set max_iterations per dataset [1557] +[2023-10-07 19:03:25,289][root][INFO] - rank=0; Multi set max_iterations 1557 +[2023-10-07 19:03:25,291][root][INFO] - Total iterations per epoch=1557 +[2023-10-07 19:03:25,294][root][INFO] - Total updates=31140 +[2023-10-07 19:03:25,296][root][INFO] - Warmup updates=1557 +[2023-10-07 19:03:25,299][root][INFO] - Eval step = 1557 +[2023-10-07 19:03:25,302][root][INFO] - ***** Training ***** +[2023-10-07 19:03:25,304][root][INFO] - ***** Epoch 0 ***** +[2023-10-07 19:03:25,310][root][INFO] - rank=0; Iteration start +[2023-10-07 19:03:25,312][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 19:03:25,315][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 19:03:25,322][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-07 19:03:28,220][root][INFO] - samples_per_shard=99698, shard_start_idx=199396, shard_end_idx=299094, max_iterations=1557 +[2023-10-07 19:03:28,223][root][INFO] - rank=2; Multi set data sizes [398792] +[2023-10-07 19:03:28,226][root][INFO] - rank=2; Multi set total data 398792 +[2023-10-07 19:03:28,228][root][INFO] - rank=2; Multi set sampling_rates None +[2023-10-07 19:03:28,231][root][INFO] - rank=2; Multi set max_iterations per dataset [1557] +[2023-10-07 19:03:28,233][root][INFO] - rank=2; Multi set max_iterations 1557 +[2023-10-07 19:03:28,236][root][INFO] - Total iterations per epoch=1557 +[2023-10-07 19:03:28,239][root][INFO] - Total updates=31140 +[2023-10-07 19:03:28,241][root][INFO] - Warmup updates=1557 +[2023-10-07 19:03:28,244][root][INFO] - Eval step = 1557 +[2023-10-07 19:03:28,246][root][INFO] - ***** Training ***** +[2023-10-07 19:03:28,249][root][INFO] - samples_per_shard=99698, shard_start_idx=299094, shard_end_idx=398792, max_iterations=1557 +[2023-10-07 19:03:28,249][root][INFO] - ***** Epoch 0 ***** +[2023-10-07 19:03:28,251][root][INFO] - rank=3; Multi set data sizes [398792] +[2023-10-07 19:03:28,254][root][INFO] - rank=3; Multi set total data 398792 +[2023-10-07 19:03:28,254][root][INFO] - rank=2; Iteration start +[2023-10-07 19:03:28,256][root][INFO] - rank=3; Multi set sampling_rates None +[2023-10-07 19:03:28,257][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 19:03:28,259][root][INFO] - rank=3; Multi set max_iterations per dataset [1557] +[2023-10-07 19:03:28,260][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 19:03:28,261][root][INFO] - rank=3; Multi set max_iterations 1557 +[2023-10-07 19:03:28,264][root][INFO] - Total iterations per epoch=1557 +[2023-10-07 19:03:28,266][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-07 19:03:28,266][root][INFO] - Total updates=31140 +[2023-10-07 19:03:28,269][root][INFO] - Warmup updates=1557 +[2023-10-07 19:03:28,271][root][INFO] - Eval step = 1557 +[2023-10-07 19:03:28,274][root][INFO] - ***** Training ***** +[2023-10-07 19:03:28,276][root][INFO] - ***** Epoch 0 ***** +[2023-10-07 19:03:28,282][root][INFO] - rank=3; Iteration start +[2023-10-07 19:03:28,284][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 19:03:28,287][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 19:03:28,293][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-07 19:03:28,328][root][INFO] - Load data size: 398792 +[2023-10-07 19:03:32,523][root][INFO] - samples_per_shard=99698, shard_start_idx=99698, shard_end_idx=199396, max_iterations=1557 +[2023-10-07 19:03:32,525][root][INFO] - rank=1; Multi set data sizes [398792] +[2023-10-07 19:03:32,528][root][INFO] - rank=1; Multi set total data 398792 +[2023-10-07 19:03:32,530][root][INFO] - rank=1; Multi set sampling_rates None +[2023-10-07 19:03:32,533][root][INFO] - rank=1; Multi set max_iterations per dataset [1557] +[2023-10-07 19:03:32,535][root][INFO] - rank=1; Multi set max_iterations 1557 +[2023-10-07 19:03:32,538][root][INFO] - Total iterations per epoch=1557 +[2023-10-07 19:03:32,541][root][INFO] - Total updates=31140 +[2023-10-07 19:03:32,543][root][INFO] - Warmup updates=1557 +[2023-10-07 19:03:32,546][root][INFO] - Eval step = 1557 +[2023-10-07 19:03:32,549][root][INFO] - ***** Training ***** +[2023-10-07 19:03:32,551][root][INFO] - ***** Epoch 0 ***** +[2023-10-07 19:03:32,557][root][INFO] - rank=1; Iteration start +[2023-10-07 19:03:32,559][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 19:03:32,562][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 19:03:32,569][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-07 19:03:33,768][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 2.36 | max: 9.16 | min: 0.23 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 3.07 | max: 10.53 | min: 0.61 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how do totally deaf people communicate with people who do not sign [SEP] ### +### [P_TEXT]: [CLS] no, not all hearing people can either! reading lips is not something specifically ### +### that deaf people can do, it is something that deaf people have done and sometimes need to do in ### +### order to exist and participate in a conversation where signing is not an option. most people can ### +### not read lips, it is also virtually impossible to read lips accurately 100 %. [SEP] ### +### ======================================= h_v_q | Gates: 29522 ======================================= ### +### ('deaf', 0, 0) ('and', 1, 9) (',', 2, 2) ('communicate', 3, 4438) ('.', 4, 4) ('the', 5, 8) ### +### ('##tooth', 6, 582) ('##ciency', 7, 4862) ('##tructing', 8, 7255) ('##dication', 9, 1763) ### +### ('a', 10, 14) ('of', 11, 18) ('communicated', 12, 2625) ('##ically', 13, 560) ('who', 14, 798) ### +### ('##ential', 15, 301) ('how', 16, 7632) ('totally', 17, 23122) ('to', 18, 30) ('##trust', 19, 618) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('deaf', 0, 0) ('lips', 3820, 1) (',', 2, 2) ('read', 818, 3) ('.', 4, 4) ('accurately', 7340, 5) ### +### ('signing', 1082, 6) ('##mity', 1437, 7) ('the', 5, 8) ('and', 1, 9) ('not', 72, 10) ### +### ('hearing', 37, 11) ('reading', 8069, 12) ('ipod', 1412, 13) ('a', 10, 14) ('mouths', 9549, 15) ### +### ('where', 610, 16) ('oder', 2246, 17) ('of', 11, 18) ('option', 21151, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('deaf', 0, 0) (',', 2, 2) ('.', 4, 4) ('and', 1, 9) ('the', 5, 8) ('a', 10, 14) ('of', 11, 18) ### +### ('hearing', 37, 11) ('to', 18, 30) ('in', 20, 38) ('not', 72, 10) ('that', 22, 56) ('is', 38, 27) ### +### ('-', 29, 44) ('##tooth', 6, 582) ('was', 23, 114) ('with', 28, 117) ('##ishly', 47, 76) ### +### ('his', 40, 92) ('"', 85, 46) ### +############################################################################################################ +[2023-10-07 19:03:33,768][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:03:33,768][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:03:34,247][root][INFO] - Epoch: 0: Step: 1/1557, loss[v]=799.714966, lr=0.000000, acc@1[1]=108.0/256=0.421875, acc@1[2]=38.5/256=0.150390625 +[2023-10-07 19:04:49,688][root][INFO] - Train batch 100 +[2023-10-07 19:04:49,689][root][INFO] - Avg. loss per last 100 batches: 604.066817 +[2023-10-07 19:04:50,379][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 2.20 | max: 9.02 | min: 0.20 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 2.51 | max: 10.25 | min: 0.38 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] function of ribosomes in eukaryotes [SEP] ### +### [P_TEXT]: [CLS] ribosomes are found in both prokaryotes and eukaryotes. they are made up of ### +### proteins and rna molecules called subunits. the predominant function of ribosomes is the ### +### translation of messenger rna ( mrna ) into proteins. both prokaryotic and eukaryotic cells have ### +### many ribosomes, but the ribosomes found in eukaryotic cells are larger and more sophisticated than ### +### those of the prokaryotic cell. nitiation of protein synthesis. ribosomes synthesize proteins in ### +### both prokaryotic and eukaryotic cells, but the process is initiated differently in each cell type. ### +### in eukaryotic cells, the starter amino acid is methionine rather than the n - formylmethionine used ### +### by prokaryotic cells. [SEP] ### +### ======================================= h_v_q | Gates: 29522 ======================================= ### +### ('##oso', 0, 5) ('.', 1, 11) ('the', 2, 16) ('and', 3, 14) (',', 4, 10) ('in', 5, 21) ('of', 6, 27) ### +### ('to', 7, 28) ('a', 8, 25) ('##tes', 9, 73) ('##mes', 10, 43) ('##kar', 11, 2) ('was', 12, 72) ### +### ('for', 13, 61) ('eu', 14, 0) ('at', 15, 58) ('her', 16, 120) ('with', 17, 47) ('an', 18, 54) ### +### ('that', 19, 57) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('eu', 14, 0) ('rna', 634, 1) ('##kar', 11, 2) ('mrna', 12007, 3) ('##met', 22111, 4) ### +### ('##oso', 0, 5) ('met', 535, 6) ('messenger', 154, 7) ('both', 158, 8) ('##oni', 14967, 9) ### +### (',', 4, 10) ('.', 1, 11) ('found', 647, 12) ('##hi', 26346, 13) ('and', 3, 14) ('##tia', 653, 15) ### +### ('the', 2, 16) ('but', 38, 17) ('differently', 14252, 18) ('synthesis', 5502, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##oso', 0, 5) ('.', 1, 11) (',', 4, 10) ('and', 3, 14) ('the', 2, 16) ('eu', 14, 0) ### +### ('##kar', 11, 2) ('in', 5, 21) ('a', 8, 25) ('of', 6, 27) ('to', 7, 28) ('but', 38, 17) ### +### ('##mes', 10, 43) ('##yo', 25, 29) ('rib', 35, 24) ('is', 22, 34) ('messenger', 154, 7) ### +### ('with', 17, 47) ('##tes', 9, 73) ('-', 24, 46) ### +############################################################################################################ +[2023-10-07 19:04:50,380][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:04:50,380][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:04:50,785][root][INFO] - Epoch: 0: Step: 101/1557, loss[v]=288.372772, lr=0.000001, acc@1[1]=112.0/256=0.4375, acc@1[2]=70.0/256=0.2734375 +[2023-10-07 19:06:07,893][root][INFO] - Train batch 200 +[2023-10-07 19:06:07,894][root][INFO] - Avg. loss per last 100 batches: 171.029334 +[2023-10-07 19:06:08,634][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 2.02 | max: 9.62 | min: 0.16 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 2.06 | max: 10.14 | min: 0.21 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] is toothpaste a convenience good [SEP] ### +### [P_TEXT]: [CLS] to choose the whitening toothpaste thatas best for you, you should first know what ### +### whitening toothpaste actually is. this post about teeth whitening reviews the basics of ahow to get ### +### your teeth whitea and tips keep them white. [SEP] ### +### ======================================= h_v_q | Gates: 29522 ======================================= ### +### ('convenience', 0, 6231) ('tooth', 1, 0) ('the', 2, 3) (',', 3, 1) ('and', 4, 4) ('.', 5, 2) ### +### ('a', 6, 7) ('to', 7, 12) ('of', 8, 9) ('in', 9, 11) ('##pas', 10, 17) ('-', 11, 14) ('is', 12, 19) ### +### ('his', 13, 18) ('teeth', 14, 6) ('that', 15, 13) ('was', 16, 15) ('with', 17, 16) ('"', 18, 26) ### +### ('an', 19, 21) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('tooth', 1, 0) (',', 3, 1) ('.', 5, 2) ('the', 2, 3) ('and', 4, 4) ('white', 2974, 5) ### +### ('teeth', 14, 6) ('a', 6, 7) ('best', 2140, 8) ('of', 8, 9) ('##ning', 1005, 10) ('in', 9, 11) ### +### ('to', 7, 12) ('that', 15, 13) ('-', 11, 14) ('was', 16, 15) ('with', 17, 16) ('##pas', 10, 17) ### +### ('his', 13, 18) ('is', 12, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('tooth', 1, 0) (',', 3, 1) ('the', 2, 3) ('.', 5, 2) ('and', 4, 4) ('a', 6, 7) ('of', 8, 9) ### +### ('in', 9, 11) ('to', 7, 12) ('teeth', 14, 6) ('##pas', 10, 17) ('-', 11, 14) ('that', 15, 13) ### +### ('was', 16, 15) ('is', 12, 19) ('his', 13, 18) ('with', 17, 16) ('an', 19, 21) ('"', 18, 26) ### +### ('(', 20, 25) ### +############################################################################################################ +[2023-10-07 19:06:08,634][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:06:08,634][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:06:09,059][root][INFO] - Epoch: 0: Step: 201/1557, loss[v]=96.330345, lr=0.000003, acc@1[1]=113.0/256=0.44140625, acc@1[2]=145.5/256=0.568359375 +[2023-10-07 19:07:26,340][root][INFO] - Train batch 300 +[2023-10-07 19:07:26,341][root][INFO] - Avg. loss per last 100 batches: 57.374392 +[2023-10-07 19:07:27,050][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 1.77 | max: 9.77 | min: 0.12 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 1.90 | max: 10.48 | min: 0.16 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is preference [SEP] ### +### [P_TEXT]: [CLS] a simple example of a preference order over three goods. in economics and other ### +### social sciences, preference is the ordering of alternatives based on their relative utility, a ### +### process which results in an optimal choice ( whether real or theoretical ). [SEP] ### +### ======================================= h_v_q | Gates: 29522 ======================================= ### +### ('.', 0, 1) ('preference', 1, 0) (',', 2, 2) ('the', 3, 8) ('and', 4, 3) ('a', 5, 13) ('of', 6, 11) ### +### ('in', 7, 19) ('to', 8, 23) ('-', 9, 30) ('preferred', 10, 108) ('that', 11, 20) ### +### ('preferences', 12, 51) ('was', 13, 34) ('an', 14, 33) ('##ential', 15, 722) (';', 16, 38) ### +### ('##s', 17, 66) ('is', 18, 29) ('his', 19, 36) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('preference', 1, 0) ('.', 0, 1) (',', 2, 2) ('and', 4, 3) ('order', 5503, 4) ('goods', 12641, 5) ### +### ('alternatives', 4241, 6) ('choice', 38, 7) ('the', 3, 8) ('economics', 20795, 9) ### +### ('optimal', 2776, 10) ('of', 6, 11) ('process', 12652, 12) ('a', 5, 13) ('ordering', 10374, 14) ### +### ('example', 10705, 15) ('examples', 7511, 16) ('social', 5014, 17) ('three', 1168, 18) ### +### ('in', 7, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('preference', 1, 0) ('.', 0, 1) (',', 2, 2) ('and', 4, 3) ('the', 3, 8) ('a', 5, 13) ('of', 6, 11) ### +### ('in', 7, 19) ('to', 8, 23) ('that', 11, 20) ('-', 9, 30) ('choice', 38, 7) ('(', 23, 26) ### +### ('is', 18, 29) ('was', 13, 34) ('an', 14, 33) ('his', 19, 36) (';', 16, 38) ('with', 25, 32) ### +### ('preferences', 12, 51) ### +############################################################################################################ +[2023-10-07 19:07:27,050][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:07:27,050][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:07:27,475][root][INFO] - Epoch: 0: Step: 301/1557, loss[v]=32.964371, lr=0.000004, acc@1[1]=126.0/256=0.4921875, acc@1[2]=197.5/256=0.771484375 +[2023-10-07 19:08:44,618][root][INFO] - Train batch 400 +[2023-10-07 19:08:44,619][root][INFO] - Avg. loss per last 100 batches: 28.009234 +[2023-10-07 19:08:45,340][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 1.06 | max: 9.75 | min: 0.04 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 1.47 | max: 10.92 | min: 0.09 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a thoracic vent used for [SEP] ### +### [P_TEXT]: [CLS] small - bore catheters and heimlich valves have been successfully used in. the ### +### treatment of pneumothoraces in several studies. 4, 5 the thoracic vent. ( see figure 1 ) is a ### +### minimally invasive device for the treatment of. pneumothorax. it consists of a polyurethane ### +### catheter connected to a. plastic chamber containing a one - way valve. [SEP] ### +### ======================================= h_v_q | Gates: 29522 ======================================= ### +### ('vent', 0, 5) ('##ac', 1, 4) ('thor', 2, 19) ('##ic', 3, 12) ('.', 4, 15) ('used', 5, 43) ### +### (',', 6, 14) ('and', 7, 17) ('a', 8, 28) ('the', 9, 20) ('use', 10, 59) ('is', 11, 49) ### +### ('of', 12, 37) ('##oc', 13, 50) ('in', 14, 18) ('vents', 15, 115) ('for', 16, 56) ('an', 17, 61) ### +### ('jet', 18, 294) ('##tic', 19, 77) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('bore', 3087, 0) ('##race', 24208, 1) ('##lich', 21329, 2) ('##otho', 28244, 3) ('##ac', 1, 4) ### +### ('vent', 0, 5) ('##het', 7627, 6) ('valve', 4271, 7) ('cat', 430, 8) ('##im', 12159, 9) ### +### ('valves', 3299, 10) ('##than', 22386, 11) ('##ic', 3, 12) ('plastic', 1561, 13) (',', 6, 14) ### +### ('.', 4, 15) ('(', 33, 16) ('and', 7, 17) ('in', 14, 18) ('thor', 2, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##ac', 1, 4) ('vent', 0, 5) ('##ic', 3, 12) ('thor', 2, 19) ('.', 4, 15) (',', 6, 14) ### +### ('and', 7, 17) ('the', 9, 20) ('a', 8, 28) ('used', 5, 43) ('in', 14, 18) ('of', 12, 37) ### +### ('(', 33, 16) ('to', 20, 36) ('use', 10, 59) ('-', 25, 34) ('is', 11, 49) ('##oc', 13, 50) ### +### ('for', 16, 56) ('an', 17, 61) ### +############################################################################################################ +[2023-10-07 19:08:45,340][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:08:45,341][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:08:45,762][root][INFO] - Epoch: 0: Step: 401/1557, loss[v]=19.865465, lr=0.000005, acc@1[1]=121.0/256=0.47265625, acc@1[2]=223.0/256=0.87109375 +[2023-10-07 19:10:02,736][root][INFO] - Train batch 500 +[2023-10-07 19:10:02,737][root][INFO] - Avg. loss per last 100 batches: 10.761707 +[2023-10-07 19:10:03,435][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.27 | max: 7.41 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.56 | max: 9.81 | min: 0.01 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] civil servants definition [SEP] ### +### [P_TEXT]: [CLS] definition of'civil servant '. civil servant. a civil servant is a person who works ### +### in the civil service in britain and some other countries, or for the local, state, or federal ### +### government in the united states.... two senior civil servants. [SEP] ### +### ======================================= h_v_q | Gates: 29522 ======================================= ### +### ('definition', 0, 6) ('civil', 1, 1) ('servants', 2, 5) ('definitions', 3, 10) ('.', 4, 7) ### +### ('postal', 5, 43) ('servant', 6, 0) (',', 7, 12) (';', 8, 21) ('the', 9, 22) ('naval', 10, 123) ### +### ('a', 11, 15) ('civilian', 12, 53) ('households', 13, 284) ('employees', 14, 370) ('and', 15, 26) ### +### ('military', 16, 48) ('police', 17, 69) ('class', 18, 102) ('howard', 19, 95) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('servant', 6, 0) ('civil', 1, 1) ('service', 128, 2) ('person', 2118, 3) ('or', 556, 4) ### +### ('servants', 2, 5) ('definition', 0, 6) ('.', 4, 7) ('works', 6057, 8) ('two', 1378, 9) ### +### ('definitions', 3, 10) ('services', 22, 11) (',', 7, 12) ("'", 15057, 13) ('work', 3201, 14) ### +### ('a', 11, 15) ('senior', 4556, 16) ('britain', 7777, 17) ('state', 7996, 18) ('in', 444, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('civil', 1, 1) ('definition', 0, 6) ('servant', 6, 0) ('servants', 2, 5) ('.', 4, 7) ### +### ('definitions', 3, 10) (',', 7, 12) ('postal', 5, 43) (';', 8, 21) ('the', 9, 22) ('a', 11, 15) ### +### ('services', 22, 11) ('service', 128, 2) ('and', 15, 26) ('civilian', 12, 53) ('military', 16, 48) ### +### ('wingspan', 21, 49) ('police', 17, 69) ('naval', 10, 123) ('or', 556, 4) ### +############################################################################################################ +[2023-10-07 19:10:03,435][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:10:03,435][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:10:03,860][root][INFO] - Epoch: 0: Step: 501/1557, loss[v]=3.742833, lr=0.000006, acc@1[1]=158.5/256=0.619140625, acc@1[2]=233.0/256=0.91015625 +[2023-10-07 19:11:21,036][root][INFO] - Train batch 600 +[2023-10-07 19:11:21,037][root][INFO] - Avg. loss per last 100 batches: 1.523697 +[2023-10-07 19:11:21,740][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29520.5/29522=99.99% | mean: 0.06 | max: 4.91 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 6.65 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the tallest building in the world [SEP] ### +### [P_TEXT]: [CLS] the shanghai tower, with 121 floors that stand 2, 073 feet tall, is the second ### +### tallest building in the world. the construction of shanghai tower began in 2006 and took eight ### +### years of work before its completion in march of 2014. it is home to the world's highest observation ### +### deck. [SEP] ### +### ======================================= h_v_q | Gates: 29484 ======================================= ### +### ('tallest', 0, 2) ('world', 1, 25) ('building', 2, 9) ('.', 3, 17) ('buildings', 4, 18) ### +### ('taller', 5, 21) ('top', 6, 26) ('nato', 7, 65) ('tower', 8, 3) ('highest', 9, 8) ### +### ('largest', 10, 45) ('anger', 11, 89) ('pan', 12, 143) ('tall', 13, 4) ('construction', 14, 5) ### +### ('atop', 15, 126) ('last', 16, 31) ('hotel', 17, 253) ('global', 18, 193) (',', 19, 39) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('shanghai', 303, 0) ('deck', 46, 1) ('tallest', 0, 2) ('tower', 8, 3) ('tall', 13, 4) ### +### ('construction', 14, 5) ('floors', 372, 6) ('observation', 10134, 7) ('highest', 9, 8) ### +### ('building', 2, 9) ('floor', 112, 10) ('2', 1526, 11) ('china', 39, 12) ('second', 77, 13) ### +### ('stand', 6413, 14) ('stood', 242, 15) ('towers', 99, 16) ('.', 3, 17) ('buildings', 4, 18) ### +### ('decks', 18951, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('tallest', 0, 2) ('building', 2, 9) ('world', 1, 25) ('.', 3, 17) ('tower', 8, 3) ### +### ('buildings', 4, 18) ('tall', 13, 4) ('highest', 9, 8) ('taller', 5, 21) ('construction', 14, 5) ### +### ('deck', 46, 1) ('top', 6, 26) ('largest', 10, 45) ('nato', 7, 65) ('shanghai', 303, 0) ### +### ('china', 39, 12) ('kent', 20, 27) ('last', 16, 31) (',', 19, 39) ('anger', 11, 89) ### +############################################################################################################ +[2023-10-07 19:11:21,741][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:11:21,741][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:11:22,163][root][INFO] - Epoch: 0: Step: 601/1557, loss[v]=0.453450, lr=0.000008, acc@1[1]=218.5/256=0.853515625, acc@1[2]=239.0/256=0.93359375 +[2023-10-07 19:12:38,308][root][INFO] - Train batch 700 +[2023-10-07 19:12:38,309][root][INFO] - Avg. loss per last 100 batches: 0.369366 +[2023-10-07 19:12:39,005][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29517.5/29522=99.98% | mean: 0.04 | max: 4.97 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.11 | max: 6.23 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] is geography science [SEP] ### +### [P_TEXT]: [CLS] the scientific process of geography. geography is considered a science and thus ### +### also uses the scientific method for data collection, analysis, and interpretation. there is no true ### +### definition of the scientific method because it varies so much between scientific disciplines. [SEP] ### +### ======================================= h_v_q | Gates: 29413 ======================================= ### +### ('geography', 0, 0) ('science', 1, 2) ('.', 2, 11) ('sciences', 3, 124) ('map', 4, 21) ### +### ('clay', 5, 1043) ('geo', 6, 12) ('maps', 7, 10) ('elliott', 8, 167) ('research', 9, 35) ### +### ('pan', 10, 14) ('wingspan', 11, 20) ('archaeology', 12, 76) ('tobacco', 13, 3715) ### +### ('survey', 14, 116) ('topography', 15, 371) ('struggled', 16, 27) ('simon', 17, 179) ### +### ('tennis', 18, 1846) ('kent', 19, 360) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('geography', 0, 0) ('scientific', 118, 1) ('science', 1, 2) ('process', 6000, 3) ### +### ('collection', 6206, 4) ('data', 3892, 5) ('method', 4165, 6) ('true', 15712, 7) ### +### ('definition', 13467, 8) ('go', 179, 9) ('maps', 7, 10) ('.', 2, 11) ('geo', 6, 12) ### +### ('so', 8597, 13) ('pan', 10, 14) ('processes', 8842, 15) ('interpretation', 17795, 16) ### +### ('analysis', 5630, 17) ('anger', 26, 18) ('theory', 49, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('geography', 0, 0) ('science', 1, 2) ('.', 2, 11) ('maps', 7, 10) ('geo', 6, 12) ('map', 4, 21) ### +### ('pan', 10, 14) ('scientific', 118, 1) ('wingspan', 11, 20) ('research', 9, 35) ### +### ('sciences', 3, 124) ('struggled', 16, 27) ('anger', 26, 18) ('leaving', 31, 23) ### +### ('archaeology', 12, 76) ('theory', 49, 19) ('scientists', 25, 33) ('elliott', 8, 167) ### +### ('go', 179, 9) ('survey', 14, 116) ### +############################################################################################################ +[2023-10-07 19:12:39,005][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:12:39,005][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:12:39,429][root][INFO] - Epoch: 0: Step: 701/1557, loss[v]=0.286910, lr=0.000009, acc@1[1]=226.5/256=0.884765625, acc@1[2]=242.0/256=0.9453125 +[2023-10-07 19:13:55,905][root][INFO] - Train batch 800 +[2023-10-07 19:13:55,906][root][INFO] - Avg. loss per last 100 batches: 0.266873 +[2023-10-07 19:13:56,638][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29513.3/29522=99.97% | mean: 0.03 | max: 4.57 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.4/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.12 | max: 6.24 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] when did hotel california first come out [SEP] ### +### [P_TEXT]: [CLS] the ambivalence of that stance came out in the song. the hotel california album was ### +### released on december 8, 1976, and went to number one in early 1977, eventually selling more than ### +### ten million copies. hotel california the song was released as a single on march 12, 1977, and also ### +### hit number one, going gold. [SEP] ### +### ======================================= h_v_q | Gates: 29325 ======================================= ### +### ('california', 0, 1) ('hotel', 1, 2) ('come', 2, 121) ('out', 3, 14) ('.', 4, 82) ### +### ('first', 5, 2239) ('came', 6, 35) ('hospital', 7, 54) ('virginia', 8, 204) ('alaska', 9, 149) ### +### ('finally', 10, 25) ('massachusetts', 11, 164) ('indiana', 12, 922) ('kent', 13, 55) ### +### ('mexico', 14, 767) ('hotels', 15, 18) ('club', 16, 1800) ('nevada', 17, 634) ('coming', 18, 3011) ### +### ('utah', 19, 29) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('stance', 11521, 0) ('california', 0, 1) ('hotel', 1, 2) ('gold', 1181, 3) ('##vale', 13363, 4) ### +### ('song', 2714, 5) ('album', 7760, 6) ('hit', 283, 7) ('early', 40, 8) ('1976', 5132, 9) ### +### ('selling', 5071, 10) ('million', 4212, 11) ('am', 3424, 12) ('1977', 4484, 13) ('out', 3, 14) ### +### ('december', 889, 15) ('2000', 136, 16) ('ten', 1751, 17) ('hotels', 15, 18) ('hitting', 1296, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('california', 0, 1) ('hotel', 1, 2) ('out', 3, 14) ('came', 6, 35) ('come', 2, 121) ### +### ('hotels', 15, 18) ('finally', 10, 25) ('.', 4, 82) ('hospital', 7, 54) ('early', 40, 8) ### +### ('utah', 19, 29) ('kent', 13, 55) ('virginia', 8, 204) ('alaska', 9, 149) ### +### ('massachusetts', 11, 164) ('film', 36, 37) ('hit', 283, 7) ('pennsylvania', 26, 58) ### +### ('2000', 136, 16) ('clark', 20, 110) ### +############################################################################################################ +[2023-10-07 19:13:56,638][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:13:56,638][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:13:57,044][root][INFO] - Epoch: 0: Step: 801/1557, loss[v]=0.270031, lr=0.000010, acc@1[1]=225.5/256=0.880859375, acc@1[2]=242.5/256=0.947265625 +[2023-10-07 19:15:14,161][root][INFO] - Train batch 900 +[2023-10-07 19:15:14,161][root][INFO] - Avg. loss per last 100 batches: 0.239704 +[2023-10-07 19:15:14,884][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29513.8/29522=99.97% | mean: 0.03 | max: 4.68 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.11 | max: 5.91 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is similar to clotrimazole lozenges [SEP] ### +### [P_TEXT]: [CLS] clotrimazole lozenges are dissolved slowly in the mouth to prevent and treat ### +### thrush. thrush, also called candidiasis or white mouth, is a fungus infection of the mouth and ### +### throat. this medicine may also be used for other problems as determined by your doctor. [SEP] ### +### ======================================= h_v_q | Gates: 29357 ======================================= ### +### ('##zen', 0, 5) ('lo', 1, 3) ('##az', 2, 8) ('##rim', 3, 11) ('##ges', 4, 21) ('##ole', 5, 10) ### +### ('similar', 6, 1874) ('similarity', 7, 6732) ('##ot', 8, 17) ('cl', 9, 18) ### +### ('similarities', 10, 6337) ('.', 11, 59) ('##sen', 12, 35) ('kent', 13, 93) ('anger', 14, 48) ### +### ('##ge', 15, 45) ('sam', 16, 173) ('preston', 17, 157) ('digital', 18, 230) ('wingspan', 19, 27) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##sh', 18541, 0) ('##sis', 7203, 1) ('candi', 24846, 2) ('lo', 1, 3) ('mouth', 12321, 4) ### +### ('##zen', 0, 5) ('##dia', 12759, 6) ('dissolved', 2190, 7) ('##az', 2, 8) ('throat', 11403, 9) ### +### ('##ole', 5, 10) ('##rim', 3, 11) ('fungus', 15897, 12) ('medicine', 6506, 13) ('thru', 29277, 14) ### +### ('white', 6777, 15) ('treat', 11971, 16) ('##ot', 8, 17) ('cl', 9, 18) ('infection', 15540, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##zen', 0, 5) ('lo', 1, 3) ('##az', 2, 8) ('##rim', 3, 11) ('##ole', 5, 10) ('##ges', 4, 21) ### +### ('##ot', 8, 17) ('cl', 9, 18) ('.', 11, 59) ('##sen', 12, 35) ('wingspan', 19, 27) ('##ge', 15, 45) ### +### ('anger', 14, 48) ('kent', 13, 93) ('similar', 6, 1874) ('archaeology', 24, 53) ('harsh', 46, 36) ### +### ('struggled', 26, 63) ('son', 56, 33) ('frederick', 29, 77) ### +############################################################################################################ +[2023-10-07 19:15:14,885][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:15:14,885][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:15:15,307][root][INFO] - Epoch: 0: Step: 901/1557, loss[v]=0.183693, lr=0.000012, acc@1[1]=229.0/256=0.89453125, acc@1[2]=248.5/256=0.970703125 +[2023-10-07 19:16:32,136][root][INFO] - Train batch 1000 +[2023-10-07 19:16:32,137][root][INFO] - Avg. loss per last 100 batches: 0.224666 +[2023-10-07 19:16:32,867][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29504.9/29522=99.94% | mean: 0.03 | max: 4.53 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.12 | max: 5.89 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how long is rn program [SEP] ### +### [P_TEXT]: [CLS] time frames. an lvn who went straight into an associate degree program after ### +### graduation might become an rn in as little as three years of full - time school - - one year to ### +### complete prerequisites and two years for the rn program itself. [SEP] ### +### ======================================= h_v_q | Gates: 29307 ======================================= ### +### ('rn', 0, 0) ('program', 1, 7) ('long', 2, 47) ('programs', 3, 32) ('minutes', 4, 53) ### +### ('programme', 5, 959) ('weeks', 6, 24) ('minute', 7, 59) ('.', 8, 229) ('is', 9, 6834) ### +### ('naval', 10, 28) ('longest', 11, 10000) ('days', 12, 91) ('length', 13, 10749) ('par', 14, 152) ### +### ('moment', 15, 105) ('project', 16, 1850) ('programmes', 17, 5323) ('short', 18, 810) ### +### ('week', 19, 26) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('rn', 0, 0) ('frames', 8303, 1) ('associate', 645, 2) ('l', 2531, 3) ('graduation', 8277, 4) ### +### ('##vn', 28484, 5) ('frame', 1252, 6) ('program', 1, 7) ('time', 35, 8) ('years', 124, 9) ### +### ('might', 8093, 10) ('degree', 1973, 11) ('##quisite', 22826, 12) ('school', 369, 13) ### +### ('full', 6881, 14) ('graduate', 1140, 15) ('year', 1535, 16) ('wingspan', 188, 17) ### +### ('two', 2831, 18) ('3', 909, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('rn', 0, 0) ('program', 1, 7) ('long', 2, 47) ('programs', 3, 32) ('weeks', 6, 24) ### +### ('minutes', 4, 53) ('naval', 10, 28) ('minute', 7, 59) ('time', 35, 8) ('.', 8, 229) ### +### ('week', 19, 26) ('programme', 5, 959) ('years', 124, 9) ('associate', 645, 2) ('days', 12, 91) ### +### ('anger', 23, 42) ('rotor', 27, 46) ('moment', 15, 105) ('centuries', 78, 30) ('harsh', 79, 31) ### +############################################################################################################ +[2023-10-07 19:16:32,868][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:16:32,868][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:16:33,296][root][INFO] - Epoch: 0: Step: 1001/1557, loss[v]=0.230787, lr=0.000013, acc@1[1]=228.5/256=0.892578125, acc@1[2]=242.5/256=0.947265625 +[2023-10-07 19:17:50,574][root][INFO] - Train batch 1100 +[2023-10-07 19:17:50,575][root][INFO] - Avg. loss per last 100 batches: 0.209086 +[2023-10-07 19:17:51,291][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29495.8/29522=99.91% | mean: 0.02 | max: 4.52 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.5/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.11 | max: 5.77 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how does creatine help with strength [SEP] ### +### [P_TEXT]: [CLS] the iranian researchers concluded that since myostatin levels were lower in the ### +### subjects taking creatine, one way that creatine may work to increase muscle size and strength is by ### +### reducing myostatin levels, which reduces the limitation that this protein places on muscle growth. ### +### [SEP] ### +### ======================================= h_v_q | Gates: 29133 ======================================= ### +### ('##tine', 0, 1) ('cr', 1, 20) ('##ea', 2, 8) ('help', 3, 102) ('strength', 4, 5) ('.', 5, 143) ### +### ('strong', 6, 58) ('helping', 7, 4659) ('helped', 8, 8647) ('how', 9, 8043) ('power', 10, 118) ### +### ('force', 11, 6706) ('does', 12, 9569) ('assistance', 13, 18338) ('anger', 14, 28) ### +### ('hall', 15, 3021) ('helps', 16, 10322) ('kent', 17, 46) ('support', 18, 8083) ('marie', 19, 234) ### +### ======================================= h_v_p | Gates: 29520 ======================================= ### +### ('##tin', 1117, 0) ('##tine', 0, 1) ('protein', 32, 2) ('muscle', 143, 3) ('iran', 293, 4) ### +### ('strength', 4, 5) ('growth', 495, 6) ('may', 4418, 7) ('##ea', 2, 8) ('size', 62, 9) ### +### ('iranian', 12196, 10) ('##osta', 26418, 11) ('lower', 1258, 12) ('reduce', 2575, 13) ### +### ('reducing', 3720, 14) ('level', 1320, 15) ('reduced', 3830, 16) ('my', 423, 17) ### +### ('proteins', 830, 18) ('increase', 4840, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##tine', 0, 1) ('##ea', 2, 8) ('strength', 4, 5) ('cr', 1, 20) ('help', 3, 102) ('strong', 6, 58) ### +### ('.', 5, 143) ('protein', 32, 2) ('anger', 14, 28) ('size', 62, 9) ('muscle', 143, 3) ### +### ('power', 10, 118) ('kent', 17, 46) ('iran', 293, 4) ('##tin', 1117, 0) ('wingspan', 85, 26) ### +### ('growth', 495, 6) ('harsh', 53, 32) ('my', 423, 17) ('archaeology', 33, 67) ### +############################################################################################################ +[2023-10-07 19:17:51,291][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:17:51,291][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:17:51,714][root][INFO] - Epoch: 0: Step: 1101/1557, loss[v]=0.138835, lr=0.000014, acc@1[1]=237.5/256=0.927734375, acc@1[2]=248.5/256=0.970703125 +[2023-10-07 19:19:08,459][root][INFO] - Train batch 1200 +[2023-10-07 19:19:08,460][root][INFO] - Avg. loss per last 100 batches: 0.199725 +[2023-10-07 19:19:09,185][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29500.2/29522=99.93% | mean: 0.02 | max: 4.61 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.5/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.13 | max: 6.04 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what can cause frequent nausea [SEP] ### +### [P_TEXT]: [CLS] frequent nausea, which may occur a few times per week or more, can be caused by ### +### foods eaten, illnesses, medications, or inner ear problems, among others. dehydration is one of the ### +### most common causes of frequent nausea. not drinking enough water throughout the day can lead to ### +### dizziness and a feeling of nausea, and it can also be potentially dangerous for the body. ### +### persistent anxiety and stress can be causes of frequent stomach upset as well. [SEP] ### +### ======================================= h_v_q | Gates: 29226 ======================================= ### +### ('nausea', 0, 0) ('frequent', 1, 1) ('can', 2, 166) ('cause', 3, 25) ('gut', 4, 199) ### +### ('caused', 5, 18) ('anger', 6, 39) ('.', 7, 773) ('kent', 8, 133) ('rebellion', 9, 63) ### +### ('causing', 10, 44) ('frequency', 11, 84) ('salt', 12, 80) ('frequently', 13, 36) ### +### ('could', 14, 1194) ('maya', 15, 834) ('occasional', 16, 57) ('causes', 17, 24) ('parker', 18, 74) ### +### ('stomach', 19, 4) ### +### ======================================= h_v_p | Gates: 29520 ======================================= ### +### ('nausea', 0, 0) ('frequent', 1, 1) ('##zziness', 2648, 2) ('upset', 23, 3) ('stomach', 19, 4) ### +### ('inner', 7224, 5) ('week', 3143, 6) ('dangerous', 503, 7) ('##dra', 15994, 8) ('feeling', 278, 9) ### +### ('water', 662, 10) ('common', 85, 11) ('foods', 8533, 12) ('##tion', 4032, 13) ('body', 5168, 14) ### +### ('weeks', 3576, 15) ('enough', 10916, 16) ('di', 5352, 17) ('caused', 5, 18) ('##hy', 25534, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('nausea', 0, 0) ('frequent', 1, 1) ('cause', 3, 25) ('caused', 5, 18) ('upset', 23, 3) ### +### ('anger', 6, 39) ('stomach', 19, 4) ('can', 2, 166) ('causing', 10, 44) ('causes', 17, 24) ### +### ('frequently', 13, 36) ('gut', 4, 199) ('rebellion', 9, 63) ('kent', 8, 133) ('occasional', 16, 57) ### +### ('frequency', 11, 84) ('common', 85, 11) ('salt', 12, 80) ('regular', 26, 49) ('harsh', 27, 48) ### +############################################################################################################ +[2023-10-07 19:19:09,185][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:19:09,186][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:19:09,610][root][INFO] - Epoch: 0: Step: 1201/1557, loss[v]=0.188901, lr=0.000015, acc@1[1]=235.5/256=0.919921875, acc@1[2]=247.0/256=0.96484375 +[2023-10-07 19:20:26,701][root][INFO] - Train batch 1300 +[2023-10-07 19:20:26,702][root][INFO] - Avg. loss per last 100 batches: 0.182401 +[2023-10-07 19:20:27,411][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29501.3/29522=99.93% | mean: 0.02 | max: 4.96 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.10 | max: 5.78 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a ring guard [SEP] ### +### [P_TEXT]: [CLS] before selecting the ring enhancer, you should consider what type of enhancer you ### +### would like - ring guard, ring wrap, curved ring or other type of ring enhancer. when selecting a ### +### ring enhancer, it is necessary to know both the size and shape of the solitaire diamond worn in ### +### conjunction with the enhancer. the shape of the enhancer must conform to the original ring and ### +### allow the center stone to fit properly into the wrap or insert. [SEP] ### +### ======================================= h_v_q | Gates: 29151 ======================================= ### +### ('ring', 0, 0) ('guard', 1, 2) ('rings', 2, 14) ('guards', 3, 604) ('.', 4, 525) ('bell', 5, 2587) ### +### ('circle', 6, 272) ('protect', 7, 2486) ('guy', 8, 749) ('wall', 9, 2585) ('band', 10, 1598) ### +### ('##guard', 11, 2357) ('string', 12, 580) ('kent', 13, 727) ('a', 14, 9821) ('unit', 15, 2542) ### +### ('bare', 16, 144) ('guarding', 17, 2681) ('network', 18, 2671) ('ringing', 19, 92) ### +### ======================================= h_v_p | Gates: 29520 ======================================= ### +### ('ring', 0, 0) ('diamond', 4759, 1) ('guard', 1, 2) ('##r', 5738, 3) ('##itaire', 23488, 4) ### +### ('enhance', 7260, 5) ('wrap', 10164, 6) ('stone', 697, 7) ('chose', 356, 8) ('fit', 8133, 9) ### +### ('size', 2197, 10) ('center', 451, 11) ('consider', 9450, 12) ('original', 4831, 13) ### +### ('rings', 2, 14) ('necessary', 1833, 15) ('selected', 2401, 16) ('sol', 5073, 17) ### +### ('##rs', 7343, 18) ('conjunction', 7239, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ring', 0, 0) ('guard', 1, 2) ('rings', 2, 14) ('guards', 3, 604) ('.', 4, 525) ('chose', 356, 8) ### +### ('circle', 6, 272) ('center', 451, 11) ('stone', 697, 7) ('pick', 35, 48) ('anger', 43, 43) ### +### ('harsh', 49, 45) ('ringing', 19, 92) ('chosen', 462, 22) ('bare', 16, 144) ('wingspan', 179, 46) ### +### ('guy', 8, 749) ('elliott', 50, 91) ('archaeology', 31, 145) ('david', 46, 113) ### +############################################################################################################ +[2023-10-07 19:20:27,412][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:20:27,412][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:20:27,814][root][INFO] - Epoch: 0: Step: 1301/1557, loss[v]=0.169373, lr=0.000017, acc@1[1]=235.0/256=0.91796875, acc@1[2]=242.0/256=0.9453125 +[2023-10-07 19:21:44,966][root][INFO] - Train batch 1400 +[2023-10-07 19:21:44,967][root][INFO] - Avg. loss per last 100 batches: 0.450249 +[2023-10-07 19:21:45,664][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29501.3/29522=99.93% | mean: 0.02 | max: 5.06 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.12 | max: 6.18 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what causes the glands straight back in your mouth to have white spots on them ### +### [SEP] ### +### [P_TEXT]: [CLS] remember : throat is a very vulnerable area of your organism. along with nasal ### +### cavity, it is the first abodyguarda against the invasion of viruses and bacteria. very red throat ### +### with white or yellow spots at its back are typical for inflamed throat, caused by bacterial ### +### infections. [SEP] ### +### ======================================= h_v_q | Gates: 29158 ======================================= ### +### ('mouth', 0, 59) ('glands', 1, 12229) ('white', 2, 22) ('spots', 3, 16) ('back', 4, 19) ### +### ('straight', 5, 19481) ('causes', 6, 88) ('them', 7, 3114) ('lips', 8, 107) ('spot', 9, 49) ### +### ('your', 10, 58) ('causing', 11, 51) ('cause', 12, 38) ('caused', 13, 26) ('forces', 14, 652) ### +### ('gland', 15, 23074) ('.', 16, 536) ('red', 17, 6) ('sharp', 18, 405) ('black', 19, 103) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('throat', 238, 0) ('nasal', 12872, 1) ('remember', 1514, 2) ('vulnerable', 10191, 3) ### +### ('invasion', 3682, 4) ('cavity', 17773, 5) ('red', 17, 6) ('##guard', 16068, 7) ### +### ('bacterial', 5401, 8) ('bacteria', 4700, 9) ('area', 3914, 10) ('typical', 4258, 11) ### +### ('##ame', 16808, 12) ('infections', 9839, 13) ('organism', 14136, 14) ('##od', 26715, 15) ### +### ('spots', 3, 16) ('ab', 9607, 17) ('remembered', 707, 18) ('back', 4, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('spots', 3, 16) ('back', 4, 19) ('white', 2, 22) ('mouth', 0, 59) ('red', 17, 6) ('causes', 6, 88) ### +### ('caused', 13, 26) ('spot', 9, 49) ('cause', 12, 38) ('causing', 11, 51) ('your', 10, 58) ### +### ('throat', 238, 0) ('lips', 8, 107) ('spotted', 25, 40) ('kent', 20, 77) ('anger', 22, 61) ### +### ('black', 19, 103) ('remember', 1514, 2) ('yellow', 358, 21) ('green', 57, 57) ### +############################################################################################################ +[2023-10-07 19:21:45,665][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:21:45,665][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:21:46,072][root][INFO] - Epoch: 0: Step: 1401/1557, loss[v]=0.095730, lr=0.000018, acc@1[1]=238.0/256=0.9296875, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 19:23:03,006][root][INFO] - Train batch 1500 +[2023-10-07 19:23:03,007][root][INFO] - Avg. loss per last 100 batches: 0.180581 +[2023-10-07 19:23:03,718][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29446.7/29522=99.74% | mean: 0.02 | max: 4.88 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.13 | max: 5.66 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] most dependable affordable cars [SEP] ### +### [P_TEXT]: [CLS] if you can look past its bargain interior and anonymous exterior, the suzuki sx4 is ### +### one of the most reliable and affordable all - wheel - drive cars. [SEP] ### +### ======================================= h_v_q | Gates: 29160 ======================================= ### +### ('affordable', 0, 6) ('##able', 1, 3524) ('depend', 2, 26061) ('cars', 3, 4) ('most', 4, 27) ### +### ('car', 5, 9) ('dependent', 6, 18181) ('private', 7, 4808) ('afford', 8, 23) ('.', 9, 4310) ### +### ('trust', 10, 6159) ('vehicles', 11, 22) (';', 12, 3686) ('indiana', 13, 6699) ('bob', 14, 4110) ### +### ('top', 15, 21) ('inexpensive', 16, 105) ('pace', 17, 124) ('many', 18, 2738) ### +### ('depended', 19, 22977) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('suzuki', 17210, 0) ('reliable', 2329, 1) ('bargain', 12583, 2) ('anonymous', 6659, 3) ### +### ('cars', 3, 4) ('##x', 8022, 5) ('affordable', 0, 6) ('##4', 4722, 7) ('wheel', 5227, 8) ### +### ('car', 5, 9) ('s', 4087, 10) ('interior', 952, 11) ('past', 805, 12) ('drive', 3503, 13) ### +### ('4', 440, 14) ('##3', 4820, 15) ('wingspan', 4245, 16) ('exterior', 9866, 17) ('##×', 7834, 18) ### +### ('3', 1250, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('affordable', 0, 6) ('cars', 3, 4) ('car', 5, 9) ('most', 4, 27) ('afford', 8, 23) ### +### ('##able', 1, 3524) ('vehicles', 11, 22) ('top', 15, 21) ('cheap', 20, 37) ('inexpensive', 16, 105) ### +### ('pace', 17, 124) ('harsh', 110, 20) ('kent', 33, 147) ('past', 805, 12) ('anger', 134, 38) ### +### ('4', 440, 14) ('malcolm', 54, 107) ('interior', 952, 11) ('sweet', 51, 125) ('expensive', 90, 61) ### +############################################################################################################ +[2023-10-07 19:23:03,718][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:23:03,718][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:23:04,141][root][INFO] - Epoch: 0: Step: 1501/1557, loss[v]=0.199587, lr=0.000019, acc@1[1]=234.5/256=0.916015625, acc@1[2]=242.5/256=0.947265625 +[2023-10-07 19:23:46,859][root][INFO] - rank=1; last iteration 1557 +[2023-10-07 19:23:46,859][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 19:23:46,859][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-07 19:23:46,861][root][INFO] - rank=2; last iteration 1557 +[2023-10-07 19:23:46,862][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 19:23:46,862][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-07 19:23:46,863][root][INFO] - rank=0; last iteration 1557 +[2023-10-07 19:23:46,864][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 19:23:46,864][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-07 19:23:46,864][root][INFO] - rank=3; last iteration 1557 +[2023-10-07 19:23:46,865][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 19:23:46,865][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-07 19:23:46,867][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 19:23:46,867][root][INFO] - Epoch finished on 1 +[2023-10-07 19:23:46,870][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 19:23:46,870][root][INFO] - Epoch finished on 2 +[2023-10-07 19:23:46,870][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 19:23:46,871][root][INFO] - Epoch finished on 0 +[2023-10-07 19:23:46,873][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 19:23:46,873][root][INFO] - Epoch finished on 3 +[2023-10-07 19:24:23,776][root][INFO] - Saved checkpoint at ./vdr_0 +[2023-10-07 19:24:23,776][root][INFO] - Saved checkpoint at ./vdr_0 +[2023-10-07 19:24:23,777][root][INFO] - Av Loss per epoch=56.210075 +[2023-10-07 19:24:23,777][root][INFO] - Av Loss per epoch=56.210075 +[2023-10-07 19:24:23,778][root][INFO] - epoch total (1) correct predictions=299313 +[2023-10-07 19:24:23,778][root][INFO] - epoch total (1) correct predictions=299313 +[2023-10-07 19:24:23,777][root][INFO] - Saved checkpoint at ./vdr_0 +[2023-10-07 19:24:23,778][root][INFO] - epoch total (2) correct predictions=332167 +[2023-10-07 19:24:23,778][root][INFO] - epoch total (2) correct predictions=332167 +[2023-10-07 19:24:23,778][root][INFO] - Av Loss per epoch=56.210075 +[2023-10-07 19:24:23,779][root][INFO] - epoch total (1) correct predictions=299313 +[2023-10-07 19:24:23,779][root][INFO] - epoch total (2) correct predictions=332167 +[2023-10-07 19:24:23,780][root][INFO] - Saved checkpoint at ./vdr_0 +[2023-10-07 19:24:23,781][root][INFO] - Av Loss per epoch=56.210075 +[2023-10-07 19:24:23,782][root][INFO] - epoch total (1) correct predictions=299313 +[2023-10-07 19:24:23,782][root][INFO] - epoch total (2) correct predictions=332167 +[2023-10-07 19:24:23,782][root][INFO] - ***** Epoch 1 ***** +[2023-10-07 19:24:23,782][root][INFO] - ***** Epoch 1 ***** +[2023-10-07 19:24:23,786][root][INFO] - ***** Epoch 1 ***** +[2023-10-07 19:24:23,788][root][INFO] - rank=1; Iteration start +[2023-10-07 19:24:23,788][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 19:24:23,788][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 19:24:23,789][root][INFO] - rank=2; Iteration start +[2023-10-07 19:24:23,789][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 19:24:23,789][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 19:24:23,790][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-07 19:24:23,789][root][INFO] - ***** Epoch 1 ***** +[2023-10-07 19:24:23,791][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-07 19:24:23,792][root][INFO] - rank=0; Iteration start +[2023-10-07 19:24:23,792][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 19:24:23,792][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 19:24:23,794][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-07 19:24:23,795][root][INFO] - rank=3; Iteration start +[2023-10-07 19:24:23,795][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 19:24:23,795][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 19:24:23,797][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-07 19:24:24,814][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29467.5/29522=99.82% | mean: 0.02 | max: 4.61 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.13 | max: 5.77 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] emilio estevez worth [SEP] ### +### [P_TEXT]: [CLS] emilio estevez net worth. emilio estevez net worth is $ 15 million. emilio estevez ### +### was born in new york and has an estimated net worth of $ 15 million dollars. an actor, director, ### +### producer, and writer, emil... [ read more ] [SEP] ### +### ======================================= h_v_q | Gates: 29167 ======================================= ### +### ('emilio', 0, 0) ('worth', 1, 6) ('##vez', 2, 1) ('este', 3, 5) ('jackson', 4, 3170) ### +### ('perry', 5, 527) ('ross', 6, 1266) ('.', 7, 774) ('lopez', 8, 94) ('martin', 9, 2899) ### +### ('empty', 10, 2374) ('war', 11, 2313) ('logan', 12, 2284) ('inverse', 13, 129) ('anne', 14, 1051) ### +### ('christopher', 15, 551) ('kent', 16, 415) ('evan', 17, 1427) ('smith', 18, 1245) ### +### ('mexico', 19, 702) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('emilio', 0, 0) ('##vez', 2, 1) ('actor', 820, 2) ('million', 1098, 3) ('net', 9591, 4) ### +### ('este', 3, 5) ('worth', 1, 6) ('born', 2238, 7) ('york', 115, 8) ('producer', 983, 9) ### +### ('$', 53, 10) ('actors', 1043, 11) ('[', 9068, 12) ('emil', 12429, 13) ('dollars', 2782, 14) ### +### ('writer', 1049, 15) ('producers', 1876, 16) ('15', 8819, 17) ('director', 524, 18) ### +### ('price', 404, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('emilio', 0, 0) ('##vez', 2, 1) ('worth', 1, 6) ('este', 3, 5) ('$', 53, 10) ('york', 115, 8) ### +### ('actor', 820, 2) ('lopez', 8, 94) ('million', 1098, 3) ('anger', 49, 38) ('inverse', 13, 129) ### +### ('harsh', 74, 31) ('perry', 5, 527) ('nelson', 28, 80) ('producer', 983, 9) ('price', 404, 19) ### +### ('antonio', 46, 62) ('actors', 1043, 11) ('director', 524, 18) ('clark', 36, 95) ### +############################################################################################################ +[2023-10-07 19:24:24,814][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:24:24,814][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:24:25,216][root][INFO] - Epoch: 1: Step: 1/1557, loss[v]=0.161955, lr=0.000020, acc@1[1]=233.0/256=0.91015625, acc@1[2]=240.5/256=0.939453125 +[2023-10-07 19:25:41,544][root][INFO] - Train batch 100 +[2023-10-07 19:25:41,545][root][INFO] - Avg. loss per last 100 batches: 0.159936 +[2023-10-07 19:25:42,238][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29454.9/29522=99.77% | mean: 0.02 | max: 4.70 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.12 | max: 5.84 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a construction company [SEP] ### +### [P_TEXT]: [CLS] construction is directly tied into the fields of civil engineering and ### +### architecture. it is a process that consists of building an infrastructure. as a result of this role ### +### or procedural description, a construction company is responsible for building structures, in both ### +### the commercial and private sectors. he laws which govern the construction process are aimed at ### +### preventing cataclysmic events in the future. explosions, bridge collapses, faulty structures, and ### +### wrongful deaths are examples of certain calamities that can occur if a construction company fails ### +### to adhere to the governing laws of their project. [SEP] ### +### ======================================= h_v_q | Gates: 28488 ======================================= ### +### ('construction', 0, 0) ('company', 1, 1) ('building', 2, 26) ('build', 3, 41) ('firm', 4, 43) ### +### ('club', 5, 226) ('team', 6, 149) ('built', 7, 38) ('design', 8, 82) ('companies', 9, 39) ### +### ('.', 10, 924) ('corporation', 11, 66) ('is', 12, 85) ('enterprise', 13, 7367) ### +### ('constructions', 14, 75) ('school', 15, 330) ('a', 16, 2499) ('constructed', 17, 81) ### +### ('partnership', 18, 1747) ('engineering', 19, 5) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('construction', 0, 0) ('company', 1, 1) ('tied', 3108, 2) ('architecture', 33, 3) ### +### ('examples', 2762, 4) ('engineering', 19, 5) ('infrastructure', 111, 6) ('structures', 242, 7) ### +### ('civil', 132, 8) ('##sm', 18492, 9) ('bridge', 332, 10) ('procedural', 16662, 11) ### +### ('responsible', 10940, 12) ('directly', 13113, 13) ('explosions', 8617, 14) ('explosion', 725, 15) ### +### ('structure', 198, 16) ('process', 339, 17) ('cat', 4254, 18) ('deaths', 14789, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('construction', 0, 0) ('company', 1, 1) ('building', 2, 26) ('build', 3, 41) ('firm', 4, 43) ### +### ('built', 7, 38) ('companies', 9, 39) ('engineering', 19, 5) ('design', 8, 82) ### +### ('corporation', 11, 66) ('architecture', 33, 3) ('team', 6, 149) ('club', 5, 226) ('is', 12, 85) ### +### ('constructions', 14, 75) ('architect', 20, 55) ('project', 35, 27) ('architectural', 32, 31) ### +### ('constructed', 17, 81) ('infrastructure', 111, 6) ### +############################################################################################################ +[2023-10-07 19:25:42,239][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:25:42,239][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:25:42,643][root][INFO] - Epoch: 1: Step: 101/1557, loss[v]=0.132998, lr=0.000020, acc@1[1]=240.0/256=0.9375, acc@1[2]=243.5/256=0.951171875 +[2023-10-07 19:26:58,980][root][INFO] - Train batch 200 +[2023-10-07 19:26:58,981][root][INFO] - Avg. loss per last 100 batches: 0.164044 +[2023-10-07 19:26:59,686][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29461.7/29522=99.80% | mean: 0.02 | max: 4.54 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 5.48 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] hanover square manhattan [SEP] ### +### [P_TEXT]: [CLS] broad street ( bmt nassau street line ) ( j z ) wall street ( irt broadway a ### +### seventh avenue line ) ( 2 3 ) south ferry a whitehall street ( 1 n r ) wall street ( irt lexington ### +### avenue line ) ( 4 5 ) hanover square park is a square and public park in the financial district, ### +### manhattan, new york city. it is triangular in shape, bordered by pearl street, stone street ( which ### +### is now pedestrian - only ) and a street named hanover square. [SEP] ### +### ======================================= h_v_q | Gates: 29121 ======================================= ### +### ('hanover', 0, 1) ('square', 1, 3) ('manhattan', 2, 10) ('brooklyn', 3, 44) ('hannover', 4, 35) ### +### ('harris', 5, 1010) ('.', 6, 1207) ('george', 7, 120) ('squares', 8, 72) ('olympic', 9, 604) ### +### ('tennis', 10, 2085) ('alliance', 11, 2828) ('park', 12, 0) ('olivier', 13, 1460) ### +### ('garden', 14, 51) ('york', 15, 13) ('kent', 16, 89) ('shaw', 17, 704) ('lee', 18, 227) ### +### ('charter', 19, 1911) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('park', 12, 0) ('hanover', 0, 1) ('ferry', 1890, 2) ('square', 1, 3) ('pearl', 3702, 4) ### +### ('nassau', 5566, 5) ('whitehall', 24181, 6) ('broad', 1452, 7) ('lexington', 8530, 8) ### +### ('triangular', 6167, 9) ('manhattan', 2, 10) ('parks', 2729, 11) ('street', 52, 12) ### +### ('york', 15, 13) ('wall', 783, 14) ('financial', 2592, 15) ('bordered', 3892, 16) ### +### ('pedestrian', 13356, 17) ('public', 1010, 18) ('##mt', 20023, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('hanover', 0, 1) ('square', 1, 3) ('manhattan', 2, 10) ('brooklyn', 3, 44) ('park', 12, 0) ### +### ('hannover', 4, 35) ('york', 15, 13) ('street', 52, 12) ('squares', 8, 72) ('george', 7, 120) ### +### ('garden', 14, 51) ('harsh', 59, 37) ('kent', 16, 89) ('harris', 5, 1010) ('carlisle', 33, 66) ### +### ('ferry', 1890, 2) ('anger', 38, 62) ('broad', 1452, 7) ('wall', 783, 14) ('.', 6, 1207) ### +############################################################################################################ +[2023-10-07 19:26:59,687][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:26:59,687][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:27:00,091][root][INFO] - Epoch: 1: Step: 201/1557, loss[v]=0.160629, lr=0.000020, acc@1[1]=235.5/256=0.919921875, acc@1[2]=244.5/256=0.955078125 +[2023-10-07 19:28:16,862][root][INFO] - Train batch 300 +[2023-10-07 19:28:16,863][root][INFO] - Avg. loss per last 100 batches: 0.157349 +[2023-10-07 19:28:17,565][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29481.4/29522=99.86% | mean: 0.02 | max: 4.79 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.13 | max: 5.89 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] definition concha [SEP] ### +### [P_TEXT]: [CLS] wiktionary ( 0. 00 / 0 votes ) rate this definition : 1 concha ( noun ) any shell - ### +### shaped structure. 2 concha ( noun ) the deepest indentation of the cartilage of the human ear, ### +### attaching to the mastoid bone. 3 concha ( noun ) an apse. [SEP] ### +### ======================================= h_v_q | Gates: 28882 ======================================= ### +### ('##cha', 0, 0) ('con', 1, 13) ('definition', 2, 10) ('##con', 3, 76) ('##cho', 4, 631) ### +### ('cha', 5, 77) ('##chi', 6, 312) ('.', 7, 2596) ('##ta', 8, 605) ('sas', 9, 1368) ### +### ('##che', 10, 119) ('##ca', 11, 652) ('bach', 12, 164) ('##ch', 13, 3816) ('alexander', 14, 618) ### +### ('murray', 15, 7734) ('julia', 16, 970) ('shaft', 17, 12861) ('definitions', 18, 31) ### +### ('canyon', 19, 6656) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('##cha', 0, 0) ('##entation', 3802, 1) ('votes', 8799, 2) ('##ry', 10098, 3) ('wi', 8447, 4) ### +### ('mast', 11546, 5) ('ear', 16196, 6) ('rate', 1753, 7) ('cart', 13775, 8) ('##kti', 27565, 9) ### +### ('definition', 2, 10) ('00', 4974, 11) ('vote', 4860, 12) ('con', 1, 13) ('##oid', 20893, 14) ### +### ('shell', 5660, 15) ('##ila', 2698, 16) ('deepest', 6994, 17) ('structure', 861, 18) ### +### ('##ona', 8121, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##cha', 0, 0) ('con', 1, 13) ('definition', 2, 10) ('##con', 3, 76) ('cha', 5, 77) ### +### ('definitions', 18, 31) ('wingspan', 20, 45) ('meaning', 194, 20) ('##chi', 6, 312) ### +### ('##che', 10, 119) ('##cho', 4, 631) ('human', 441, 23) ('turnout', 41, 50) ('bach', 12, 164) ### +### ('deep', 120, 43) ('harsh', 99, 47) ('anger', 35, 65) ('##ta', 8, 605) ('tan', 39, 72) ### +### ('structure', 861, 18) ### +############################################################################################################ +[2023-10-07 19:28:17,566][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:28:17,566][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:28:17,988][root][INFO] - Epoch: 1: Step: 301/1557, loss[v]=0.196555, lr=0.000020, acc@1[1]=242.5/256=0.947265625, acc@1[2]=247.0/256=0.96484375 +[2023-10-07 19:29:34,128][root][INFO] - Train batch 400 +[2023-10-07 19:29:34,129][root][INFO] - Avg. loss per last 100 batches: 0.156380 +[2023-10-07 19:29:34,823][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29460.2/29522=99.79% | mean: 0.02 | max: 4.77 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.12 | max: 5.88 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a male version of a mistress called [SEP] ### +### [P_TEXT]: [CLS] mistress derives from norman french maistresse. its male equivalent was maistre. in ### +### modern english, the latter is the master of a kept woman. he could also be a keptomaniac. paul ### +### roberts, lake cathie if his wife finds out, usually a rotten bastard. alastair wilson, balmain ### +### dunno, but a mistress is between a mister and a mattress. geoff linn, penrith historically, women ### +### didn't have the status or power to have extramarital sexual partners that men have had, so most ### +### terms imply the male as financially and socially superior. ` master'or ` consort'are generally ### +### considered the male equivalent. [SEP] ### +### ======================================= h_v_q | Gates: 28872 ======================================= ### +### ('male', 0, 8) ('mistress', 1, 3) ('version', 2, 243) ('called', 3, 4501) ('males', 4, 34) ### +### ('female', 5, 92) ('call', 6, 1727) ('versions', 7, 786) ('calling', 8, 6564) ('wife', 9, 21) ### +### ('men', 10, 53) ('master', 11, 7) ('.', 12, 115) ('variant', 13, 1123) ('mother', 14, 111) ### +### ('lover', 15, 88) ('calls', 16, 14174) ('teacher', 17, 220) ('man', 18, 200) ('lady', 19, 85) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('bastard', 1368, 0) ('rotten', 25278, 1) ('lin', 3691, 2) ('mistress', 1, 3) ('norman', 3510, 4) ### +### ('consort', 890, 5) ('equivalent', 566, 6) ('master', 11, 7) ('male', 0, 8) ('##ress', 788, 9) ### +### ('##main', 19911, 10) ('##oman', 26498, 11) ('##sta', 17498, 12) ('wilson', 2306, 13) ### +### ('roberts', 6417, 14) ('derives', 1640, 15) ('latter', 5248, 16) ('ala', 12828, 17) ### +### ('##rith', 17878, 18) ('mai', 21019, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('mistress', 1, 3) ('male', 0, 8) ('males', 4, 34) ('wife', 9, 21) ('master', 11, 7) ### +### ('version', 2, 243) ('men', 10, 53) ('female', 5, 92) ('.', 12, 115) ('versions', 7, 786) ### +### ('called', 3, 4501) ('lover', 15, 88) ('lady', 19, 85) ('mother', 14, 111) ('woman', 80, 42) ### +### ('equivalent', 566, 6) ('call', 6, 1727) ('encompasses', 45, 57) ('bastard', 1368, 0) ### +### ('consort', 890, 5) ### +############################################################################################################ +[2023-10-07 19:29:34,824][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:29:34,824][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:29:35,249][root][INFO] - Epoch: 1: Step: 401/1557, loss[v]=0.166930, lr=0.000020, acc@1[1]=233.0/256=0.91015625, acc@1[2]=243.0/256=0.94921875 +[2023-10-07 19:30:52,214][root][INFO] - Train batch 500 +[2023-10-07 19:30:52,216][root][INFO] - Avg. loss per last 100 batches: 0.147740 +[2023-10-07 19:30:52,919][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29456.8/29522=99.78% | mean: 0.02 | max: 4.64 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.13 | max: 5.96 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] most popular medium haircuts [SEP] ### +### [P_TEXT]: [CLS] keywords : curly hairstyles medium length, medium length curly hair, most popular ### +### medium length hairstyles, shoulder length curly haircuts, female hairstyles medium length, medium ### +### wavy hair with bangs, mid length layered hairstyles, halle berry 2013, medium wavy layered ### +### hairstyles, wavy layered haircuts. [SEP] ### +### ======================================= h_v_q | Gates: 28577 ======================================= ### +### ('medium', 0, 9) ('hair', 1, 2) ('popular', 2, 5) ('##cut', 3, 20) ('most', 4, 35) ('cut', 5, 16) ### +### ('popularity', 6, 17) ('##s', 7, 132) ('famous', 8, 48) ('frequent', 9, 1508) ('.', 10, 3798) ### +### ('common', 11, 170) ('middle', 12, 46) ('cutting', 13, 63) ('skin', 14, 1087) ('heavy', 15, 83) ### +### ('head', 16, 663) ('pop', 17, 59) ('mid', 18, 7) ('moderate', 19, 36) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('halle', 15999, 0) ('##tyle', 26158, 1) ('hair', 1, 2) ('curly', 23281, 3) ('hairs', 458, 4) ### +### ('popular', 2, 5) ('shoulder', 1708, 6) ('mid', 18, 7) ('layered', 18071, 8) ('medium', 0, 9) ### +### ('length', 2475, 10) ('berry', 6732, 11) ('##words', 21479, 12) ('key', 31, 13) ('wavy', 26866, 14) ### +### ('female', 9938, 15) ('cut', 5, 16) ('popularity', 6, 17) ('bangs', 25026, 18) ('male', 1075, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('hair', 1, 2) ('medium', 0, 9) ('popular', 2, 5) ('##cut', 3, 20) ('cut', 5, 16) ('most', 4, 35) ### +### ('popularity', 6, 17) ('mid', 18, 7) ('##s', 7, 132) ('key', 31, 13) ('famous', 8, 48) ### +### ('hairs', 458, 4) ('middle', 12, 46) ('moderate', 19, 36) ('cutting', 13, 63) ('pop', 17, 59) ### +### ('heavy', 15, 83) ('harsh', 106, 25) ('common', 11, 170) ('anger', 50, 44) ### +############################################################################################################ +[2023-10-07 19:30:52,920][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:30:52,920][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:30:53,341][root][INFO] - Epoch: 1: Step: 501/1557, loss[v]=0.226694, lr=0.000020, acc@1[1]=230.0/256=0.8984375, acc@1[2]=239.5/256=0.935546875 +[2023-10-07 19:32:09,248][root][INFO] - Train batch 600 +[2023-10-07 19:32:09,249][root][INFO] - Avg. loss per last 100 batches: 0.152008 +[2023-10-07 19:32:09,950][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29462.4/29522=99.80% | mean: 0.02 | max: 4.94 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 5.92 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the purpose of social security [SEP] ### +### [P_TEXT]: [CLS] what is the purpose of social security. the purpose of social security is to ### +### strengthen and support an individual who has faced an unwanted and tragic incident in his life. ### +### this may include loss of a dear one or losing a job or a sudden accident. keep reading to know more ### +### about its purpose. [SEP] ### +### ======================================= h_v_q | Gates: 28794 ======================================= ### +### ('social', 0, 3) ('purpose', 1, 1) ('security', 2, 0) ('encompasses', 3, 36) ('socio', 4, 14) ### +### ('is', 5, 1425) ('reason', 6, 45) ('society', 7, 739) ('mission', 8, 98) ('meaning', 9, 17) ### +### ('protection', 10, 175) ('.', 11, 3252) ('xavier', 12, 199) ('intent', 13, 47) ('barely', 14, 692) ### +### ('safety', 15, 257) ('target', 16, 185) ('of', 17, 11346) ('health', 18, 1799) ('ball', 19, 1504) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('security', 2, 0) ('purpose', 1, 1) ('dear', 4745, 2) ('social', 0, 3) ('tragic', 3557, 4) ### +### ('loss', 1116, 5) ('reading', 1857, 6) ('sudden', 2940, 7) ('unwanted', 24918, 8) ### +### ('accident', 5837, 9) ('losing', 3503, 10) ('faced', 4984, 11) ('life', 534, 12) ### +### ('strengthen', 13892, 13) ('socio', 4, 14) ('incident', 9097, 15) ('support', 185, 16) ### +### ('meaning', 9, 17) ('definition', 23, 18) ('harsh', 86, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('social', 0, 3) ('purpose', 1, 1) ('security', 2, 0) ('socio', 4, 14) ('meaning', 9, 17) ### +### ('encompasses', 3, 36) ('reason', 6, 45) ('definition', 23, 18) ('intent', 13, 47) ### +### ('mission', 8, 98) ('anger', 31, 27) ('knowing', 26, 43) ('support', 185, 16) ('harsh', 86, 19) ### +### ('protection', 10, 175) ('wingspan', 62, 34) ('purposes', 122, 22) ('hated', 42, 51) ### +### ('sociology', 75, 33) ('xavier', 12, 199) ### +############################################################################################################ +[2023-10-07 19:32:09,950][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:32:09,950][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:32:10,374][root][INFO] - Epoch: 1: Step: 601/1557, loss[v]=0.164881, lr=0.000020, acc@1[1]=230.0/256=0.8984375, acc@1[2]=244.0/256=0.953125 +[2023-10-07 19:33:26,787][root][INFO] - Train batch 700 +[2023-10-07 19:33:26,788][root][INFO] - Avg. loss per last 100 batches: 0.147032 +[2023-10-07 19:33:27,526][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29421.2/29522=99.66% | mean: 0.02 | max: 4.54 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.4/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.13 | max: 5.65 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what area is phone code 0203 [SEP] ### +### [P_TEXT]: [CLS] the 0203 area code is the prefix for a large number of london phone numbers. if you ### +### see any 0203 phone numbers it will be for a location either in london or the surrounding area. ### +### [SEP] ### +### ======================================= h_v_q | Gates: 28632 ======================================= ### +### ('phone', 0, 4) ('code', 1, 2) ('##0', 2, 6) ('02', 3, 8) ('area', 4, 3) ('##3', 5, 5) ### +### ('2002', 6, 44) ('codes', 7, 10) ('3', 8, 11) ('01', 9, 15) ('03', 10, 28) ('territory', 11, 1399) ### +### ('region', 12, 42) ('three', 13, 130) ('telephone', 14, 22) ('barely', 15, 247) ('.', 16, 1240) ### +### ('edward', 17, 807) ('trust', 18, 3633) ('district', 19, 655) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('london', 409, 0) ('prefix', 7926, 1) ('code', 1, 2) ('area', 4, 3) ('phone', 0, 4) ('##3', 5, 5) ### +### ('##0', 2, 6) ('numbers', 2784, 7) ('02', 3, 8) ('numbering', 8936, 9) ('codes', 7, 10) ### +### ('3', 8, 11) ('see', 2506, 12) ('define', 19740, 13) ('surrounding', 6112, 14) ('01', 9, 15) ### +### ('number', 82, 16) ('##8', 53, 17) ('large', 5259, 18) ('definition', 92, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('code', 1, 2) ('phone', 0, 4) ('##0', 2, 6) ('area', 4, 3) ('02', 3, 8) ('##3', 5, 5) ### +### ('codes', 7, 10) ('2002', 6, 44) ('3', 8, 11) ('01', 9, 15) ('03', 10, 28) ('telephone', 14, 22) ### +### ('region', 12, 42) ('london', 409, 0) ('##4', 20, 27) ('encompasses', 25, 25) ('anger', 26, 33) ### +### ('areas', 24, 40) ('##8', 53, 17) ('number', 82, 16) ### +############################################################################################################ +[2023-10-07 19:33:27,526][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:33:27,526][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:33:27,933][root][INFO] - Epoch: 1: Step: 701/1557, loss[v]=0.113773, lr=0.000020, acc@1[1]=239.0/256=0.93359375, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 19:34:44,750][root][INFO] - Train batch 800 +[2023-10-07 19:34:44,751][root][INFO] - Avg. loss per last 100 batches: 0.140518 +[2023-10-07 19:34:45,438][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29468.1/29522=99.82% | mean: 0.02 | max: 4.83 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.13 | max: 5.97 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how long to microwave fresh ear of corn [SEP] ### +### [P_TEXT]: [CLS] wet a paper towel, and wring out. wrap the ear of corn in the moist towel, and ### +### place on a dinner plate. cook in the microwave for 5 minutes. carefully remove paper towel, and ### +### enjoy! [SEP] ### +### ======================================= h_v_q | Gates: 28825 ======================================= ### +### ('microwave', 0, 9) ('corn', 1, 0) ('ear', 2, 5) ('fresh', 3, 2930) ('minutes', 4, 12) ### +### ('weeks', 5, 39) ('days', 6, 113) ('ears', 7, 15) ('years', 8, 86) ('simon', 9, 296) ### +### ('nathan', 10, 395) ('long', 11, 22) ('bare', 12, 280) ('hours', 13, 76) ('archaeology', 14, 96) ### +### ('wingspan', 15, 37) ('frederick', 16, 650) ('kent', 17, 84) ('warren', 18, 341) ### +### ('manchester', 19, 2511) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('corn', 1, 0) ('wet', 876, 1) ('towel', 13992, 2) ('dinner', 4642, 3) ('enjoy', 11733, 4) ### +### ('ear', 2, 5) ('wr', 9030, 6) ('wrap', 11411, 7) ('paper', 3676, 8) ('microwave', 0, 9) ### +### ('plate', 1290, 10) ('out', 7855, 11) ('minutes', 4, 12) ('carefully', 1913, 13) ### +### ('moist', 13103, 14) ('ears', 7, 15) ('towels', 21669, 16) ('cook', 8358, 17) ### +### ('wrapping', 9808, 18) ('remove', 10048, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('corn', 1, 0) ('microwave', 0, 9) ('ear', 2, 5) ('minutes', 4, 12) ('weeks', 5, 39) ### +### ('ears', 7, 15) ('days', 6, 113) ('long', 11, 22) ('fresh', 3, 2930) ('years', 8, 86) ### +### ('minute', 27, 26) ('wingspan', 15, 37) ('wet', 876, 1) ('anger', 31, 34) ('tightly', 61, 33) ### +### ('hours', 13, 76) ('##ο', 71, 35) ('kent', 17, 84) ('archaeology', 14, 96) ('harsh', 47, 57) ### +############################################################################################################ +[2023-10-07 19:34:45,439][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:34:45,439][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:34:45,861][root][INFO] - Epoch: 1: Step: 801/1557, loss[v]=0.156300, lr=0.000019, acc@1[1]=235.5/256=0.919921875, acc@1[2]=244.0/256=0.953125 +[2023-10-07 19:36:02,245][root][INFO] - Train batch 900 +[2023-10-07 19:36:02,246][root][INFO] - Avg. loss per last 100 batches: 0.140739 +[2023-10-07 19:36:02,953][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29439.2/29522=99.72% | mean: 0.02 | max: 4.31 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 5.80 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] why can water molecule pass through a cell membrane [SEP] ### +### [P_TEXT]: [CLS] rating newest oldest. best answer : the cell membrane has a hydrophopic area which ### +### enable the water to diffuse by osmosis, besides that, cell membrane also contain aquaporins ( a ### +### water transport channels ) which enable the water to pass though it. source ( s ) : my ### +### matriculation lecturer. unknown a · 7 years ago. thumbs up. 1. thumbs down. 0. [SEP] ### +### ======================================= h_v_q | Gates: 28635 ======================================= ### +### ('water', 0, 6) ('membrane', 1, 0) ('molecule', 2, 23402) ('pass', 3, 31) ('why', 4, 912) ### +### ('through', 5, 3697) ('cell', 6, 11) ('passing', 7, 143) ('passes', 8, 141) ('passed', 9, 73) ### +### ('molecules', 10, 10436) ('can', 11, 12099) ('cells', 12, 32) ('membranes', 13, 15) ### +### ('couldn', 14, 1211) ('trail', 15, 2346) ('via', 16, 1117) ('molecular', 17, 13782) ('.', 18, 2942) ### +### ('club', 19, 11736) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('membrane', 1, 0) ('answer', 1308, 1) ('thumbs', 25379, 2) ('·', 7033, 3) ('hydro', 954, 4) ### +### ('##ulation', 4609, 5) ('water', 0, 6) ('diffuse', 13690, 7) ('oldest', 6666, 8) ### +### ('lecturer', 10093, 9) ('os', 9504, 10) ('cell', 6, 11) ('besides', 4561, 12) ('source', 3584, 13) ### +### ('best', 3220, 14) ('membranes', 13, 15) ('##mos', 25327, 16) ('transport', 1510, 17) ### +### ('unknown', 1807, 18) ('ago', 7999, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('membrane', 1, 0) ('water', 0, 6) ('pass', 3, 31) ('cell', 6, 11) ('why', 4, 912) ### +### ('cells', 12, 32) ('membranes', 13, 15) ('passed', 9, 73) ('passing', 7, 143) ('passes', 8, 141) ### +### ('through', 5, 3697) ('anger', 24, 54) ('wingspan', 43, 36) ('ˈ', 48, 60) ('receptor', 79, 50) ### +### ('answer', 1308, 1) ('##ο', 143, 35) ('sharply', 91, 49) ('waters', 44, 86) ('able', 58, 74) ### +############################################################################################################ +[2023-10-07 19:36:02,954][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:36:02,954][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:36:03,359][root][INFO] - Epoch: 1: Step: 901/1557, loss[v]=0.135061, lr=0.000019, acc@1[1]=238.0/256=0.9296875, acc@1[2]=248.0/256=0.96875 +[2023-10-07 19:37:19,786][root][INFO] - Train batch 1000 +[2023-10-07 19:37:19,787][root][INFO] - Avg. loss per last 100 batches: 0.138420 +[2023-10-07 19:37:20,490][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29443.9/29522=99.74% | mean: 0.02 | max: 5.00 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 6.04 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what does establish justice mean [SEP] ### +### [P_TEXT]: [CLS] in the preamble to the us constitution, one aim of the us is to establish justice, ### +### which means to set up a fair and equitable system of laws - under which all persons are treated ### +### fairly and equally, and are guaranteed that their rights are respected by others. [SEP] ### +### ======================================= h_v_q | Gates: 28909 ======================================= ### +### ('justice', 0, 0) ('establish', 1, 7) ('established', 2, 10) ('establishing', 3, 28) ### +### ('definition', 4, 47) ('meaning', 5, 20) ('means', 6, 14) ('.', 7, 1780) ('justices', 8, 24) ### +### ('establishes', 9, 142) ('peace', 10, 3708) ('truth', 11, 200) ('establishment', 12, 264) ### +### ('sense', 13, 2729) ('does', 14, 8020) ('encompasses', 15, 27) ('founded', 16, 64) ### +### ('create', 17, 59) ('noble', 18, 5195) ('mean', 19, 88) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('justice', 0, 0) ('constitution', 3241, 1) ('##quitable', 27841, 2) ('fair', 723, 3) ### +### ('us', 3568, 4) ('aim', 10870, 5) ('laws', 1379, 6) ('establish', 1, 7) ('persons', 8220, 8) ### +### ('equally', 13275, 9) ('established', 2, 10) ('rights', 3300, 11) ('under', 5365, 12) ### +### ('respected', 6599, 13) ('means', 6, 14) ('pre', 6370, 15) ('person', 2282, 16) ('##am', 13671, 17) ### +### ('set', 80, 18) ('usa', 10494, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('justice', 0, 0) ('establish', 1, 7) ('established', 2, 10) ('establishing', 3, 28) ### +### ('means', 6, 14) ('meaning', 5, 20) ('definition', 4, 47) ('justices', 8, 24) ### +### ('encompasses', 15, 27) ('law', 20, 30) ('founded', 16, 64) ('create', 17, 59) ('set', 80, 18) ### +### ('meant', 32, 36) ('establishes', 9, 142) ('mean', 19, 88) ('truth', 11, 200) ('bare', 21, 103) ### +### ('fair', 723, 3) ('establishment', 12, 264) ### +############################################################################################################ +[2023-10-07 19:37:20,490][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:37:20,490][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:37:20,912][root][INFO] - Epoch: 1: Step: 1001/1557, loss[v]=0.175545, lr=0.000019, acc@1[1]=236.5/256=0.923828125, acc@1[2]=243.5/256=0.951171875 +[2023-10-07 19:38:37,337][root][INFO] - Train batch 1100 +[2023-10-07 19:38:37,339][root][INFO] - Avg. loss per last 100 batches: 0.135678 +[2023-10-07 19:38:38,025][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29422.6/29522=99.66% | mean: 0.02 | max: 5.09 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 6.09 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] which county huntingtown md belongs to [SEP] ### +### [P_TEXT]: [CLS] huntingtown is a census - designated place ( cdp ) in calvert county, maryland, ### +### united states. the population was 3, 311 at the 2010 census, [ 1 ] up from 2, 436 at the 2000 ### +### census. many large estate homes have recently been built in small developments off maryland route 2 ### +### / 4. it has a public high school called huntingtown high. the calverton school is located just ### +### south of the town center. [SEP] ### +### ======================================= h_v_q | Gates: 28674 ======================================= ### +### ('hunting', 0, 0) ('county', 1, 11) ('##town', 2, 1) ('belongs', 3, 22588) ('md', 4, 9) ### +### ('maryland', 5, 4) ('hunt', 6, 7) ('belonged', 7, 13529) ('hunter', 8, 103) ('belonging', 9, 5065) ### +### ('hunters', 10, 20) ('hunted', 11, 12) ('fishing', 12, 52) ('belong', 13, 21027) ### +### ('counties', 14, 35) ('assigned', 15, 1480) ('shooting', 16, 482) ('bishop', 17, 1449) ### +### ('owns', 18, 1073) ('district', 19, 2171) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('hunting', 0, 0) ('##town', 2, 1) ('cdp', 15770, 2) ('calvert', 9777, 3) ('maryland', 5, 4) ### +### ('census', 4108, 5) ('designated', 281, 6) ('hunt', 6, 7) ('estate', 76, 8) ('md', 4, 9) ### +### ('place', 3198, 10) ('county', 1, 11) ('hunted', 11, 12) ('developments', 15778, 13) ### +### ('route', 5494, 14) ('homes', 15711, 15) ('population', 1254, 16) ('recently', 11972, 17) ### +### ('3', 5182, 18) ('hunts', 52, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('hunting', 0, 0) ('##town', 2, 1) ('county', 1, 11) ('maryland', 5, 4) ('md', 4, 9) ('hunt', 6, 7) ### +### ('hunters', 10, 20) ('hunted', 11, 12) ('hunter', 8, 103) ('fishing', 12, 52) ('counties', 14, 35) ### +### ('estate', 76, 8) ('town', 30, 28) ('hunts', 52, 19) ('designated', 281, 6) ('killing', 26, 65) ### +### ('harsh', 57, 43) ('encompasses', 96, 25) ('wingspan', 64, 39) ('anger', 68, 44) ### +############################################################################################################ +[2023-10-07 19:38:38,026][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:38:38,026][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:38:38,450][root][INFO] - Epoch: 1: Step: 1101/1557, loss[v]=0.145346, lr=0.000019, acc@1[1]=236.0/256=0.921875, acc@1[2]=247.0/256=0.96484375 +[2023-10-07 19:39:55,296][root][INFO] - Train batch 1200 +[2023-10-07 19:39:55,297][root][INFO] - Avg. loss per last 100 batches: 0.136784 +[2023-10-07 19:39:56,011][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29398.6/29522=99.58% | mean: 0.01 | max: 4.92 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 6.07 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] who was the wife of hercules [SEP] ### +### [P_TEXT]: [CLS] megara was the first wife of the greek hero herakles ( better known as hercules ). ### +### she was the daughter of king creon of thebes who gave her in marriage to hercules in gratitude for ### +### his help in winning back creon's kingdom from the minyans. [SEP] ### +### ======================================= h_v_q | Gates: 28546 ======================================= ### +### ('hercules', 0, 0) ('wife', 1, 3) ('was', 2, 24) ('husband', 3, 18) ('woman', 4, 72) ('who', 5, 48) ### +### ('daughter', 6, 12) ('widow', 7, 17) ('wives', 8, 22) ('women', 9, 79) ('married', 10, 32) ### +### ('.', 11, 1303) ('of', 12, 1736) ('her', 13, 13) ('george', 14, 203) ('marriage', 15, 25) ### +### ('mother', 16, 77) ('love', 17, 597) ('she', 18, 19) ('kent', 19, 55) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('hercules', 0, 0) ('mega', 8066, 1) ('##ra', 4491, 2) ('wife', 1, 3) ('##bes', 22808, 4) ### +### ('##eon', 13377, 5) ('min', 18828, 6) ('greek', 634, 7) ('hero', 991, 8) ('first', 1810, 9) ### +### ('##les', 7632, 10) ('king', 718, 11) ('daughter', 6, 12) ('her', 13, 13) ('kingdom', 5492, 14) ### +### ('greece', 173, 15) ('gratitude', 4606, 16) ('widow', 7, 17) ('husband', 3, 18) ('she', 18, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('hercules', 0, 0) ('wife', 1, 3) ('husband', 3, 18) ('was', 2, 24) ('daughter', 6, 12) ### +### ('widow', 7, 17) ('wives', 8, 22) ('who', 5, 48) ('her', 13, 13) ('woman', 4, 72) ### +### ('married', 10, 32) ('she', 18, 19) ('marriage', 15, 25) ('women', 9, 79) ('daughters', 20, 30) ### +### ('kent', 19, 55) ('mother', 16, 77) ('mom', 31, 38) ('greece', 173, 15) ('were', 52, 36) ### +############################################################################################################ +[2023-10-07 19:39:56,012][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:39:56,012][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:39:56,417][root][INFO] - Epoch: 1: Step: 1201/1557, loss[v]=0.133078, lr=0.000019, acc@1[1]=237.5/256=0.927734375, acc@1[2]=245.5/256=0.958984375 +[2023-10-07 19:41:12,712][root][INFO] - Train batch 1300 +[2023-10-07 19:41:12,712][root][INFO] - Avg. loss per last 100 batches: 0.135919 +[2023-10-07 19:41:13,426][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29434.9/29522=99.70% | mean: 0.02 | max: 4.81 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 5.96 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] synonym word for scavenger [SEP] ### +### [P_TEXT]: [CLS] princeton's wordnet ( 0. 00 / 0 votes ) rate these synonyms : scavenger ( noun ). a ### +### chemical agent that is added to a chemical mixture to counteract the effects of impurities. ### +### synonyms : magpie, pack rat. magpie, scavenger, pack rat ( noun ). someone who collects things that ### +### have been discarded by others. [SEP] ### +### ======================================= h_v_q | Gates: 28894 ======================================= ### +### ('##nger', 0, 1) ('sc', 1, 6) ('synonym', 2, 13) ('##ave', 3, 40) ('word', 4, 15) ('bare', 5, 254) ### +### ('name', 6, 8321) ('hunter', 7, 3636) ('term', 8, 346) ('.', 9, 5377) ('letter', 10, 3023) ### +### ('cooper', 11, 7212) ('berlin', 12, 67) ('kent', 13, 791) ('warren', 14, 482) ### +### ('manchester', 15, 713) ('frederick', 16, 700) ('liberal', 17, 1702) ('definition', 18, 14) ### +### ('paul', 19, 5753) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('princeton', 3311, 0) ('##nger', 0, 1) ('pack', 804, 2) ('rat', 359, 3) ('mag', 4115, 4) ### +### ('##net', 8029, 5) ('sc', 1, 6) ('votes', 5914, 7) ('##pie', 6986, 8) ('##urities', 13569, 9) ### +### ('rate', 5910, 10) ('chemical', 1802, 11) ('agent', 937, 12) ('synonym', 2, 13) ### +### ('definition', 18, 14) ('word', 4, 15) ('added', 934, 16) ('meaning', 779, 17) ('00', 7515, 18) ### +### ('define', 15883, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##nger', 0, 1) ('sc', 1, 6) ('synonym', 2, 13) ('word', 4, 15) ('##ave', 3, 40) ### +### ('definition', 18, 14) ('rat', 359, 3) ('bare', 5, 254) ('encompasses', 58, 25) ('pack', 804, 2) ### +### ('berlin', 12, 67) ('harsh', 26, 55) ('words', 22, 97) ('agent', 937, 12) ('scrapped', 261, 38) ### +### ('meaning', 779, 17) ('term', 8, 346) ('##ο', 146, 49) ('added', 934, 16) ('princeton', 3311, 0) ### +############################################################################################################ +[2023-10-07 19:41:13,426][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:41:13,426][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:41:13,848][root][INFO] - Epoch: 1: Step: 1301/1557, loss[v]=0.116727, lr=0.000019, acc@1[1]=238.0/256=0.9296875, acc@1[2]=247.0/256=0.96484375 +[2023-10-07 19:42:30,034][root][INFO] - Train batch 1400 +[2023-10-07 19:42:30,035][root][INFO] - Avg. loss per last 100 batches: 0.134656 +[2023-10-07 19:42:30,742][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29453.5/29522=99.77% | mean: 0.02 | max: 4.77 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 5.89 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how do coyotes survive in the desert [SEP] ### +### [P_TEXT]: [CLS] diet. coyotes are omnivores, which means they will eat or try to eat just about ### +### anything. in the sonoran desert coyotes vary their diet with the seasons. cactus fruit, mesquite ### +### beans, flowers, insects, rodents, lizards, rabbits, birds, and snakes make up some of their dietary ### +### choices. daptations. coyotes adjust their hunting style to what foods are available. when they hunt ### +### small prey alone, they usually stalk it and then pounce. if the prey is larger like a deer, they ### +### will often hunt in small packs and work together to kill the prey. [SEP] ### +### ======================================= h_v_q | Gates: 28927 ======================================= ### +### ('coyotes', 0, 0) ('desert', 1, 1) ('survive', 2, 1122) ('arizona', 3, 73) ('tucson', 4, 22) ### +### ('survival', 5, 4546) ('coyote', 6, 16) ('.', 7, 1059) ('##brush', 8, 1218) ('survived', 9, 12468) ### +### ('angeles', 10, 5347) ('##hawks', 11, 224) ('cardinals', 12, 102) ('dodgers', 13, 144) ### +### ('deputy', 14, 637) ('do', 15, 13612) ('nelson', 16, 468) ('2002', 17, 738) ('mexican', 18, 270) ### +### ('juarez', 19, 335) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('coyotes', 0, 0) ('desert', 1, 1) ('hunting', 148, 2) ('prey', 4148, 3) ('snakes', 3361, 4) ### +### ('diet', 26387, 5) ('dietary', 20750, 6) ('##unce', 25666, 7) ('rabbits', 95, 8) ### +### ('cactus', 5210, 9) ('sonora', 2731, 10) ('##vor', 14874, 11) ('hunt', 1033, 12) ('eat', 1848, 13) ### +### ('om', 9393, 14) ('da', 9533, 15) ('coyote', 6, 16) ('##es', 9181, 17) ('choices', 4364, 18) ### +### ('anything', 14197, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('coyotes', 0, 0) ('desert', 1, 1) ('tucson', 4, 22) ('coyote', 6, 16) ('arizona', 3, 73) ### +### ('survive', 2, 1122) ('hunting', 148, 2) ('rabbits', 95, 8) ('navajo', 29, 36) ('″', 31, 53) ### +### ('hunters', 82, 33) ('cardinals', 12, 102) ('dodgers', 13, 144) ('anger', 42, 68) ### +### ('sirens', 44, 70) ('##hawks', 11, 224) ('tigers', 202, 31) ('##izzly', 40, 84) ('arid', 23, 175) ### +### ('##ο', 156, 54) ### +############################################################################################################ +[2023-10-07 19:42:30,743][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:42:30,743][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:42:31,154][root][INFO] - Epoch: 1: Step: 1401/1557, loss[v]=0.084432, lr=0.000019, acc@1[1]=238.0/256=0.9296875, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 19:43:48,012][root][INFO] - Train batch 1500 +[2023-10-07 19:43:48,013][root][INFO] - Avg. loss per last 100 batches: 0.130302 +[2023-10-07 19:43:48,693][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29393.5/29522=99.56% | mean: 0.01 | max: 4.78 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.13 | max: 5.83 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what are silo armaments [SEP] ### +### [P_TEXT]: [CLS] a missile launch facility, also known as an underground missile silo or launch ### +### facilityalf, is a vertical cylindrical structure constructed underground, for the storage and ### +### launching of intercontinental ballistic missiles ( icbms ). ost silos were based in colorado, ### +### arizona, nebraska, north dakota, south dakota, montana, wyoming and other western states away from ### +### heavily populated areas. they had many defense systems to keep out intruders and other defense ### +### systems to prevent destruction ( see project safeguard ). [SEP] ### +### ======================================= h_v_q | Gates: 28726 ======================================= ### +### ('##lo', 0, 11) ('si', 1, 7) ('armament', 2, 8811) ('##s', 3, 572) ('are', 4, 931) ### +### ('weapon', 5, 9275) ('.', 6, 254) ('weapons', 7, 1788) ('encompasses', 8, 26) ('arm', 9, 4131) ### +### ('include', 10, 493) ('lo', 11, 107) ('were', 12, 50) ('##no', 13, 63) ('##la', 14, 106) ### +### ('ball', 15, 304) ('park', 16, 406) ('navy', 17, 2723) ('army', 18, 3415) ('##co', 19, 3993) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##los', 232, 0) ('underground', 6690, 1) ('missile', 281, 2) ('os', 8475, 3) ('launch', 1277, 4) ### +### ('missiles', 3536, 5) ('##f', 388, 6) ('si', 1, 7) ('##bm', 27243, 8) ('storage', 2812, 9) ### +### ('montana', 4815, 10) ('##lo', 0, 11) ('dakota', 114, 12) ('intercontinental', 11159, 13) ### +### ('vertical', 10588, 14) ('colorado', 253, 15) ('ic', 9474, 16) ('##al', 7628, 17) ### +### ('intruder', 6850, 18) ('ballistic', 3009, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('si', 1, 7) ('##lo', 0, 11) ('encompasses', 8, 26) ('##s', 3, 572) ('armament', 2, 8811) ### +### ('##los', 232, 0) ('were', 12, 50) ('dakota', 114, 12) ('missile', 281, 2) ('##f', 388, 6) ### +### ('.', 6, 254) ('##no', 13, 63) ('colorado', 253, 15) ('lo', 11, 107) ('arizona', 312, 20) ### +### ('##la', 14, 106) ('launch', 1277, 4) ('definition', 395, 24) ('are', 4, 931) ('##ο', 52, 54) ### +############################################################################################################ +[2023-10-07 19:43:48,694][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:43:48,694][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:43:49,117][root][INFO] - Epoch: 1: Step: 1501/1557, loss[v]=0.141627, lr=0.000019, acc@1[1]=241.5/256=0.943359375, acc@1[2]=250.5/256=0.978515625 +[2023-10-07 19:44:32,184][root][INFO] - rank=3; last iteration 1557 +[2023-10-07 19:44:32,184][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 19:44:32,185][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-07 19:44:32,187][root][INFO] - rank=2; last iteration 1557 +[2023-10-07 19:44:32,187][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 19:44:32,187][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-07 19:44:32,188][root][INFO] - rank=1; last iteration 1557 +[2023-10-07 19:44:32,188][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 19:44:32,188][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-07 19:44:32,191][root][INFO] - rank=0; last iteration 1557 +[2023-10-07 19:44:32,191][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 19:44:32,192][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-07 19:44:32,192][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 19:44:32,193][root][INFO] - Epoch finished on 3 +[2023-10-07 19:44:32,195][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 19:44:32,195][root][INFO] - Epoch finished on 2 +[2023-10-07 19:44:32,196][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 19:44:32,196][root][INFO] - Epoch finished on 1 +[2023-10-07 19:44:32,198][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 19:44:32,198][root][INFO] - Epoch finished on 0 +[2023-10-07 19:44:45,592][root][INFO] - Saved checkpoint at ./vdr_1 +[2023-10-07 19:44:45,593][root][INFO] - Av Loss per epoch=0.144343 +[2023-10-07 19:44:45,593][root][INFO] - epoch total (1) correct predictions=367962 +[2023-10-07 19:44:45,593][root][INFO] - epoch total (2) correct predictions=382303 +[2023-10-07 19:44:45,593][root][INFO] - Saved checkpoint at ./vdr_1 +[2023-10-07 19:44:45,594][root][INFO] - Av Loss per epoch=0.144343 +[2023-10-07 19:44:45,594][root][INFO] - epoch total (1) correct predictions=367962 +[2023-10-07 19:44:45,594][root][INFO] - epoch total (2) correct predictions=382303 +[2023-10-07 19:44:45,594][root][INFO] - Saved checkpoint at ./vdr_1 +[2023-10-07 19:44:45,594][root][INFO] - Av Loss per epoch=0.144343 +[2023-10-07 19:44:45,594][root][INFO] - epoch total (1) correct predictions=367962 +[2023-10-07 19:44:45,595][root][INFO] - epoch total (2) correct predictions=382303 +[2023-10-07 19:44:45,596][root][INFO] - ***** Epoch 2 ***** +[2023-10-07 19:44:45,596][root][INFO] - Saved checkpoint at ./vdr_1 +[2023-10-07 19:44:45,597][root][INFO] - Av Loss per epoch=0.144343 +[2023-10-07 19:44:45,598][root][INFO] - epoch total (1) correct predictions=367962 +[2023-10-07 19:44:45,598][root][INFO] - epoch total (2) correct predictions=382303 +[2023-10-07 19:44:45,597][root][INFO] - ***** Epoch 2 ***** +[2023-10-07 19:44:45,597][root][INFO] - ***** Epoch 2 ***** +[2023-10-07 19:44:45,601][root][INFO] - rank=2; Iteration start +[2023-10-07 19:44:45,601][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 19:44:45,601][root][INFO] - rank=0; Iteration start +[2023-10-07 19:44:45,601][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 19:44:45,601][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 19:44:45,601][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 19:44:45,601][root][INFO] - rank=3; Iteration start +[2023-10-07 19:44:45,602][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 19:44:45,602][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 19:44:45,602][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-07 19:44:45,602][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-07 19:44:45,603][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-07 19:44:45,606][root][INFO] - ***** Epoch 2 ***** +[2023-10-07 19:44:45,612][root][INFO] - rank=1; Iteration start +[2023-10-07 19:44:45,612][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 19:44:45,613][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 19:44:45,614][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-07 19:44:46,555][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29433.7/29522=99.70% | mean: 0.01 | max: 4.95 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 5.84 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is compliance elf [SEP] ### +### [P_TEXT]: [CLS] compliance elf compliance elf ( employee level filing ) streamlines the compliance ### +### process related to personal trading / dealing monitoring, affirmations, gift and political ### +### contribution reporting and much more a for both employees and the compliance team. [SEP] ### +### ======================================= h_v_q | Gates: 28945 ======================================= ### +### ('elf', 0, 0) ('compliance', 1, 2) ('definition', 2, 23) ('encompasses', 3, 69) ('elves', 4, 6) ### +### ('is', 5, 7390) ('comply', 6, 18) ('.', 7, 6575) ('enforcement', 8, 68) ('accordance', 9, 30) ### +### ('resistance', 10, 320) ('loyalty', 11, 906) ('pilot', 12, 4563) ('liberal', 13, 1023) ### +### ('complied', 14, 261) ('communication', 15, 2132) ('violations', 16, 133) ('alien', 17, 529) ### +### (';', 18, 5320) ('respect', 19, 2139) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('elf', 0, 0) ('filing', 6168, 1) ('compliance', 1, 2) ('employee', 7042, 3) ('trading', 4128, 4) ### +### ('dealing', 1476, 5) ('elves', 4, 6) ('monitoring', 133, 7) ('filed', 4247, 8) ('gift', 4742, 9) ### +### ('personal', 1285, 10) ('reporting', 9731, 11) ('employees', 6603, 12) ('level', 1555, 13) ### +### ('stream', 7664, 14) ('process', 64, 15) ('related', 561, 16) ('contribution', 1144, 17) ### +### ('comply', 6, 18) ('##firm', 21190, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('elf', 0, 0) ('compliance', 1, 2) ('elves', 4, 6) ('definition', 2, 23) ('comply', 6, 18) ### +### ('encompasses', 3, 69) ('accordance', 9, 30) ('enforcement', 8, 68) ('process', 64, 15) ### +### ('monitoring', 133, 7) ('compliant', 20, 29) ('odd', 54, 31) ('signature', 23, 70) ### +### ('related', 561, 16) ('violations', 16, 133) ('anger', 105, 39) ('resistance', 10, 320) ### +### ('dealing', 1476, 5) ('complied', 14, 261) ('secretary', 86, 67) ### +############################################################################################################ +[2023-10-07 19:44:46,556][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:44:46,556][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:44:46,940][root][INFO] - Epoch: 2: Step: 1/1557, loss[v]=0.070385, lr=0.000019, acc@1[1]=245.5/256=0.958984375, acc@1[2]=253.0/256=0.98828125 +[2023-10-07 19:46:03,439][root][INFO] - Train batch 100 +[2023-10-07 19:46:03,439][root][INFO] - Avg. loss per last 100 batches: 0.122176 +[2023-10-07 19:46:04,160][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29415.8/29522=99.64% | mean: 0.01 | max: 5.00 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 5.69 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] will apple watch work with 4s iphone [SEP] ### +### [P_TEXT]: [CLS] it has to do with the fact that the watch requires bluetooth 4. 0, which the iphone ### +### 4s does not support. iphone 4s & apple watch both have bluetooth low energy ( ble ) 4. 0. level 1 ( ### +### 0 points ). hey, brandon. [SEP] ### +### ======================================= h_v_q | Gates: 28467 ======================================= ### +### ('apple', 0, 6) ('watch', 1, 2) ('iphone', 2, 3) ('4', 3, 10) ('work', 4, 14413) ('##s', 5, 226) ### +### ('four', 6, 35) ('watching', 7, 13) ('4th', 8, 55) ('working', 9, 11296) ('##4', 10, 56) ### +### ('watched', 11, 27) ('phone', 12, 71) ('3', 13, 25) ('5', 14, 47) ('6', 15, 45) ('will', 16, 18715) ### +### ('.', 17, 9961) ('albert', 18, 1084) ('film', 19, 378) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('brandon', 5850, 0) ('##tooth', 21655, 1) ('watch', 1, 2) ('iphone', 2, 3) ('blue', 713, 4) ### +### ('low', 1009, 5) ('apple', 0, 6) ('##le', 8624, 7) ('hey', 4101, 8) ('b', 5050, 9) ('4', 3, 10) ### +### ('watches', 47, 11) ('level', 2493, 12) ('watching', 7, 13) ('energy', 3072, 14) ('both', 4635, 15) ### +### ('requires', 4810, 16) ('points', 12216, 17) ('fact', 9547, 18) ('ˈ', 988, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('watch', 1, 2) ('apple', 0, 6) ('iphone', 2, 3) ('4', 3, 10) ('watching', 7, 13) ('##s', 5, 226) ### +### ('four', 6, 35) ('4th', 8, 55) ('watched', 11, 27) ('3', 13, 25) ('##4', 10, 56) ### +### ('watches', 47, 11) ('6', 15, 45) ('5', 14, 47) ('phone', 12, 71) ('##ο', 43, 21) ('2', 25, 46) ### +### ('bother', 32, 38) ('1', 28, 54) ('blue', 713, 4) ### +############################################################################################################ +[2023-10-07 19:46:04,160][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:46:04,160][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:46:04,582][root][INFO] - Epoch: 2: Step: 101/1557, loss[v]=0.148942, lr=0.000019, acc@1[1]=237.0/256=0.92578125, acc@1[2]=247.0/256=0.96484375 +[2023-10-07 19:47:21,658][root][INFO] - Train batch 200 +[2023-10-07 19:47:21,659][root][INFO] - Avg. loss per last 100 batches: 0.119579 +[2023-10-07 19:47:22,362][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29431.5/29522=99.69% | mean: 0.02 | max: 4.66 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 5.68 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a good dry sherry to cook with? [SEP] ### +### [P_TEXT]: [CLS] cooking sherry has added preservatives ( salt ) to increase it's shelf - life and ### +### is not suitable for drinking. types of sherry : fino is a pale straw and gold color, with a ### +### delicate crisp aroma ( nutty ). it is ideal with tapas, soups, seafood, fish, ham and mild cheese. ### +### manzanilla : straw colored, has a crisp aroma, and it is dry and light. [SEP] ### +### ======================================= h_v_q | Gates: 28592 ======================================= ### +### ('sherry', 0, 0) ('dry', 1, 19) ('cook', 2, 118) ('good', 3, 358) ('wet', 4, 145) ('to', 5, 19919) ### +### ('is', 6, 1006) ('warren', 7, 110) ('kent', 8, 252) ('bother', 9, 33) ('eat', 10, 257) ### +### ('?', 11, 20762) ('massachusetts', 12, 10834) ('frederick', 13, 367) ('helen', 14, 98) ### +### ('with', 15, 528) ('definition', 16, 161) ('wine', 17, 81) ('albert', 18, 307) ### +### ('conservative', 19, 632) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('sherry', 0, 0) ('straw', 7094, 1) ('fin', 1814, 2) ('delicate', 449, 3) ('cooking', 28, 4) ### +### ('tap', 1502, 5) ('crisp', 4813, 6) ('salt', 169, 7) ('##o', 12884, 8) ('pale', 4597, 9) ### +### ('gold', 1762, 10) ('ideal', 2052, 11) ('cheese', 9170, 12) ('drinking', 6568, 13) ### +### ('color', 4067, 14) ('aroma', 18926, 15) ('types', 5631, 16) ('##ves', 12044, 17) ('fish', 644, 18) ### +### ('dry', 1, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('sherry', 0, 0) ('dry', 1, 19) ('cook', 2, 118) ('wet', 4, 145) ('cooking', 28, 4) ### +### ('good', 3, 358) ('bother', 9, 33) ('warren', 7, 110) ('salt', 169, 7) ('encompasses', 31, 38) ### +### ('delicate', 449, 3) ('wine', 17, 81) ('helen', 14, 98) ('richard', 20, 86) ('kent', 8, 252) ### +### ('anger', 33, 59) ('bare', 23, 76) ('ruins', 57, 41) ('ছ', 97, 31) ('afraid', 112, 34) ### +############################################################################################################ +[2023-10-07 19:47:22,362][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:47:22,362][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:47:22,779][root][INFO] - Epoch: 2: Step: 201/1557, loss[v]=0.101179, lr=0.000019, acc@1[1]=242.0/256=0.9453125, acc@1[2]=249.5/256=0.974609375 +[2023-10-07 19:48:38,943][root][INFO] - Train batch 300 +[2023-10-07 19:48:38,944][root][INFO] - Avg. loss per last 100 batches: 0.120842 +[2023-10-07 19:48:39,637][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29400.3/29522=99.59% | mean: 0.01 | max: 4.86 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 5.59 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] where is glasgow gla international airport? [SEP] ### +### [P_TEXT]: [CLS] glasgow airport, also unofficially glasgow international airport, formerly ### +### abbotsinch airport, is an international airport in scotland, located 8. 6 nautical miles west of ### +### glasgow city centre. in 2016, the airport handled nearly 9. 4 million passengers, a 7 % annual ### +### increase, making it the second - busiest in scotland, after edinburgh airport, and the eighth - ### +### busiest airport in the united kingdom. it is the primary airport serving the west of scotland and ### +### is the principal transatlantic and direct long [SEP] ### +### ======================================= h_v_q | Gates: 28293 ======================================= ### +### ('glasgow', 0, 0) ('g', 1, 1113) ('airport', 2, 1) ('##la', 3, 8432) ('international', 4, 11) ### +### ('edinburgh', 5, 6) ('located', 6, 96) ('scotland', 7, 5) ('scottish', 8, 13) ('is', 9, 55) ### +### ('london', 10, 73) ('manchester', 11, 47) ('.', 12, 1291) ('where', 13, 90) ('place', 14, 2295) ### +### ('hospital', 15, 95) ('airfield', 16, 8) ('situated', 17, 167) ('bother', 18, 51) ### +### ('duncan', 19, 77) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('glasgow', 0, 0) ('airport', 2, 1) ('busiest', 15840, 2) ('##sin', 8584, 3) ('abbot', 2605, 4) ### +### ('scotland', 7, 5) ('edinburgh', 5, 6) ('passengers', 4814, 7) ('airfield', 16, 8) ### +### ('##ch', 13322, 9) ('formerly', 1622, 10) ('international', 4, 11) ('passenger', 709, 12) ### +### ('scottish', 8, 13) ('nearly', 3348, 14) ('airports', 32, 15) ('million', 4774, 16) ### +### ('west', 202, 17) ('annual', 1283, 18) ('handled', 7918, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('glasgow', 0, 0) ('airport', 2, 1) ('international', 4, 11) ('edinburgh', 5, 6) ('scotland', 7, 5) ### +### ('scottish', 8, 13) ('g', 1, 1113) ('airfield', 16, 8) ('located', 6, 96) ('is', 9, 55) ### +### ('manchester', 11, 47) ('london', 10, 73) ('where', 13, 90) ('##la', 3, 8432) ('hospital', 15, 95) ### +### ('airports', 32, 15) ('bother', 18, 51) ('duncan', 19, 77) ('celtic', 25, 89) ('situated', 17, 167) ### +############################################################################################################ +[2023-10-07 19:48:39,638][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:48:39,638][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:48:40,057][root][INFO] - Epoch: 2: Step: 301/1557, loss[v]=0.118322, lr=0.000019, acc@1[1]=238.0/256=0.9296875, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 19:49:55,754][root][INFO] - Train batch 400 +[2023-10-07 19:49:55,755][root][INFO] - Avg. loss per last 100 batches: 0.117815 +[2023-10-07 19:49:56,474][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29377.9/29522=99.51% | mean: 0.01 | max: 4.74 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 6.00 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] quicklime price per ton [SEP] ### +### [P_TEXT]: [CLS] quicklime costs $ 120 per ton. lime is used in large quantities as a building and ### +### engineering material. the rocks and minerals from which these materials are derived, typically ### +### limestone or chalk, are composed primarily of calcium carbonate. they may be cut, crushed or ### +### pulverized and chemically altered. [SEP] ### +### ======================================= h_v_q | Gates: 28286 ======================================= ### +### ('##lim', 0, 4) ('ton', 1, 1) ('quick', 2, 3) ('price', 3, 14) ('per', 4, 199) ('$', 5, 25) ### +### ('##e', 6, 23) ('prices', 7, 29) ('slow', 8, 53) ('kent', 9, 190) ('quickly', 10, 85) ### +### ('##е', 11, 253) ('...', 12, 2173) ('albert', 13, 194) ('fast', 14, 30) ('swift', 15, 136) ### +### ('bare', 16, 446) ('julian', 17, 159) ('simon', 18, 87) ('sam', 19, 754) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('lime', 3646, 0) ('ton', 1, 1) ('chalk', 3289, 2) ('quick', 2, 3) ('##lim', 0, 4) ### +### ('calcium', 10120, 5) ('costs', 3144, 6) ('engineering', 2830, 7) ('carbonate', 15577, 8) ### +### ('cost', 20, 9) ('material', 4713, 10) ('minerals', 8912, 11) ('rocks', 7760, 12) ### +### ('derived', 2657, 13) ('price', 3, 14) ('used', 2329, 15) ('limestone', 718, 16) ### +### ('quantity', 720, 17) ('crushed', 1156, 18) ('mineral', 1046, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##lim', 0, 4) ('ton', 1, 1) ('quick', 2, 3) ('price', 3, 14) ('$', 5, 25) ('##e', 6, 23) ### +### ('per', 4, 199) ('prices', 7, 29) ('cost', 20, 9) ('slow', 8, 53) ('fast', 14, 30) ('tons', 21, 31) ### +### ('quickly', 10, 85) ('kent', 9, 190) ('money', 39, 41) ('simon', 18, 87) ('bother', 31, 61) ### +### ('swift', 15, 136) ('##е', 11, 253) ('helen', 23, 115) ### +############################################################################################################ +[2023-10-07 19:49:56,474][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:49:56,474][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:49:56,902][root][INFO] - Epoch: 2: Step: 401/1557, loss[v]=0.072890, lr=0.000019, acc@1[1]=241.5/256=0.943359375, acc@1[2]=248.5/256=0.970703125 +[2023-10-07 19:51:13,265][root][INFO] - Train batch 500 +[2023-10-07 19:51:13,266][root][INFO] - Avg. loss per last 100 batches: 0.122476 +[2023-10-07 19:51:13,951][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29388.6/29522=99.55% | mean: 0.01 | max: 4.64 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.5/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 5.70 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a function pointer [SEP] ### +### [P_TEXT]: [CLS] 1. int foo ( ) ; if you guessed that foo is actually a constant pointer to a ### +### function, you are correct. when a function is called ( via the ( ) operator ), the function pointer ### +### is dereferenced, and execution branches to the function. [SEP] ### +### ======================================= h_v_q | Gates: 28985 ======================================= ### +### ('pointer', 0, 1) ('function', 1, 3) ('functions', 2, 12) ('definition', 3, 20) ('is', 4, 296) ### +### ('encompasses', 5, 52) ('functional', 6, 312) ('a', 7, 5577) ('.', 8, 3258) ('job', 9, 1210) ### +### ('information', 10, 3952) ('reference', 11, 168) ('sensitive', 12, 266) ('service', 13, 5071) ### +### ('instruction', 14, 6127) (';', 15, 1183) ('operation', 16, 7245) ('capacity', 17, 8297) ### +### ('defined', 18, 323) ('baker', 19, 5074) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('foo', 6867, 0) ('pointer', 0, 1) ('constant', 3065, 2) ('function', 1, 3) ('operator', 484, 4) ### +### ('int', 16959, 5) ('der', 4235, 6) ('guessed', 3752, 7) ('via', 162, 8) ('guess', 2374, 9) ### +### ('##rence', 27732, 10) ('operators', 2716, 11) ('functions', 2, 12) ('branch', 78, 13) ### +### ('define', 14261, 14) ('branches', 15377, 15) ('call', 1053, 16) ('called', 2278, 17) ### +### ('correct', 10376, 18) ('actually', 7547, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('pointer', 0, 1) ('function', 1, 3) ('functions', 2, 12) ('definition', 3, 20) ('is', 4, 296) ### +### ('encompasses', 5, 52) ('via', 162, 8) ('branch', 78, 13) ('operator', 484, 4) ### +### ('functional', 6, 312) ('reference', 11, 168) ('warren', 45, 54) ('sensitive', 12, 266) ### +### ('hates', 62, 55) ('ছ', 382, 23) ('bare', 24, 255) ('defined', 18, 323) ('constant', 3065, 2) ### +### ('job', 9, 1210) ('thomas', 33, 325) ### +############################################################################################################ +[2023-10-07 19:51:13,951][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:51:13,951][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:51:14,373][root][INFO] - Epoch: 2: Step: 501/1557, loss[v]=0.113179, lr=0.000019, acc@1[1]=237.5/256=0.927734375, acc@1[2]=245.5/256=0.958984375 +[2023-10-07 19:52:30,240][root][INFO] - Train batch 600 +[2023-10-07 19:52:30,240][root][INFO] - Avg. loss per last 100 batches: 0.114460 +[2023-10-07 19:52:30,936][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29422.0/29522=99.66% | mean: 0.01 | max: 4.73 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.17 | max: 5.95 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how much is cuba gooding worth [SEP] ### +### [P_TEXT]: [CLS] wondering how much cuba gooding jr is worth? or maybe youare curious about cuba ### +### gooding jras salary this year? fortunately for you, weave got the details on cuba gooding jras net ### +### worth for 2018. in 2018, cuba gooding jras net worth was estimated to be $ 15 million. who is cuba ### +### gooding jr? [SEP] ### +### ======================================= h_v_q | Gates: 28502 ======================================= ### +### ('cuba', 0, 3) ('worth', 1, 1) ('$', 2, 15) ('good', 3, 11) ('##ing', 4, 47) ('cuban', 5, 14) ### +### ('much', 6, 10) ('havana', 7, 53) ('%', 8, 2869) ('familiarity', 9, 20038) ('is', 10, 5819) ### +### ('panama', 11, 166) ('iran', 12, 100) ('kent', 13, 328) ('million', 14, 16) ### +### ('definition', 15, 3393) ('julian', 16, 699) ('excellent', 17, 147) ('province', 18, 892) ### +### ('simon', 19, 194) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('jr', 11620, 0) ('worth', 1, 1) ('weave', 15040, 2) ('cuba', 0, 3) ('wondering', 2847, 4) ### +### ('salary', 11741, 5) ('net', 6702, 6) ('estimated', 4735, 7) ('price', 87, 8) ('wondered', 899, 9) ### +### ('much', 6, 10) ('good', 3, 11) ('details', 4739, 12) ('2018', 4005, 13) ('cuban', 5, 14) ### +### ('$', 2, 15) ('million', 14, 16) ('weaving', 947, 17) ('fortunately', 18483, 18) ### +### ('luckily', 16154, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('cuba', 0, 3) ('worth', 1, 1) ('$', 2, 15) ('good', 3, 11) ('cuban', 5, 14) ('much', 6, 10) ### +### ('##ing', 4, 47) ('havana', 7, 53) ('million', 14, 16) ('price', 87, 8) ('considerable', 36, 22) ### +### ('money', 22, 43) ('iran', 12, 100) ('ছ', 33, 35) ('panama', 11, 166) ('dodgers', 52, 41) ### +### ('argentina', 28, 75) ('cost', 182, 21) ('excellent', 17, 147) ('nice', 24, 135) ### +############################################################################################################ +[2023-10-07 19:52:30,936][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:52:30,936][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:52:31,340][root][INFO] - Epoch: 2: Step: 601/1557, loss[v]=0.130520, lr=0.000019, acc@1[1]=243.5/256=0.951171875, acc@1[2]=248.0/256=0.96875 +[2023-10-07 19:53:47,823][root][INFO] - Train batch 700 +[2023-10-07 19:53:47,824][root][INFO] - Avg. loss per last 100 batches: 0.124821 +[2023-10-07 19:53:48,554][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29430.5/29522=99.69% | mean: 0.01 | max: 5.25 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 5.98 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is e coli origin of replication [SEP] ### +### [P_TEXT]: [CLS] e. coli replication origin. the e. coli replication origin oric is an a240 - bp dna ### +### segment present at the start site for replication of e. coli chromosomal dna. plasmids or any other ### +### circular dna containing oric are capable of independent and controlled replication in e. coli ### +### cells.. coli replication origin. the e. coli replication origin oric is an a240 - bp dna segment ### +### present at the start site for replication of e. coli chromosomal dna. plasmids or any other ### +### circular dna containing oric are capable of independent and controlled replication in e. coli ### +### cells. [SEP] ### +### ======================================= h_v_q | Gates: 28604 ======================================= ### +### ('replication', 0, 1) ('coli', 1, 0) ('origin', 2, 2) ('e', 3, 5) ('origins', 4, 23) ### +### ('familiarity', 5, 21662) ('encompasses', 6, 25) ('of', 7, 21231) ('is', 8, 298) ('replica', 9, 49) ### +### ('definition', 10, 184) ('development', 11, 896) ('evolution', 12, 571) ('alfred', 13, 582) ### +### ('heritage', 14, 3988) ('kent', 15, 672) ('repeating', 16, 916) ('.', 17, 931) ### +### ('manufacturing', 18, 1978) ('warren', 19, 98) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('coli', 1, 0) ('replication', 0, 1) ('origin', 2, 2) ('dna', 1478, 3) ('##ic', 17796, 4) ### +### ('e', 3, 5) ('##osomal', 12419, 6) ('capable', 5948, 7) ('a2', 12360, 8) ('##40', 19793, 9) ### +### ('segment', 1003, 10) ('or', 132, 11) ('pl', 10534, 12) ('present', 3216, 13) ('circular', 736, 14) ### +### ('else', 783, 15) ('controlled', 3032, 16) ('independent', 6803, 17) ('segments', 682, 18) ### +### ('bp', 19985, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('replication', 0, 1) ('coli', 1, 0) ('origin', 2, 2) ('e', 3, 5) ('origins', 4, 23) ### +### ('encompasses', 6, 25) ('replica', 9, 49) ('ছ', 22, 27) ('definition', 10, 184) ('is', 8, 298) ### +### ('or', 132, 11) ('warren', 19, 98) ('##ο', 92, 29) ('simon', 27, 85) ('dodgers', 91, 38) ### +### ('replicate', 46, 66) ('wingspan', 196, 28) ('development', 11, 896) ('dna', 1478, 3) ### +### ('segment', 1003, 10) ### +############################################################################################################ +[2023-10-07 19:53:48,555][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:53:48,555][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:53:48,958][root][INFO] - Epoch: 2: Step: 701/1557, loss[v]=0.177256, lr=0.000018, acc@1[1]=230.5/256=0.900390625, acc@1[2]=245.0/256=0.95703125 +[2023-10-07 19:55:05,589][root][INFO] - Train batch 800 +[2023-10-07 19:55:05,591][root][INFO] - Avg. loss per last 100 batches: 0.122220 +[2023-10-07 19:55:06,286][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29425.0/29522=99.67% | mean: 0.01 | max: 4.89 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.13 | max: 5.62 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] another name for dracula [SEP] ### +### [P_TEXT]: [CLS] however, it is bram stoker's 1897 novel dracula which is remembered as the ### +### quintessential vampire novel and provided the basis of the modern vampire legend. the success of ### +### this book spawned a distinctive vampire genre, still popular in the 21st century, with books, ### +### films, and television shows. [SEP] ### +### ======================================= h_v_q | Gates: 28990 ======================================= ### +### ('dracula', 0, 0) ('name', 1, 16460) ('another', 2, 242) ('surname', 3, 8150) ('names', 4, 9007) ### +### ('nickname', 5, 5080) ('title', 6, 8375) ('vampire', 7, 2) ('renamed', 8, 1638) ### +### ('familiarity', 9, 24246) ('.', 10, 1887) ('named', 11, 8918) ('word', 12, 4357) ### +### ('identity', 13, 3421) ('nazi', 14, 4625) ('designation', 15, 5721) ('for', 16, 14737) ### +### ('vampires', 17, 13) ('brand', 18, 848) ('pseudonym', 19, 358) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('dracula', 0, 0) ('##tial', 18802, 1) ('vampire', 7, 2) ('stoke', 1346, 3) ('remembered', 2594, 4) ### +### ('novel', 1529, 5) ('legend', 2535, 6) ('##ntes', 26567, 7) ('bram', 7703, 8) ('##sen', 2783, 9) ### +### ('genre', 4934, 10) ('book', 1806, 11) ('novels', 488, 12) ('vampires', 17, 13) ### +### ('popular', 1968, 14) ('basis', 1726, 15) ('qui', 10625, 16) ('success', 3535, 17) ### +### ('however', 14008, 18) ('spawned', 15385, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('dracula', 0, 0) ('vampire', 7, 2) ('another', 2, 242) ('vampires', 17, 13) ('tigers', 29, 69) ### +### ('name', 1, 16460) ('−', 47, 80) ('novels', 488, 12) ('frankenstein', 56, 79) ('fangs', 60, 75) ### +### ('modern', 332, 22) ('stoke', 1346, 3) ('hated', 74, 70) ('surname', 3, 8150) ('horror', 20, 230) ### +### ('tuberculosis', 64, 99) ('novel', 1529, 5) ('pseudonym', 19, 358) ('fame', 394, 24) ### +### ('names', 4, 9007) ### +############################################################################################################ +[2023-10-07 19:55:06,287][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:55:06,287][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:55:06,709][root][INFO] - Epoch: 2: Step: 801/1557, loss[v]=0.131200, lr=0.000018, acc@1[1]=234.0/256=0.9140625, acc@1[2]=242.0/256=0.9453125 +[2023-10-07 19:56:22,930][root][INFO] - Train batch 900 +[2023-10-07 19:56:22,931][root][INFO] - Avg. loss per last 100 batches: 0.117502 +[2023-10-07 19:56:23,659][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29415.2/29522=99.64% | mean: 0.01 | max: 4.59 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 5.74 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what can cause a hemorrhage over the optic nerve [SEP] ### +### [P_TEXT]: [CLS] a neurological exam or eye exam, which can show swelling of the optic nerve, may ### +### also be performed. a lumbar puncture is usually not performed, as it may be dangerous and make ### +### things worse. treatment for bleeding in the brain depends on the location, cause, and extent of the ### +### hemorrhage. [SEP] ### +### ======================================= h_v_q | Gates: 28903 ======================================= ### +### ('optic', 0, 10) ('hem', 1, 17) ('##or', 2, 72) ('nerve', 3, 5) ('over', 4, 9646) ('##age', 5, 38) ### +### ('familiarity', 6, 21485) ('##rh', 7, 70) ('cause', 8, 826) ('caused', 9, 4598) ('can', 10, 1130) ### +### ('couldn', 11, 191) ('causes', 12, 332) ('kent', 13, 134) ('optical', 14, 63) ('warren', 15, 102) ### +### ('could', 16, 673) ('##ors', 17, 463) ('sam', 18, 2197) ('ammunition', 19, 204) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('bleeding', 5435, 0) ('##cture', 14361, 1) ('swelling', 19424, 2) ('exam', 14280, 3) ### +### ('brain', 2322, 4) ('nerve', 3, 5) ('neurological', 14538, 6) ('dangerous', 1455, 7) ### +### ('lu', 4326, 8) ('eye', 2888, 9) ('optic', 0, 10) ('##mba', 25310, 11) ('worse', 10723, 12) ### +### ('depends', 16520, 13) ('swell', 17354, 14) ('not', 18347, 15) ('pun', 7754, 16) ('hem', 1, 17) ### +### ('treatment', 7803, 18) ('extent', 10416, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('optic', 0, 10) ('hem', 1, 17) ('nerve', 3, 5) ('##or', 2, 72) ('##age', 5, 38) ('##rh', 7, 70) ### +### ('cause', 8, 826) ('optical', 14, 63) ('kent', 13, 134) ('warren', 15, 102) ('couldn', 11, 191) ### +### ('causes', 12, 332) ('wingspan', 39, 23) ('over', 4, 9646) ('ruins', 61, 27) ('ছ', 41, 37) ### +### ('can', 10, 1130) ('caused', 9, 4598) ('bother', 36, 48) ('afraid', 37, 53) ### +############################################################################################################ +[2023-10-07 19:56:23,660][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:56:23,660][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:56:24,083][root][INFO] - Epoch: 2: Step: 901/1557, loss[v]=0.082952, lr=0.000018, acc@1[1]=239.0/256=0.93359375, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 19:57:40,513][root][INFO] - Train batch 1000 +[2023-10-07 19:57:40,514][root][INFO] - Avg. loss per last 100 batches: 0.113435 +[2023-10-07 19:57:41,210][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29425.1/29522=99.67% | mean: 0.01 | max: 5.39 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 6.29 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how do i know if my tax refund will be garnished for student loans [SEP] ### +### [P_TEXT]: [CLS] the department of education is authorized to seize up to 10 percent of disposable ### +### earnings to repay defaulted federal student loans. for private loans, up to 25 percent of ### +### disposable income may be garnished. if your tax return is garnished, your lender will send you a ### +### notice that a claim has been filed against you. the notice includes : 1 your defaulted loan amount. ### +### 2 your rights as a student loan borrower. 3 how you can avoid tax offset. f your tax return is ### +### garnished, your lender will send you a notice that a claim has been filed against you. the notice ### +### includes : 1 your defaulted loan amount. 2 your rights as a student loan borrower. 3 how you can ### +### avoid tax offset. [SEP] ### +### ======================================= h_v_q | Gates: 28873 ======================================= ### +### ('tax', 0, 4) ('##ished', 1, 35) ('ga', 2, 3) ('student', 3, 24) ('knew', 4, 179) ('ref', 5, 23743) ### +### ('loans', 6, 37) ('familiarity', 7, 16250) ('loan', 8, 14) ('##und', 9, 16654) ('##rn', 10, 49) ### +### ('know', 11, 1303) ('knowing', 12, 623) ('knowledge', 13, 9113) ('students', 14, 42) ### +### ('my', 15, 111) ('teacher', 16, 366) ('##ishing', 17, 2734) ('knows', 18, 2984) ('fame', 19, 507) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('notice', 7017, 0) ('avoid', 1392, 1) ('borrow', 3637, 2) ('ga', 2, 3) ('tax', 0, 4) ### +### ('seize', 17935, 5) ('department', 3351, 6) ('percent', 4287, 7) ('offset', 17975, 8) ### +### ('income', 2762, 9) ('private', 2609, 10) ('authorized', 7537, 11) ('claim', 5990, 12) ### +### ('repay', 12548, 13) ('loan', 8, 14) ('seized', 14665, 15) ('borrowed', 777, 16) ### +### ('seizing', 21241, 17) ('federal', 4399, 18) ('education', 1045, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('tax', 0, 4) ('ga', 2, 3) ('##ished', 1, 35) ('student', 3, 24) ('loans', 6, 37) ('loan', 8, 14) ### +### ('knew', 4, 179) ('##rn', 10, 49) ('students', 14, 42) ('my', 15, 111) ('##元', 22, 29) ### +### ('know', 11, 1303) ('knowing', 12, 623) ('taxes', 29, 51) ('wingspan', 26, 56) ('tired', 27, 55) ### +### ('ˈ', 37, 57) ('##ο', 72, 43) ('teacher', 16, 366) ('bother', 31, 120) ### +############################################################################################################ +[2023-10-07 19:57:41,210][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:57:41,210][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:57:41,634][root][INFO] - Epoch: 2: Step: 1001/1557, loss[v]=0.101248, lr=0.000018, acc@1[1]=241.5/256=0.943359375, acc@1[2]=245.0/256=0.95703125 +[2023-10-07 19:58:57,699][root][INFO] - Train batch 1100 +[2023-10-07 19:58:57,700][root][INFO] - Avg. loss per last 100 batches: 0.113905 +[2023-10-07 19:58:58,418][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29382.9/29522=99.53% | mean: 0.01 | max: 4.97 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 5.88 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] bakersfield animal control phone number [SEP] ### +### [P_TEXT]: [CLS] visit animal control bakersfield on the given address : 1601 truxtun ave, ### +### bakersfield, ca 93301, united states contact animal control bakersfield on the given contact number ### +### : + 1 661 - 326 - 3436. if the contact number or email address of animal control bakersfield is ### +### incorrect, please tell us here send animal control bakersfield email on given email address : open ### +### animal control bakersfield website by given website address : [SEP] ### +### ======================================= h_v_q | Gates: 28017 ======================================= ### +### ('##sfield', 0, 1) ('baker', 1, 2) ('animal', 2, 0) ('phone', 3, 4278) ('control', 4, 7) ### +### ('number', 5, 11) ('animals', 6, 6) ('familiarity', 7, 7936) ('telephone', 8, 10093) ### +### ('smith', 9, 1101) ('beast', 10, 93) ('numbers', 11, 23) ('kent', 12, 189) ('radio', 13, 9944) ### +### (';', 14, 3802) ('controlled', 15, 41) ('.', 16, 4505) ('taylor', 17, 1062) ('##ト', 18, 143) ### +### ('dog', 19, 125) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('animal', 2, 0) ('##sfield', 0, 1) ('baker', 1, 2) ('tr', 7773, 3) ('##1', 504, 4) ### +### ('address', 493, 5) ('animals', 6, 6) ('control', 4, 7) ('+', 3937, 8) ('visit', 7875, 9) ### +### ('contact', 811, 10) ('number', 5, 11) ('##ux', 17554, 12) ('incorrect', 8859, 13) ### +### ('##30', 6196, 14) ('email', 6423, 15) ('addresses', 5408, 16) ('open', 126, 17) ('##6', 3950, 18) ### +### ('ca', 1734, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##sfield', 0, 1) ('baker', 1, 2) ('animal', 2, 0) ('control', 4, 7) ('number', 5, 11) ### +### ('animals', 6, 6) ('phone', 3, 4278) ('numbers', 11, 23) ('controlled', 15, 41) ('beast', 10, 93) ### +### ('##ο', 61, 24) ('open', 126, 17) ('afraid', 56, 36) ('##1', 504, 4) ('kent', 12, 189) ### +### ('bother', 51, 52) ('address', 493, 5) ('warren', 20, 112) ('ruins', 72, 42) ('dog', 19, 125) ### +############################################################################################################ +[2023-10-07 19:58:58,418][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 19:58:58,418][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 19:58:58,845][root][INFO] - Epoch: 2: Step: 1101/1557, loss[v]=0.062555, lr=0.000018, acc@1[1]=241.0/256=0.94140625, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 20:00:15,311][root][INFO] - Train batch 1200 +[2023-10-07 20:00:15,311][root][INFO] - Avg. loss per last 100 batches: 0.116213 +[2023-10-07 20:00:15,994][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29385.4/29522=99.54% | mean: 0.01 | max: 5.06 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.15 | max: 5.97 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is common cold pathogen [SEP] ### +### [P_TEXT]: [CLS] the common cold ( also known as nasopharyngitis, rhinopharyngitis, acute coryza, ### +### head cold, or simply a cold ) is a viral infectious disease of the upper respiratory tract which ### +### primarily affects the nose. he common cold may occasionally lead to pneumonia, either viral ### +### pneumonia or secondary bacterial pneumonia. no cure for the common cold exists, but the symptoms ### +### can be treated. [SEP] ### +### ======================================= h_v_q | Gates: 28341 ======================================= ### +### ('pathogen', 0, 2422) ('cold', 1, 0) ('common', 2, 2) ('familiarity', 3, 16529) ### +### ('encompasses', 4, 14) ('is', 5, 649) ('definition', 6, 39) ('outbreak', 7, 113) ('cool', 8, 32) ### +### ('frequent', 9, 40) ('hot', 10, 157) ('.', 11, 8005) ('warm', 12, 27) ('ammunition', 13, 344) ### +### ('heat', 14, 103) ('host', 15, 100) ('systematic', 16, 345) ('alfred', 17, 412) ### +### ('something', 18, 1679) ('commonly', 19, 36) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('cold', 1, 0) ('nas', 2180, 1) ('common', 2, 2) ('nose', 1724, 3) ('head', 6809, 4) ### +### ('pneumonia', 1032, 5) ('acute', 4709, 6) ('respiratory', 7597, 7) ('##za', 11474, 8) ### +### ('rhino', 27680, 9) ('viral', 284, 10) ('tract', 7295, 11) ('cory', 18596, 12) ('disease', 26, 13) ### +### ('encompasses', 4, 14) ('heads', 16228, 15) ('infectious', 4584, 16) ('cure', 7241, 17) ### +### ('##git', 28808, 18) ('bacterial', 673, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('cold', 1, 0) ('common', 2, 2) ('encompasses', 4, 14) ('pathogen', 0, 2422) ('definition', 6, 39) ### +### ('cool', 8, 32) ('frequent', 9, 40) ('outbreak', 7, 113) ('disease', 26, 13) ('warm', 12, 27) ### +### ('frost', 29, 22) ('hot', 10, 157) ('commonly', 19, 36) ('is', 5, 649) ('host', 15, 100) ### +### ('heat', 14, 103) ('viral', 284, 10) ('widespread', 20, 93) ('colder', 98, 25) ('−', 37, 60) ### +############################################################################################################ +[2023-10-07 20:00:15,995][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:00:15,995][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:00:16,398][root][INFO] - Epoch: 2: Step: 1201/1557, loss[v]=0.111024, lr=0.000018, acc@1[1]=238.5/256=0.931640625, acc@1[2]=247.0/256=0.96484375 +[2023-10-07 20:01:33,052][root][INFO] - Train batch 1300 +[2023-10-07 20:01:33,053][root][INFO] - Avg. loss per last 100 batches: 0.118470 +[2023-10-07 20:01:33,773][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29357.5/29522=99.44% | mean: 0.01 | max: 5.00 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 5.76 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what county is radcliff, ky [SEP] ### +### [P_TEXT]: [CLS] radcliff, kentucky. radcliff is a home rule - class city in hardin county, ### +### kentucky, in the united states. the population was 21, 961 during the year 2000 u. s. census. it is ### +### included in the elizabethtown metropolitan area. its economy is largely dominated by the adjacent ### +### u. s. army base fort knox, and by the nearby city of elizabethtown. [SEP] ### +### ======================================= h_v_q | Gates: 28710 ======================================= ### +### ('county', 0, 12) ('ra', 1, 1) ('##dc', 2, 6) ('##li', 3, 17) ('##ff', 4, 2) ('ky', 5, 13) ### +### ('kentucky', 6, 0) ('familiarity', 7, 18126) ('indiana', 8, 99) ('ohio', 9, 685) ### +### ('missouri', 10, 1646) ('counties', 11, 34) ('##f', 12, 51) ('illinois', 13, 1038) ('is', 14, 102) ### +### ('parish', 15, 257) ('rural', 16, 399) ('pennsylvania', 17, 1143) ('municipal', 18, 302) ### +### ('.', 19, 1825) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('kentucky', 6, 0) ('ra', 1, 1) ('##ff', 4, 2) ('hardin', 22204, 3) ('##town', 5121, 4) ### +### ('knox', 596, 5) ('##dc', 2, 6) ('elizabeth', 2871, 7) ('population', 936, 8) ### +### ('metropolitan', 3861, 9) ('nearby', 478, 10) ('rule', 7109, 11) ('county', 0, 12) ('ky', 5, 13) ### +### ('economy', 4666, 14) ('class', 2061, 15) ('21', 3299, 16) ('##li', 3, 17) ('2000', 2060, 18) ### +### ('20', 5815, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ra', 1, 1) ('##dc', 2, 6) ('county', 0, 12) ('##ff', 4, 2) ('kentucky', 6, 0) ('##li', 3, 17) ### +### ('ky', 5, 13) ('counties', 11, 34) ('indiana', 8, 99) ('##f', 12, 51) ('afraid', 72, 27) ### +### ('is', 14, 102) ('encompasses', 20, 78) ('city', 113, 22) ('ohio', 9, 685) ('knox', 596, 5) ### +### ('crashed', 66, 42) ('−', 81, 33) ('nearby', 478, 10) ('ˈ', 74, 41) ### +############################################################################################################ +[2023-10-07 20:01:33,773][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:01:33,773][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:01:34,178][root][INFO] - Epoch: 2: Step: 1301/1557, loss[v]=0.122519, lr=0.000018, acc@1[1]=236.5/256=0.923828125, acc@1[2]=247.0/256=0.96484375 +[2023-10-07 20:02:51,873][root][INFO] - Train batch 1400 +[2023-10-07 20:02:51,874][root][INFO] - Avg. loss per last 100 batches: 0.115277 +[2023-10-07 20:02:52,580][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29305.5/29522=99.27% | mean: 0.01 | max: 5.04 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 6.06 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] list of current jewish athletes [SEP] ### +### [P_TEXT]: [CLS] carew has never actually converted to judaism formally, although he did marry a ### +### jewish woman and his children were raised in the jewish tradition. in honor of the jewish holiday, ### +### this is a list of the top 15 current jewish athletes in sports. amar'e stoudemire. stoudemire has ### +### to be considered the top jewish player in the nba after signing a $ 100 million contract during the ### +### offseason with the new york knicks. [SEP] ### +### ======================================= h_v_q | Gates: 27538 ======================================= ### +### ('jewish', 0, 1) ('athletes', 1, 4) ('current', 2, 20) ('list', 3, 76) ('jews', 4, 6) ### +### ('athlete', 5, 18) ('familiarity', 6, 21881) ('hebrew', 7, 44) ('synagogue', 8, 36) ('.', 9, 3476) ### +### ('olympic', 10, 111) ('athletics', 11, 97) ('israel', 12, 39) ('currently', 13, 302) ### +### ('players', 14, 41) ('former', 15, 517) ('team', 16, 514) ('men', 17, 6573) ('games', 18, 527) ### +### ('people', 19, 4542) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('amar', 21326, 0) ('jewish', 0, 1) ('knicks', 23475, 2) ('judaism', 112, 3) ('athletes', 1, 4) ### +### ('care', 8655, 5) ('jews', 4, 6) ('sports', 29, 7) ('never', 1350, 8) ('nba', 8715, 9) ### +### ('converted', 4052, 10) ('##w', 8656, 11) ('offs', 7334, 12) ('conversion', 7562, 13) ### +### ('holiday', 13463, 14) ('marry', 11991, 15) ('player', 643, 16) ('married', 360, 17) ### +### ('athlete', 5, 18) ('tradition', 993, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('jewish', 0, 1) ('athletes', 1, 4) ('current', 2, 20) ('jews', 4, 6) ('athlete', 5, 18) ### +### ('list', 3, 76) ('synagogue', 8, 36) ('hebrew', 7, 44) ('israel', 12, 39) ('sports', 29, 7) ### +### ('players', 14, 41) ('olympic', 10, 111) ('athletics', 11, 97) ('judaism', 112, 3) ### +### ('israeli', 20, 65) ('rabbi', 65, 33) ('currently', 13, 302) ('−', 53, 48) ('muslim', 25, 139) ### +### ('elite', 47, 87) ### +############################################################################################################ +[2023-10-07 20:02:52,581][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:02:52,581][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:02:53,004][root][INFO] - Epoch: 2: Step: 1401/1557, loss[v]=0.147837, lr=0.000018, acc@1[1]=240.0/256=0.9375, acc@1[2]=247.0/256=0.96484375 +[2023-10-07 20:04:10,219][root][INFO] - Train batch 1500 +[2023-10-07 20:04:10,220][root][INFO] - Avg. loss per last 100 batches: 0.114740 +[2023-10-07 20:04:10,927][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29338.2/29522=99.38% | mean: 0.01 | max: 4.50 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 5.70 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] can apache openoffice save in *. docx format? [SEP] ### +### [P_TEXT]: [CLS] edit answer ( for another - 22 minute ) it is possible to open the docx document in ### +### word and use word to save it as a pdf. there are also other office suites that can open the docx ### +### format and save it as a pdf. look at apache openoffice. org. it appears you are trying to open the ### +### docx with adobe reader. adobe reader can only open pdf files. [SEP] ### +### ======================================= h_v_q | Gates: 28406 ======================================= ### +### ('apache', 0, 2) ('open', 1, 5) ('##ice', 2, 36) ('##off', 3, 14) ('##x', 4, 7) ('save', 5, 12) ### +### ('doc', 6, 0) ('format', 7, 9) ('familiarity', 8, 20510) ('*', 9, 18018) ('opened', 10, 76) ### +### ('arizona', 11, 749) ('can', 12, 1387) ('able', 13, 22) ('.', 14, 7223) ('couldn', 15, 61) ### +### ('wearing', 16, 194) ('?', 17, 26590) ('x', 18, 281) ('off', 19, 965) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('doc', 6, 0) ('pdf', 18022, 1) ('apache', 0, 2) ('answer', 2290, 3) ('edit', 10685, 4) ### +### ('open', 1, 5) ('document', 4156, 6) ('##x', 4, 7) ('adobe', 10850, 8) ('format', 7, 9) ### +### ('suites', 23629, 10) ('suite', 8285, 11) ('save', 5, 12) ('office', 2061, 13) ('##off', 3, 14) ### +### ('minute', 3836, 15) ('documents', 12777, 16) ('minutes', 2065, 17) ('word', 5159, 18) ### +### ('files', 7067, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('apache', 0, 2) ('open', 1, 5) ('doc', 6, 0) ('##x', 4, 7) ('##off', 3, 14) ('save', 5, 12) ### +### ('format', 7, 9) ('##ice', 2, 36) ('able', 13, 22) ('opened', 10, 76) ('afraid', 20, 20) ### +### ('couldn', 15, 61) ('ˈ', 29, 27) ('crashing', 39, 25) ('formats', 41, 29) ('wingspan', 50, 26) ### +### ('ছ', 42, 40) ('−', 28, 66) ('unwilling', 36, 43) ('hugh', 67, 31) ### +############################################################################################################ +[2023-10-07 20:04:10,927][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:04:10,927][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:04:11,351][root][INFO] - Epoch: 2: Step: 1501/1557, loss[v]=0.129842, lr=0.000018, acc@1[1]=237.0/256=0.92578125, acc@1[2]=242.5/256=0.947265625 +[2023-10-07 20:04:55,436][root][INFO] - rank=3; last iteration 1557 +[2023-10-07 20:04:55,437][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 20:04:55,437][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-07 20:04:55,441][root][INFO] - rank=1; last iteration 1557 +[2023-10-07 20:04:55,441][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 20:04:55,441][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-07 20:04:55,442][root][INFO] - rank=0; last iteration 1557 +[2023-10-07 20:04:55,443][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 20:04:55,443][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-07 20:04:55,443][root][INFO] - rank=2; last iteration 1557 +[2023-10-07 20:04:55,444][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 20:04:55,444][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-07 20:04:55,445][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 20:04:55,446][root][INFO] - Epoch finished on 3 +[2023-10-07 20:04:55,449][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 20:04:55,449][root][INFO] - Epoch finished on 1 +[2023-10-07 20:04:55,450][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 20:04:55,450][root][INFO] - Epoch finished on 0 +[2023-10-07 20:04:55,452][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 20:04:55,452][root][INFO] - Epoch finished on 2 +[2023-10-07 20:05:42,656][root][INFO] - Saved checkpoint at ./vdr_2 +[2023-10-07 20:05:42,656][root][INFO] - Saved checkpoint at ./vdr_2 +[2023-10-07 20:05:42,657][root][INFO] - Av Loss per epoch=0.118200 +[2023-10-07 20:05:42,657][root][INFO] - Av Loss per epoch=0.118200 +[2023-10-07 20:05:42,657][root][INFO] - epoch total (1) correct predictions=371306 +[2023-10-07 20:05:42,657][root][INFO] - epoch total (1) correct predictions=371306 +[2023-10-07 20:05:42,657][root][INFO] - Saved checkpoint at ./vdr_2 +[2023-10-07 20:05:42,657][root][INFO] - epoch total (2) correct predictions=384758 +[2023-10-07 20:05:42,657][root][INFO] - epoch total (2) correct predictions=384758 +[2023-10-07 20:05:42,657][root][INFO] - Av Loss per epoch=0.118200 +[2023-10-07 20:05:42,658][root][INFO] - epoch total (1) correct predictions=371306 +[2023-10-07 20:05:42,658][root][INFO] - epoch total (2) correct predictions=384758 +[2023-10-07 20:05:42,658][root][INFO] - Saved checkpoint at ./vdr_2 +[2023-10-07 20:05:42,659][root][INFO] - Av Loss per epoch=0.118200 +[2023-10-07 20:05:42,659][root][INFO] - epoch total (1) correct predictions=371306 +[2023-10-07 20:05:42,659][root][INFO] - epoch total (2) correct predictions=384758 +[2023-10-07 20:05:42,661][root][INFO] - ***** Epoch 3 ***** +[2023-10-07 20:05:42,661][root][INFO] - ***** Epoch 3 ***** +[2023-10-07 20:05:42,663][root][INFO] - ***** Epoch 3 ***** +[2023-10-07 20:05:42,667][root][INFO] - rank=1; Iteration start +[2023-10-07 20:05:42,667][root][INFO] - rank=0; Iteration start +[2023-10-07 20:05:42,668][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 20:05:42,668][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 20:05:42,668][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 20:05:42,668][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 20:05:42,667][root][INFO] - ***** Epoch 3 ***** +[2023-10-07 20:05:42,669][root][INFO] - rank=3; Iteration start +[2023-10-07 20:05:42,669][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 20:05:42,669][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-07 20:05:42,669][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 20:05:42,669][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-07 20:05:42,671][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-07 20:05:42,673][root][INFO] - rank=2; Iteration start +[2023-10-07 20:05:42,673][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 20:05:42,673][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 20:05:42,675][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-07 20:05:43,646][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29311.4/29522=99.29% | mean: 0.01 | max: 4.81 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 5.50 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] akron children's hospital number [SEP] ### +### [P_TEXT]: [CLS] akron childrens hospital address and contact number akron childrens hospital ### +### contact phone number is : 330 - 543 - 1000 and address is one perkins square, akron, ohio, united ### +### states of america akron children's hospital is a children hospital situated in akron, ohio. akron ### +### children's hospital is has branches in 78 locations and is the largest pediatric center in ### +### northeast ohio. the hospital has all the medical facilities such as trauma center and intensive ### +### care center for child's and adults. [SEP] ### +### ======================================= h_v_q | Gates: 28094 ======================================= ### +### ('akron', 0, 0) ('hospital', 1, 2) ('children', 2, 3) ('number', 3, 21) ('hospitals', 4, 6) ### +### ('familiarity', 5, 18185) ('child', 6, 7) ('son', 7, 167) ('.', 8, 2553) ('toledo', 9, 312) ### +### ('cemetery', 10, 329) ('trail', 11, 209) ('ohio', 12, 4) ('kids', 13, 34) ('s', 14, 1668) ### +### ('health', 15, 696) ('medical', 16, 37) ('elevation', 17, 156) ('hotel', 18, 32) ### +### ('cleveland', 19, 30) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('akron', 0, 0) ('perkins', 3019, 1) ('hospital', 1, 2) ('children', 2, 3) ('ohio', 12, 4) ### +### ('contact', 3402, 5) ('hospitals', 4, 6) ('child', 6, 7) ('address', 3979, 8) ('##3', 6422, 9) ### +### ('1000', 1810, 10) ('pediatric', 9636, 11) ('largest', 968, 12) ('situated', 3407, 13) ### +### ('northeast', 2706, 14) ('trauma', 3486, 15) ('square', 912, 16) ('phone', 236, 17) ### +### ('center', 169, 18) ('addresses', 10383, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('akron', 0, 0) ('hospital', 1, 2) ('children', 2, 3) ('number', 3, 21) ('hospitals', 4, 6) ### +### ('child', 6, 7) ('ohio', 12, 4) ('kids', 13, 34) ('medical', 16, 37) ('hotel', 18, 32) ### +### ('cleveland', 19, 30) ('son', 7, 167) ('numbers', 22, 52) ('trail', 11, 209) ('elevation', 17, 156) ### +### ('toledo', 9, 312) ('cemetery', 10, 329) ('afraid', 83, 22) ('−', 38, 70) ('healthcare', 56, 42) ### +############################################################################################################ +[2023-10-07 20:05:43,647][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:05:43,647][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:05:44,043][root][INFO] - Epoch: 3: Step: 1/1557, loss[v]=0.120999, lr=0.000018, acc@1[1]=240.5/256=0.939453125, acc@1[2]=244.0/256=0.953125 +[2023-10-07 20:07:00,220][root][INFO] - Train batch 100 +[2023-10-07 20:07:00,221][root][INFO] - Avg. loss per last 100 batches: 0.107578 +[2023-10-07 20:07:00,943][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29397.1/29522=99.58% | mean: 0.01 | max: 5.06 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 5.92 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] do pizza hut delivery drivers get part of the delivery fee? [SEP] ### +### [P_TEXT]: [CLS] the suit alleges that pizza hut's mandatory delivery fees give customers the ### +### impression that they are part of the tip, but drivers receive no portion of it, says the associated ### +### press. he hut isn't the only pizza chain to get backlash from employees on shady fee practices : ### +### last year, u. k. chain pizza express came under fire for keeping 8 percent of its workers'credit ### +### card tips, which it claimed was to cover credit card processing fees. [SEP] ### +### ======================================= h_v_q | Gates: 28232 ======================================= ### +### ('pizza', 0, 1) ('hut', 1, 0) ('delivery', 2, 6) ('drivers', 3, 17) ('part', 4, 18) ('fee', 5, 2) ### +### ('$', 6, 3136) ('familiarity', 7, 17699) ('got', 8, 208) ('driver', 9, 21) ('get', 10, 252) ### +### ('gets', 11, 420) ('delivered', 12, 65) ('tax', 13, 1844) ('pilots', 14, 141) ('grant', 15, 297) ### +### ('driving', 16, 69) ('price', 17, 70) ('pilot', 18, 364) ('parts', 19, 133) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('hut', 1, 0) ('pizza', 0, 1) ('fee', 5, 2) ('tip', 2539, 3) ('express', 6174, 4) ### +### ('mandatory', 7953, 5) ('delivery', 2, 6) ('fees', 36, 7) ('backlash', 10423, 8) ('chain', 1618, 9) ### +### ('associated', 6151, 10) ('impression', 4254, 11) ('shady', 27412, 12) ('lawsuit', 1346, 13) ### +### ('suit', 14673, 14) ('card', 4209, 15) ('practices', 7297, 16) ('drivers', 3, 17) ('part', 4, 18) ### +### ('credit', 170, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('hut', 1, 0) ('pizza', 0, 1) ('delivery', 2, 6) ('fee', 5, 2) ('drivers', 3, 17) ('part', 4, 18) ### +### ('driver', 9, 21) ('got', 8, 208) ('delivered', 12, 65) ('get', 10, 252) ('fees', 36, 7) ### +### ('driving', 16, 69) ('price', 17, 70) ('pilots', 14, 141) ('$', 6, 3136) ('gets', 11, 420) ### +### ('parts', 19, 133) ('pie', 22, 96) ('vehicles', 20, 167) ('grant', 15, 297) ### +############################################################################################################ +[2023-10-07 20:07:00,944][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:07:00,944][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:07:01,373][root][INFO] - Epoch: 3: Step: 101/1557, loss[v]=0.069595, lr=0.000018, acc@1[1]=240.5/256=0.939453125, acc@1[2]=254.0/256=0.9921875 +[2023-10-07 20:08:18,263][root][INFO] - Train batch 200 +[2023-10-07 20:08:18,263][root][INFO] - Avg. loss per last 100 batches: 0.107572 +[2023-10-07 20:08:18,952][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29372.3/29522=99.49% | mean: 0.01 | max: 4.80 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.17 | max: 6.00 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a literature review in sociology [SEP] ### +### [P_TEXT]: [CLS] what is a literature review? a literature review is an account of published ### +### research by accredited scholars and researchers. in your review you will show the relevance of ### +### previously published research to your topic and explain how your research fits in to the larger ### +### field of study. [SEP] ### +### ======================================= h_v_q | Gates: 28023 ======================================= ### +### ('review', 0, 1) ('literature', 1, 0) ('sociology', 2, 6733) ('definition', 3, 12) ### +### ('encompasses', 4, 4) ('reviews', 5, 8) ('a', 6, 498) ('journal', 7, 826) ('.', 8, 4627) ### +### ('familiarity', 9, 22049) ('social', 10, 13698) ('anthropology', 11, 11008) ('is', 12, 130) ### +### ('sociologist', 13, 10974) ('chicago', 14, 350) ('noun', 15, 14150) ('defined', 16, 639) ### +### ('philosophy', 17, 3399) ('something', 18, 686) ('reviewing', 19, 10) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('literature', 1, 0) ('review', 0, 1) ('accredited', 17170, 2) ('research', 529, 3) ### +### ('encompasses', 4, 4) ('account', 5380, 5) ('define', 11741, 6) ('scholars', 2610, 7) ### +### ('reviews', 5, 8) ('researchers', 6983, 9) ('reviewing', 19, 10) ('relevance', 10861, 11) ### +### ('definition', 3, 12) ('scholar', 2205, 13) ('previously', 9101, 14) ('crashing', 231, 15) ### +### ('−', 32, 16) ('definitions', 1688, 17) ('published', 744, 18) ('reviewer', 125, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('literature', 1, 0) ('review', 0, 1) ('definition', 3, 12) ('encompasses', 4, 4) ('reviews', 5, 8) ### +### ('sociology', 2, 6733) ('reviewing', 19, 10) ('a', 6, 498) ('−', 32, 16) ('is', 12, 130) ### +### ('criticism', 30, 22) ('journal', 7, 826) ('study', 35, 25) ('critic', 38, 31) ('reviewed', 41, 34) ### +### ('hugh', 52, 30) ('what', 27, 59) ('research', 529, 3) ('afraid', 71, 28) ('wingspan', 88, 24) ### +############################################################################################################ +[2023-10-07 20:08:18,953][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:08:18,953][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:08:19,374][root][INFO] - Epoch: 3: Step: 201/1557, loss[v]=0.153889, lr=0.000018, acc@1[1]=233.5/256=0.912109375, acc@1[2]=245.5/256=0.958984375 +[2023-10-07 20:09:35,965][root][INFO] - Train batch 300 +[2023-10-07 20:09:35,966][root][INFO] - Avg. loss per last 100 batches: 0.103725 +[2023-10-07 20:09:36,661][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29382.8/29522=99.53% | mean: 0.01 | max: 5.40 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.13 | max: 6.12 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] definitive mutation [SEP] ### +### [P_TEXT]: [CLS] a case study of coevolution. mutations. mutation is a change in dna, the hereditary ### +### material of life. an organism's dna affects how it looks, how it behaves, and its physiology a all ### +### aspects of its life. so a change in an organism's dna can cause changes in all aspects of its life. ### +### mutations are random. mutations can be beneficial, neutral, or harmful for the organism, but ### +### mutations do not try to supply what the organism needs.. in this respect, mutations are random a ### +### whether a particular mutation happens or not is unrelated to how useful that mutation would be. ### +### [SEP] ### +### ======================================= h_v_q | Gates: 28303 ======================================= ### +### ('definitive', 0, 12339) ('mutation', 1, 0) ('mutations', 2, 1) ('or', 3, 14995) ('gene', 4, 157) ### +### ('familiarity', 5, 24849) (';', 6, 6885) ('.', 7, 6107) ('ultimate', 8, 3573) ('variation', 9, 134) ### +### ('steel', 10, 2732) ('authoritative', 11, 21454) ('london', 12, 2399) ('authority', 13, 12155) ### +### ('striking', 14, 13332) ('descendant', 15, 3362) ('complete', 16, 2730) ('breeding', 17, 115) ### +### ('british', 18, 9925) ('variant', 19, 317) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('mutation', 1, 0) ('mutations', 2, 1) ('hereditary', 6531, 2) ('dna', 2562, 3) ('coe', 10956, 4) ### +### ('organism', 8753, 5) ('define', 22611, 6) ('change', 43, 7) ('definition', 859, 8) ### +### ('encompasses', 1041, 9) ('random', 585, 10) ('aspects', 1585, 11) ('life', 1022, 12) ### +### ('##tion', 1546, 13) ('useful', 6412, 14) ('crashing', 74, 15) ('##vo', 22570, 16) ### +### ('neutral', 1109, 17) ('dangerous', 4944, 18) ('unrelated', 6036, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('mutation', 1, 0) ('mutations', 2, 1) ('definitive', 0, 12339) ('gene', 4, 157) ('change', 43, 7) ### +### ('wingspan', 20, 21) ('variation', 9, 134) ('crashing', 74, 15) ('genetic', 54, 32) ### +### ('crashed', 51, 43) ('breeding', 17, 115) ('−', 172, 20) ('random', 585, 10) ('definition', 859, 8) ### +### ('hugh', 167, 46) ('mutant', 79, 92) ('##ο', 318, 30) ('encompasses', 1041, 9) ### +### ('evolution', 63, 113) ('julia', 106, 94) ### +############################################################################################################ +[2023-10-07 20:09:36,661][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:09:36,661][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:09:37,066][root][INFO] - Epoch: 3: Step: 301/1557, loss[v]=0.120207, lr=0.000018, acc@1[1]=242.0/256=0.9453125, acc@1[2]=248.5/256=0.970703125 +[2023-10-07 20:10:53,749][root][INFO] - Train batch 400 +[2023-10-07 20:10:53,750][root][INFO] - Avg. loss per last 100 batches: 0.104421 +[2023-10-07 20:10:54,451][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29364.5/29522=99.47% | mean: 0.01 | max: 5.02 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 5.88 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] who is luke hochevar playing for [SEP] ### +### [P_TEXT]: [CLS] luke hochevar. luke anthony hochevar ( / ehoeteeeªver / ; born september 15, 1983 ) ### +### is an american professional baseball pitcher who is a free agent. he played college baseball at the ### +### university of tennessee, and has played in major league baseball ( mlb ) for the kansas city ### +### royals. 1 early life. [SEP] ### +### ======================================= h_v_q | Gates: 28479 ======================================= ### +### ('hoc', 0, 4) ('luke', 1, 1) ('playing', 2, 45) ('##var', 3, 9) ('##he', 4, 13) ('played', 5, 29) ### +### ('who', 6, 21) ('familiarity', 7, 22033) ('player', 8, 89) ('is', 9, 106) ('coaching', 10, 656) ### +### ('play', 11, 75) ('wingspan', 12, 64) ('thinking', 13, 1789) ('operating', 14, 4111) ### +### ('league', 15, 31) ('encompasses', 16, 22) ('wearing', 17, 219) ('game', 18, 904) ### +### ('definition', 19, 477) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##ª', 17933, 0) ('luke', 1, 1) ('anthony', 339, 2) ('tennessee', 1130, 3) ('hoc', 0, 4) ### +### ('pitcher', 10590, 5) ('royals', 16681, 6) ('baseball', 267, 7) ('mlb', 8714, 8) ('##var', 3, 9) ### +### ('kansas', 5500, 10) ('##oe', 9417, 11) ('college', 997, 12) ('##he', 4, 13) ('eh', 19250, 14) ### +### ('free', 2011, 15) ('1983', 7309, 16) ('agent', 1002, 17) ('september', 523, 18) ### +### ('early', 3978, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('luke', 1, 1) ('hoc', 0, 4) ('##var', 3, 9) ('##he', 4, 13) ('playing', 2, 45) ('played', 5, 29) ### +### ('who', 6, 21) ('encompasses', 16, 22) ('league', 15, 31) ('player', 8, 89) ('wingspan', 12, 64) ### +### ('is', 9, 106) ('play', 11, 75) ('anthony', 339, 2) ('baseball', 267, 7) ('−', 44, 41) ### +### ('hugh', 42, 49) ('##ο', 66, 50) ('tennessee', 1130, 3) ('players', 21, 153) ### +############################################################################################################ +[2023-10-07 20:10:54,452][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:10:54,452][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:10:54,876][root][INFO] - Epoch: 3: Step: 401/1557, loss[v]=0.072426, lr=0.000018, acc@1[1]=239.5/256=0.935546875, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 20:12:10,837][root][INFO] - Train batch 500 +[2023-10-07 20:12:10,838][root][INFO] - Avg. loss per last 100 batches: 0.102115 +[2023-10-07 20:12:11,543][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29284.6/29522=99.20% | mean: 0.01 | max: 4.82 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 5.81 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] types of pliers and their uses [SEP] ### +### [P_TEXT]: [CLS] these are the pliers to turn to when you need to get into tight spaces. chain - ### +### nose pliers : chain - nose pliers... have a smooth flat surface on the interior of the jaws. the ### +### small tapered point allows you to get into small areas. you will typically use this tool for ### +### gripping jewelry findings and working with wire. [SEP] ### +### ======================================= h_v_q | Gates: 28207 ======================================= ### +### ('pl', 0, 8) ('##iers', 1, 0) ('types', 2, 717) ('uses', 3, 173) ('their', 4, 20567) ### +### ('use', 5, 332) ('##ier', 6, 11) ('used', 7, 745) ('type', 8, 12733) ('usage', 9, 1327) ### +### ('applications', 10, 9976) ('familiarity', 11, 24918) ('styles', 12, 4511) ('employs', 13, 61) ### +### ('classes', 14, 2624) ('users', 15, 11104) ('examples', 16, 389) ('them', 17, 3567) ### +### ('categories', 18, 3286) ('species', 19, 6106) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##iers', 1, 0) ('chain', 4255, 1) ('tight', 2017, 2) ('jewelry', 17065, 3) ('gripping', 9611, 4) ### +### ('tape', 20030, 5) ('space', 2322, 6) ('nose', 9005, 7) ('pl', 0, 8) ('jaws', 17243, 9) ### +### ('spaces', 3005, 10) ('##ier', 6, 11) ('gripped', 1990, 12) ('findings', 10873, 13) ### +### ('surface', 8271, 14) ('tool', 1708, 15) ('turn', 4679, 16) ('tools', 418, 17) ('flat', 2898, 18) ### +### ('wire', 21182, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##iers', 1, 0) ('pl', 0, 8) ('##ier', 6, 11) ('uses', 3, 173) ('types', 2, 717) ('use', 5, 332) ### +### ('employs', 13, 61) ('wingspan', 40, 22) ('used', 7, 745) ('afraid', 39, 23) ('parker', 27, 150) ### +### ('fernando', 122, 32) ('examples', 16, 389) ('crashing', 90, 50) ('−', 107, 38) ('crashed', 66, 79) ### +### ('usage', 9, 1327) ('tools', 418, 17) ('julia', 115, 53) ('hesitated', 183, 36) ### +############################################################################################################ +[2023-10-07 20:12:11,544][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:12:11,544][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:12:11,967][root][INFO] - Epoch: 3: Step: 501/1557, loss[v]=0.100799, lr=0.000018, acc@1[1]=237.0/256=0.92578125, acc@1[2]=246.5/256=0.962890625 +[2023-10-07 20:13:29,178][root][INFO] - Train batch 600 +[2023-10-07 20:13:29,179][root][INFO] - Avg. loss per last 100 batches: 0.101321 +[2023-10-07 20:13:29,900][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29332.9/29522=99.36% | mean: 0.01 | max: 5.37 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 6.22 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] who wrote boston legal theme song [SEP] ### +### [P_TEXT]: [CLS] david e. kelley is a man who loves his music. he's married up his writing with the ### +### perfect song in many of his series. he put the concept of having a personal theme song in the pop ### +### culture vernacular when ally mcbeal decided tell him by the exciters was her personal musical ### +### motivator. boston legal is scored by danny lux with the vocal stylings of billy valentine. [SEP] ### +### ======================================= h_v_q | Gates: 27564 ======================================= ### +### ('boston', 0, 7) ('song', 1, 20) ('theme', 2, 13) ('legal', 3, 5) ('wrote', 4, 126) ### +### ('written', 5, 129) ('writing', 6, 22) ('massachusetts', 7, 122) ('law', 8, 63) ### +### ('familiarity', 9, 22440) ('who', 10, 53) ('songs', 11, 26) ('themes', 12, 34) ('.', 13, 3416) ### +### ('published', 14, 8386) ('authored', 15, 2234) ('music', 16, 6) ('write', 17, 481) ### +### ('george', 18, 1298) ('film', 19, 242) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('lux', 17889, 0) ('kelley', 12982, 1) ('perfect', 1546, 2) ('david', 111, 3) ### +### ('valentine', 8829, 4) ('legal', 3, 5) ('music', 16, 6) ('boston', 0, 7) ('mc', 5443, 8) ### +### ('pop', 3545, 9) ('##cite', 26095, 10) ('personal', 305, 11) ('danny', 5163, 12) ('theme', 2, 13) ### +### ('billy', 366, 14) ('vocal', 2744, 15) ('man', 857, 16) ('vernacular', 4195, 17) ### +### ('married', 2202, 18) ('concept', 911, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('boston', 0, 7) ('legal', 3, 5) ('theme', 2, 13) ('song', 1, 20) ('wrote', 4, 126) ### +### ('writing', 6, 22) ('written', 5, 129) ('songs', 11, 26) ('music', 16, 6) ('themes', 12, 34) ### +### ('who', 10, 53) ('law', 8, 63) ('massachusetts', 7, 122) ('david', 111, 3) ('lawyer', 38, 30) ### +### ('lawyers', 55, 35) ('musical', 83, 33) ('personal', 305, 11) ('melody', 36, 78) ('billy', 366, 14) ### +############################################################################################################ +[2023-10-07 20:13:29,900][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:13:29,900][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:13:30,324][root][INFO] - Epoch: 3: Step: 601/1557, loss[v]=0.095291, lr=0.000017, acc@1[1]=241.0/256=0.94140625, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 20:14:47,022][root][INFO] - Train batch 700 +[2023-10-07 20:14:47,023][root][INFO] - Avg. loss per last 100 batches: 0.106207 +[2023-10-07 20:14:47,739][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29317.9/29522=99.31% | mean: 0.01 | max: 5.08 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 5.76 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what war was going on during the movie atonement [SEP] ### +### [P_TEXT]: [CLS] as depicted in atonement. in atonement, directed by joe wright and adapted by ### +### christopher hampton from ian mcewan's novel, we see visualized one of the low points in the ### +### allies'war against hitler a the british evacuation of dunkirk between may 26 and june 4, 1940. ### +### robbie turner ( james mcavoy ) arrives in bray - dunes outside of dunkirk, france. at the time, 50 ### +### years had passed since the end of that war. ) the birth of a nation, adapted from thomas dixon's ### +### 1905 novel, the clansman, an homage to the ku klux klan, was the most profitable film ever made - ### +### that is up until 1937 when it was surpassed by snow white and the seven dwarfs. [SEP] ### +### ======================================= h_v_q | Gates: 28147 ======================================= ### +### ('##one', 0, 0) ('war', 1, 28) ('##ment', 2, 9) ('film', 3, 15) ('movie', 4, 32) ### +### ('during', 5, 8569) ('at', 6, 40) ('##ments', 7, 23) ('was', 8, 140) ('battle', 9, 594) ### +### ('familiarity', 10, 27878) ('.', 11, 2301) ('going', 12, 6783) ('movies', 13, 133) ### +### ('onto', 14, 278) ('warfare', 15, 688) ('fought', 16, 571) ('combat', 17, 1489) ('on', 18, 11128) ### +### ('hope', 19, 5507) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##one', 0, 0) ('dunkirk', 20845, 1) ('evacuation', 11228, 2) ('turner', 1407, 3) ### +### ('bray', 17103, 4) ('mca', 20448, 5) ('hitler', 15728, 6) ('robbie', 8539, 7) ('dunes', 9449, 8) ### +### ('##ment', 2, 9) ('allies', 3254, 10) ('hampton', 8522, 11) ('joe', 682, 12) ('homage', 12367, 13) ### +### ('birth', 6249, 14) ('film', 3, 15) ('klan', 23291, 16) ('directed', 379, 17) ('ku', 8204, 18) ### +### ('low', 3311, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##one', 0, 0) ('##ment', 2, 9) ('war', 1, 28) ('film', 3, 15) ('movie', 4, 32) ('at', 6, 40) ### +### ('##ments', 7, 23) ('was', 8, 140) ('movies', 13, 133) ('battle', 9, 594) ('films', 29, 92) ### +### ('june', 243, 20) ('onto', 14, 278) ('knew', 24, 124) ('during', 5, 8569) ('directed', 379, 17) ### +### ('afraid', 53, 69) ('joe', 682, 12) ('filming', 42, 108) ('wingspan', 89, 57) ### +############################################################################################################ +[2023-10-07 20:14:47,740][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:14:47,740][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:14:48,164][root][INFO] - Epoch: 3: Step: 701/1557, loss[v]=0.068961, lr=0.000017, acc@1[1]=241.0/256=0.94140625, acc@1[2]=252.0/256=0.984375 +[2023-10-07 20:16:04,160][root][INFO] - Train batch 800 +[2023-10-07 20:16:04,161][root][INFO] - Avg. loss per last 100 batches: 0.107507 +[2023-10-07 20:16:04,880][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29351.4/29522=99.42% | mean: 0.01 | max: 5.10 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 6.18 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] sata to usb adapter [SEP] ### +### [P_TEXT]: [CLS] the sata / ide - usb adapter gives you the flexibility of connecting almost any ### +### type of drive to your computer via usb. sata / ide to usb 2. 0 adapter the sata / ide - usb adapter ### +### gives you the flexibility of connecting almost any type of drive to your computer via usb. [SEP] ### +### ======================================= h_v_q | Gates: 28488 ======================================= ### +### ('usb', 0, 0) ('sat', 1, 2) ('##a', 2, 11) ('##er', 3, 18) ('adapt', 4, 3) ('sit', 5, 20) ### +### ('familiarity', 6, 26896) ('sitting', 7, 15) ('to', 8, 3286) ('adaptation', 9, 344) ### +### ('ammunition', 10, 247) ('sits', 11, 36) ('##ma', 12, 3708) ('.', 13, 6668) ('##ers', 14, 124) ### +### ('##а', 15, 78) ('simon', 16, 128) ('daniel', 17, 3929) ('onto', 18, 299) ('##α', 19, 24) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('usb', 0, 0) ('flexibility', 2991, 1) ('sat', 1, 2) ('adapt', 4, 3) ('almost', 6583, 4) ### +### ('drive', 2713, 5) ('via', 8053, 6) ('id', 12244, 7) ('computer', 368, 8) ('nearly', 3107, 9) ### +### ('##e', 55, 10) ('##a', 2, 11) ('types', 5897, 12) ('wingspan', 83, 13) ('ˈ', 573, 14) ### +### ('sitting', 7, 15) ('ছ', 123, 16) ('presenter', 269, 17) ('##er', 3, 18) ('practically', 7827, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('usb', 0, 0) ('sat', 1, 2) ('##a', 2, 11) ('adapt', 4, 3) ('##er', 3, 18) ('sit', 5, 20) ### +### ('sitting', 7, 15) ('sits', 11, 36) ('##e', 55, 10) ('##α', 19, 24) ('wingspan', 83, 13) ### +### ('afraid', 61, 21) ('##а', 15, 78) ('ছ', 123, 16) ('##ers', 14, 124) ('computer', 368, 8) ### +### ('simon', 16, 128) ('ammunition', 10, 247) ('adaptation', 9, 344) ('unwilling', 173, 25) ### +############################################################################################################ +[2023-10-07 20:16:04,880][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:16:04,880][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:16:05,306][root][INFO] - Epoch: 3: Step: 801/1557, loss[v]=0.082309, lr=0.000017, acc@1[1]=245.5/256=0.958984375, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 20:17:21,643][root][INFO] - Train batch 900 +[2023-10-07 20:17:21,643][root][INFO] - Avg. loss per last 100 batches: 0.108458 +[2023-10-07 20:17:22,339][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29274.6/29522=99.16% | mean: 0.01 | max: 4.72 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 5.74 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] where is bhutan located [SEP] ### +### [P_TEXT]: [CLS] bhutan's capital city, thimpu, is centrally located towards the country's western ### +### border with india. bhutan shares a 605 - kilometer ( 376 - mile ) border with india and a 470 - ### +### kilometer ( 292 - mile ) border with china. in 2000 the population of bhutan was estimated at 2, ### +### 005, 222 by the cia world factbook. hutan's capital city, thimpu, is centrally located towards the ### +### country's western border with india. [SEP] ### +### ======================================= h_v_q | Gates: 27453 ======================================= ### +### ('bhutan', 0, 0) ('located', 1, 21) ('nepal', 2, 20) ('is', 3, 219) ('india', 4, 7) ### +### ('situated', 5, 43) ('familiarity', 6, 25043) ('.', 7, 2449) ('where', 8, 52) ('coast', 9, 1593) ### +### ('tibet', 10, 38) ('location', 11, 183) ('massachusetts', 12, 10534) ('founded', 13, 1166) ### +### ('connecticut', 14, 1219) ('nepali', 15, 117) ('based', 16, 314) ('established', 17, 770) ### +### ('england', 18, 817) ('center', 19, 1922) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('bhutan', 0, 0) ('hut', 6699, 1) ('##pu', 13363, 2) ('capital', 145, 3) ('border', 2658, 4) ### +### ('centrally', 8325, 5) ('th', 7073, 6) ('india', 4, 7) ('population', 875, 8) ('towards', 1067, 9) ### +### ('cia', 2810, 10) ('shared', 7193, 11) ('##im', 12314, 12) ('estimated', 2761, 13) ('lama', 25, 14) ### +### ('cities', 5564, 15) ('china', 78, 16) ('shares', 12332, 17) ('city', 3254, 18) ### +### ('country', 652, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('bhutan', 0, 0) ('located', 1, 21) ('nepal', 2, 20) ('india', 4, 7) ('situated', 5, 43) ### +### ('where', 8, 52) ('tibet', 10, 38) ('is', 3, 219) ('lama', 25, 14) ('capital', 145, 3) ### +### ('tibetan', 35, 41) ('china', 78, 16) ('western', 64, 26) ('central', 27, 78) ('nepali', 15, 117) ### +### ('location', 11, 183) ('west', 39, 89) ('−', 133, 34) ('encompasses', 31, 121) ('afghan', 30, 128) ### +############################################################################################################ +[2023-10-07 20:17:22,340][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:17:22,340][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:17:22,743][root][INFO] - Epoch: 3: Step: 901/1557, loss[v]=0.091676, lr=0.000017, acc@1[1]=240.0/256=0.9375, acc@1[2]=246.0/256=0.9609375 +[2023-10-07 20:18:39,979][root][INFO] - Train batch 1000 +[2023-10-07 20:18:39,980][root][INFO] - Avg. loss per last 100 batches: 0.106214 +[2023-10-07 20:18:40,668][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29392.4/29522=99.56% | mean: 0.01 | max: 5.01 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 6.07 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] definition of interior angles [SEP] ### +### [P_TEXT]: [CLS] an interior angle is an angle inside a shape. since triangles have three angles, ### +### they have three interior angles. in this triangle, angles a, b and c are all interior angles. just ### +### as the pieces in a jigsaw puzzle fit together perfectly, the interior angles in a triangle must fit ### +### with each other. the sum of the interior angles is always 180 degrees. in other words, a + b + c = ### +### 180 degrees. [SEP] ### +### ======================================= h_v_q | Gates: 27629 ======================================= ### +### ('interior', 0, 2) ('angles', 1, 0) ('definition', 2, 9) ('noun', 3, 19821) ('interiors', 4, 41) ### +### ('angle', 5, 1) ('internal', 6, 28) ('defined', 7, 240) ('inside', 8, 15) ('of', 9, 24015) ### +### ('exterior', 10, 100) ('familiarity', 11, 26503) ('or', 12, 12573) ('.', 13, 4563) ### +### ('lines', 14, 3467) ('foreign', 15, 5792) (';', 16, 3730) ('forces', 17, 580) ('roger', 18, 801) ### +### ('##°', 19, 56) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('angles', 1, 0) ('angle', 5, 1) ('interior', 0, 2) ('triangle', 14761, 3) ('triangles', 4865, 4) ### +### ('puzzle', 5679, 5) ('fit', 4926, 6) ('ji', 7315, 7) ('degrees', 10286, 8) ('definition', 2, 9) ### +### ('encompasses', 208, 10) ('define', 2543, 11) ('shape', 2992, 12) ('##gs', 12954, 13) ### +### ('puzzles', 11897, 14) ('inside', 8, 15) ('three', 2018, 16) ('−', 350, 17) ('+', 5274, 18) ### +### ('b', 3561, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('interior', 0, 2) ('angles', 1, 0) ('definition', 2, 9) ('angle', 5, 1) ('interiors', 4, 41) ### +### ('internal', 6, 28) ('inside', 8, 15) ('exterior', 10, 100) ('defined', 7, 240) ('odds', 25, 33) ### +### ('##°', 19, 56) ('definitions', 36, 31) ('encompasses', 208, 10) ('meaning', 99, 38) ### +### ('stab', 47, 67) ('outside', 35, 83) ('−', 350, 17) ('angled', 148, 40) ('wingspan', 316, 25) ### +### ('stark', 114, 66) ### +############################################################################################################ +[2023-10-07 20:18:40,668][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:18:40,668][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:18:41,089][root][INFO] - Epoch: 3: Step: 1001/1557, loss[v]=0.171585, lr=0.000017, acc@1[1]=234.5/256=0.916015625, acc@1[2]=243.5/256=0.951171875 +[2023-10-07 20:19:57,800][root][INFO] - Train batch 1100 +[2023-10-07 20:19:57,801][root][INFO] - Avg. loss per last 100 batches: 0.107280 +[2023-10-07 20:19:58,511][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29352.2/29522=99.42% | mean: 0.01 | max: 5.22 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 5.74 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] middle aged actors [SEP] ### +### [P_TEXT]: [CLS] the cast members of at middle age have been in many other movies, so use this list ### +### as a starting point to find actors or actresses that you may not be familiar with. the list you're ### +### viewing has a variety of actors, like pan hong and shichang da, in it. [SEP] ### +### ======================================= h_v_q | Gates: 27782 ======================================= ### +### ('middle', 0, 8) ('actors', 1, 3) ('aged', 2, 180) ('age', 3, 0) ('actor', 4, 11) ('mid', 5, 145) ### +### ('familiarity', 6, 2646) (';', 7, 6050) ('players', 8, 156) ('upper', 9, 388) ('.', 10, 6127) ### +### ('actress', 11, 18) ('film', 12, 37) ('characters', 13, 191) ('old', 14, 167) ('comedy', 15, 1780) ### +### ('ages', 16, 81) ('center', 17, 1379) ('young', 18, 168) ('elementary', 19, 1595) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('age', 3, 0) ('cast', 97, 1) ('hong', 4730, 2) ('actors', 1, 3) ('actresses', 180, 4) ### +### ('viewing', 5649, 5) ('pan', 68, 6) ('list', 6336, 7) ('middle', 0, 8) ('movies', 1007, 9) ### +### ('da', 2372, 10) ('actor', 4, 11) ('##chang', 18874, 12) ('familiar', 2312, 13) ('shi', 5668, 14) ### +### ('movie', 863, 15) ('members', 201, 16) ('variety', 5331, 17) ('actress', 11, 18) ('##ο', 610, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('middle', 0, 8) ('actors', 1, 3) ('age', 3, 0) ('actor', 4, 11) ('aged', 2, 180) ### +### ('actress', 11, 18) ('mid', 5, 145) ('film', 12, 37) ('cast', 97, 1) ('pan', 68, 6) ### +### ('older', 31, 33) ('ages', 16, 81) ('players', 8, 156) ('actresses', 180, 4) ('old', 14, 167) ### +### ('characters', 13, 191) ('ছ', 81, 29) ('filming', 30, 101) ('members', 201, 16) ('upper', 9, 388) ### +############################################################################################################ +[2023-10-07 20:19:58,511][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:19:58,511][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:19:58,918][root][INFO] - Epoch: 3: Step: 1101/1557, loss[v]=0.112765, lr=0.000017, acc@1[1]=240.0/256=0.9375, acc@1[2]=248.0/256=0.96875 +[2023-10-07 20:21:15,577][root][INFO] - Train batch 1200 +[2023-10-07 20:21:15,578][root][INFO] - Avg. loss per last 100 batches: 0.106334 +[2023-10-07 20:21:16,274][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29290.2/29522=99.21% | mean: 0.01 | max: 4.96 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 5.74 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how to boil ears of corn [SEP] ### +### [P_TEXT]: [CLS] just before cooking, husk the corn, pull off the silky threads, and cut out any ### +### blemishes with a pointed knife. drop the corn into a large pot filled with boiling salted water. ### +### cover the pot and let the water return to a boil again, then turn off the heat and keep the pot ### +### covered. after about 5 minutes, remove enough ears for a first serving. you can keep the remaining ### +### corn warm in the water for another 10 minutes without its becoming tough. serve with lots of butter ### +### and salt. [SEP] ### +### ======================================= h_v_q | Gates: 27958 ======================================= ### +### ('corn', 0, 0) ('ears', 1, 1) ('boil', 2, 18) ('to', 3, 20647) ('ear', 4, 137) ('eager', 5, 353) ### +### ('manual', 6, 230) ('jeremy', 7, 143) ('maize', 8, 30) ('familiarity', 9, 25314) ('boiling', 10, 9) ### +### ('ability', 11, 2748) ('of', 12, 27426) ('onto', 13, 38) ('software', 14, 4565) ('how', 15, 22173) ### +### ('tap', 16, 8974) ('.', 17, 7521) ('boiled', 18, 1335) ('tightly', 19, 110) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('corn', 0, 0) ('ears', 1, 1) ('hu', 1532, 2) ('warm', 578, 3) ('pot', 33, 4) ('minutes', 280, 5) ### +### ('cooking', 12778, 6) ('salt', 450, 7) ('serving', 5176, 8) ('boiling', 10, 9) ('cover', 4771, 10) ### +### ('serve', 2108, 11) ('tough', 7019, 12) ('remaining', 7281, 13) ('water', 1050, 14) ('ˈ', 623, 15) ### +### ('before', 5825, 16) ('long', 3014, 17) ('boil', 2, 18) ('threads', 4384, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('corn', 0, 0) ('ears', 1, 1) ('boil', 2, 18) ('maize', 8, 30) ('boiling', 10, 9) ('ear', 4, 137) ### +### ('onto', 13, 38) ('jeremy', 7, 143) ('pot', 33, 4) ('manual', 6, 230) ('eager', 5, 353) ### +### ('wheat', 29, 54) ('wingspan', 44, 29) ('tightly', 19, 110) ('##α', 48, 32) ('afraid', 45, 43) ### +### ('ছ', 49, 48) ('##₂', 54, 46) ('able', 25, 120) ('presenter', 72, 27) ### +############################################################################################################ +[2023-10-07 20:21:16,275][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:21:16,275][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:21:16,697][root][INFO] - Epoch: 3: Step: 1201/1557, loss[v]=0.156451, lr=0.000017, acc@1[1]=236.5/256=0.923828125, acc@1[2]=243.5/256=0.951171875 +[2023-10-07 20:22:32,653][root][INFO] - Train batch 1300 +[2023-10-07 20:22:32,653][root][INFO] - Avg. loss per last 100 batches: 0.104722 +[2023-10-07 20:22:33,367][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29244.1/29522=99.06% | mean: 0.01 | max: 4.54 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.6/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 5.80 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what kind of medication is ambril [SEP] ### +### [P_TEXT]: [CLS] some medications need not be prescribed by healthcare practitioners and can be ### +### purchased and used without prescription by the patients ; these are called over - the - counter ### +### medications. read the drug prescription information of ambril before taking it. ambril uses ambril ### +### is a mucolytic agent, prescribed for various respiratory diseases such as emphysema with bronchitis ### +### pneumoconiosis, chronic inflammatory pulmonary conditions, t [SEP] ### +### ======================================= h_v_q | Gates: 27841 ======================================= ### +### ('am', 0, 2) ('##bri', 1, 5) ('##l', 2, 27) ('medication', 3, 3) ('is', 4, 1257) ('kind', 5, 8406) ### +### ('medications', 6, 1) ('type', 7, 11913) ('class', 8, 8294) ('types', 9, 765) ### +### ('familiarity', 10, 27631) ('encompasses', 11, 35) ('definition', 12, 689) ('medicine', 13, 65) ### +### ('ram', 14, 855) ('sort', 15, 1144) ('software', 16, 3330) ('afraid', 17, 59) ('alfred', 18, 355) ### +### ('im', 19, 990) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('prescription', 321, 0) ('medications', 6, 1) ('am', 0, 2) ('medication', 3, 3) ### +### ('healthcare', 2053, 4) ('##bri', 1, 5) ('mu', 2473, 6) ('##ytic', 22170, 7) ### +### ('prescribed', 10861, 8) ('respiratory', 16052, 9) ('ˈ', 620, 10) ('counter', 2429, 11) ### +### ('inflammatory', 16207, 12) ('drug', 175, 13) ('−', 37, 14) ('pulmonary', 3136, 15) ### +### ('diseases', 4104, 16) ('##sis', 20155, 17) ('drugs', 53, 18) ('over', 5108, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('am', 0, 2) ('##bri', 1, 5) ('medication', 3, 3) ('##l', 2, 27) ('medications', 6, 1) ### +### ('encompasses', 11, 35) ('medicine', 13, 65) ('afraid', 17, 59) ('−', 37, 14) ('##ل', 26, 33) ### +### ('drugs', 53, 18) ('agent', 43, 23) ('is', 4, 1257) ('##α', 31, 43) ('prescription', 321, 0) ### +### ('hated', 41, 52) ('wingspan', 46, 53) ('##ο', 108, 29) ('types', 9, 765) ('ছ', 61, 45) ### +############################################################################################################ +[2023-10-07 20:22:33,368][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:22:33,368][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:22:33,774][root][INFO] - Epoch: 3: Step: 1301/1557, loss[v]=0.100971, lr=0.000017, acc@1[1]=240.0/256=0.9375, acc@1[2]=246.5/256=0.962890625 +[2023-10-07 20:23:50,863][root][INFO] - Train batch 1400 +[2023-10-07 20:23:50,864][root][INFO] - Avg. loss per last 100 batches: 0.100103 +[2023-10-07 20:23:51,562][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29337.1/29522=99.37% | mean: 0.01 | max: 4.89 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 5.98 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a legal retainer agreement [SEP] ### +### [P_TEXT]: [CLS] a retainer agreement is a work for hire contract. it falls between a one - time ### +### contract and full - time employment. its distinguishing feature is that the employer pays in ### +### advance for work to be specified later. additional contracts regarding the performance of this work ### +### may also apply. [SEP] ### +### ======================================= h_v_q | Gates: 28012 ======================================= ### +### ('agreement', 0, 2) ('retain', 1, 1) ('##er', 2, 9) ('legal', 3, 3668) ('definition', 4, 17) ### +### ('retained', 5, 6) ('is', 6, 206) ('familiarity', 7, 24868) ('encompasses', 8, 8) ### +### ('contract', 9, 3) ('agreed', 10, 14) ('##ers', 11, 76) ('##r', 12, 136) ('retaining', 13, 16) ### +### ('law', 14, 5681) ('maintain', 15, 154) ('a', 16, 511) ('retention', 17, 23) ### +### ('association', 18, 2510) ('noun', 19, 20247) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('hire', 2649, 0) ('retain', 1, 1) ('agreement', 0, 2) ('contract', 9, 3) ('work', 279, 4) ### +### ('employer', 15667, 5) ('retained', 5, 6) ('advance', 719, 7) ('encompasses', 8, 8) ('##er', 2, 9) ### +### ('distinguishing', 19401, 10) ('full', 1621, 11) ('features', 807, 12) ('feature', 5494, 13) ### +### ('agreed', 10, 14) ('retains', 37, 15) ('retaining', 13, 16) ('definition', 4, 17) ### +### ('specified', 655, 18) ('define', 9706, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('retain', 1, 1) ('agreement', 0, 2) ('##er', 2, 9) ('retained', 5, 6) ('definition', 4, 17) ### +### ('contract', 9, 3) ('encompasses', 8, 8) ('agreed', 10, 14) ('retaining', 13, 16) ### +### ('retention', 17, 23) ('legal', 3, 3668) ('agreements', 21, 30) ('##ers', 11, 76) ('is', 6, 206) ### +### ('retains', 37, 15) ('treaty', 23, 41) ('agree', 38, 27) ('deal', 27, 51) ('##r', 12, 136) ### +### ('plan', 22, 80) ### +############################################################################################################ +[2023-10-07 20:23:51,562][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:23:51,562][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:23:51,987][root][INFO] - Epoch: 3: Step: 1401/1557, loss[v]=0.143409, lr=0.000017, acc@1[1]=239.5/256=0.935546875, acc@1[2]=243.0/256=0.94921875 +[2023-10-07 20:25:08,002][root][INFO] - Train batch 1500 +[2023-10-07 20:25:08,002][root][INFO] - Avg. loss per last 100 batches: 0.098905 +[2023-10-07 20:25:08,717][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29281.3/29522=99.18% | mean: 0.01 | max: 5.33 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 6.07 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is moist heat [SEP] ### +### [P_TEXT]: [CLS] quick answer. moist heat sterilization is a technique that uses heated water and ### +### superheated steam to disinfect contaminated materials, such as culture media, surgical instruments, ### +### glassware and other non - thermally sensitive items. [SEP] ### +### ======================================= h_v_q | Gates: 27000 ======================================= ### +### ('heat', 0, 0) ('moist', 1, 3) ('is', 2, 429) ('heats', 3, 49) ('warm', 4, 33) ('heated', 5, 7) ### +### ('warmth', 6, 31) ('hot', 7, 55) ('definition', 8, 25) ('wet', 9, 47) ('.', 10, 6503) ### +### ('heating', 11, 120) ('cold', 12, 116) ('familiarity', 13, 28636) ('dry', 14, 46) ('damp', 15, 48) ### +### ('sun', 16, 2153) ('something', 17, 851) ('encompasses', 18, 14) ('temperature', 19, 61) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('heat', 0, 0) ('steam', 52, 1) ('##ware', 13783, 2) ('moist', 1, 3) ('contaminated', 4035, 4) ### +### ('##ization', 6827, 5) ('instruments', 4542, 6) ('heated', 5, 7) ('techniques', 3110, 8) ### +### ('##ril', 22988, 9) ('sensitive', 6204, 10) ('ste', 927, 11) ('technique', 5880, 12) ### +### ('thermal', 28, 13) ('encompasses', 18, 14) ('instrument', 5605, 15) ('answer', 2480, 16) ### +### ('surgical', 10664, 17) ('materials', 2595, 18) ('water', 60, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('heat', 0, 0) ('moist', 1, 3) ('heated', 5, 7) ('warmth', 6, 31) ('warm', 4, 33) ('heats', 3, 49) ### +### ('definition', 8, 25) ('encompasses', 18, 14) ('hot', 7, 55) ('wet', 9, 47) ('is', 2, 429) ### +### ('dry', 14, 46) ('damp', 15, 48) ('heating', 11, 120) ('cold', 12, 116) ('temperature', 19, 61) ### +### ('steam', 52, 1) ('thermal', 28, 13) ('moisture', 24, 51) ('water', 60, 19) ### +############################################################################################################ +[2023-10-07 20:25:08,718][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:25:08,718][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:25:09,126][root][INFO] - Epoch: 3: Step: 1501/1557, loss[v]=0.095878, lr=0.000017, acc@1[1]=241.5/256=0.943359375, acc@1[2]=247.0/256=0.96484375 +[2023-10-07 20:25:52,690][root][INFO] - rank=1; last iteration 1557 +[2023-10-07 20:25:52,691][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 20:25:52,691][root][INFO] - rank=3; last iteration 1557 +[2023-10-07 20:25:52,691][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-07 20:25:52,691][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 20:25:52,691][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-07 20:25:52,692][root][INFO] - rank=2; last iteration 1557 +[2023-10-07 20:25:52,692][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 20:25:52,692][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-07 20:25:52,693][root][INFO] - rank=0; last iteration 1557 +[2023-10-07 20:25:52,693][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 20:25:52,693][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-07 20:25:52,699][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 20:25:52,699][root][INFO] - Epoch finished on 1 +[2023-10-07 20:25:52,700][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 20:25:52,700][root][INFO] - Epoch finished on 3 +[2023-10-07 20:25:52,700][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 20:25:52,701][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 20:25:52,701][root][INFO] - Epoch finished on 2 +[2023-10-07 20:25:52,701][root][INFO] - Epoch finished on 0 +[2023-10-07 20:26:03,872][root][INFO] - Saved checkpoint at ./vdr_3 +[2023-10-07 20:26:03,873][root][INFO] - Saved checkpoint at ./vdr_3 +[2023-10-07 20:26:03,873][root][INFO] - Av Loss per epoch=0.104726 +[2023-10-07 20:26:03,874][root][INFO] - Av Loss per epoch=0.104726 +[2023-10-07 20:26:03,874][root][INFO] - epoch total (1) correct predictions=373080 +[2023-10-07 20:26:03,873][root][INFO] - Saved checkpoint at ./vdr_3 +[2023-10-07 20:26:03,874][root][INFO] - epoch total (1) correct predictions=373080 +[2023-10-07 20:26:03,874][root][INFO] - epoch total (2) correct predictions=386170 +[2023-10-07 20:26:03,874][root][INFO] - Av Loss per epoch=0.104726 +[2023-10-07 20:26:03,874][root][INFO] - epoch total (2) correct predictions=386170 +[2023-10-07 20:26:03,874][root][INFO] - epoch total (1) correct predictions=373080 +[2023-10-07 20:26:03,874][root][INFO] - epoch total (2) correct predictions=386170 +[2023-10-07 20:26:03,874][root][INFO] - Saved checkpoint at ./vdr_3 +[2023-10-07 20:26:03,875][root][INFO] - Av Loss per epoch=0.104726 +[2023-10-07 20:26:03,875][root][INFO] - epoch total (1) correct predictions=373080 +[2023-10-07 20:26:03,875][root][INFO] - epoch total (2) correct predictions=386170 +[2023-10-07 20:26:03,877][root][INFO] - ***** Epoch 4 ***** +[2023-10-07 20:26:03,877][root][INFO] - ***** Epoch 4 ***** +[2023-10-07 20:26:03,877][root][INFO] - ***** Epoch 4 ***** +[2023-10-07 20:26:03,878][root][INFO] - ***** Epoch 4 ***** +[2023-10-07 20:26:03,881][root][INFO] - rank=3; Iteration start +[2023-10-07 20:26:03,881][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 20:26:03,881][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 20:26:03,882][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-07 20:26:03,884][root][INFO] - rank=1; Iteration start +[2023-10-07 20:26:03,884][root][INFO] - rank=2; Iteration start +[2023-10-07 20:26:03,884][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 20:26:03,884][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 20:26:03,884][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 20:26:03,884][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 20:26:03,884][root][INFO] - rank=0; Iteration start +[2023-10-07 20:26:03,885][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 20:26:03,885][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 20:26:03,886][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-07 20:26:03,886][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-07 20:26:03,886][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-07 20:26:14,315][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29270.4/29522=99.15% | mean: 0.01 | max: 5.22 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 5.80 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a exhaust valves [SEP] ### +### [P_TEXT]: [CLS] exhaust valve - a valve through which burned gases from a cylinder escape into the ### +### exhaust manifold. exhaust system, exhaust - system consisting of the parts of an engine through ### +### which burned gases or steam are discharged. valve - control consisting of a mechanical device for ### +### controlling the flow of a fluid. [SEP] ### +### ======================================= h_v_q | Gates: 27508 ======================================= ### +### ('exhaust', 0, 1) ('valves', 1, 4) ('valve', 2, 0) ('definition', 3, 8) ('is', 4, 386) ### +### ('.', 5, 4108) ('a', 6, 15689) ('forces', 7, 11488) ('familiarity', 8, 28854) ('something', 9, 743) ### +### ('encompasses', 10, 20) ('engine', 11, 7) ('defined', 12, 171) ('noun', 13, 13596) ### +### ('systems', 14, 35) ('...', 15, 1165) ('gates', 16, 5846) ('ocean', 17, 7615) ('fuel', 18, 139) ### +### ('nelson', 19, 274) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('valve', 2, 0) ('exhaust', 0, 1) ('steam', 62, 2) ('manifold', 10141, 3) ('valves', 1, 4) ### +### ('gases', 1129, 5) ('cylinder', 694, 6) ('engine', 11, 7) ('definition', 3, 8) ('burned', 4487, 9) ### +### ('gas', 391, 10) ('define', 12464, 11) ('escape', 35, 12) ('fluid', 2374, 13) ('flow', 314, 14) ### +### ('discharged', 6276, 15) ('meaning', 984, 16) ('parts', 1176, 17) ('engines', 354, 18) ### +### ('−', 77, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('exhaust', 0, 1) ('valves', 1, 4) ('valve', 2, 0) ('definition', 3, 8) ('engine', 11, 7) ### +### ('encompasses', 10, 20) ('is', 4, 386) ('steam', 62, 2) ('escape', 35, 12) ('systems', 14, 35) ### +### ('exit', 30, 38) ('−', 77, 19) ('smoke', 21, 82) ('what', 42, 36) ('defined', 12, 171) ### +### ('gas', 391, 10) ('flow', 314, 14) ('cylinder', 694, 6) ('fuel', 18, 139) ('mechanical', 148, 27) ### +############################################################################################################ +[2023-10-07 20:26:14,316][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:26:14,316][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:26:14,712][root][INFO] - Epoch: 4: Step: 1/1557, loss[v]=0.063145, lr=0.000017, acc@1[1]=245.0/256=0.95703125, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 20:27:31,584][root][INFO] - Train batch 100 +[2023-10-07 20:27:31,585][root][INFO] - Avg. loss per last 100 batches: 0.097321 +[2023-10-07 20:27:32,275][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29313.5/29522=99.29% | mean: 0.01 | max: 4.69 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.8/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.14 | max: 5.75 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] average surgery scheduler salary charlotte nc [SEP] ### +### [P_TEXT]: [CLS] a surgery scheduler in charlotte, north carolina earns an average wage of $ 15. 65 ### +### per hour. [SEP] ### +### ======================================= h_v_q | Gates: 28038 ======================================= ### +### ('charlotte', 0, 1) ('surgery', 1, 2) ('schedule', 2, 0) ('average', 3, 7) ('##r', 4, 27) ### +### ('carolina', 5, 5) ('$', 6, 8) ('salary', 7, 19) ('familiarity', 8, 27312) ('nc', 9, 18) ### +### ('hampshire', 10, 9472) ('georgia', 11, 1801) ('virginia', 12, 436) ('arkansas', 13, 3072) ### +### ('colorado', 14, 9853) ('massachusetts', 15, 10360) ('louisiana', 16, 2161) ('##rs', 17, 96) ### +### ('busy', 18, 187) ('pennsylvania', 19, 956) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('schedule', 2, 0) ('charlotte', 0, 1) ('surgery', 1, 2) ('earn', 863, 3) ('wage', 2173, 4) ### +### ('carolina', 5, 5) ('surgical', 22, 6) ('average', 3, 7) ('$', 6, 8) ('earning', 135, 9) ### +### ('hour', 2871, 10) ('price', 151, 11) ('speed', 460, 12) ('ˈ', 120, 13) ('wages', 947, 14) ### +### ('earned', 1629, 15) ('cost', 535, 16) ('prices', 351, 17) ('nc', 9, 18) ('salary', 7, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('charlotte', 0, 1) ('schedule', 2, 0) ('surgery', 1, 2) ('average', 3, 7) ('carolina', 5, 5) ### +### ('##r', 4, 27) ('$', 6, 8) ('salary', 7, 19) ('nc', 9, 18) ('surgical', 22, 6) ### +### ('scheduled', 20, 20) ('##rs', 17, 96) ('afraid', 44, 32) ('earning', 135, 9) ('−', 47, 30) ### +### ('surgeon', 56, 26) ('wingspan', 63, 23) ('##α', 79, 21) ('ˈ', 120, 13) ('price', 151, 11) ### +############################################################################################################ +[2023-10-07 20:27:32,275][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:27:32,276][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:27:32,698][root][INFO] - Epoch: 4: Step: 101/1557, loss[v]=0.064457, lr=0.000017, acc@1[1]=245.0/256=0.95703125, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 20:28:49,489][root][INFO] - Train batch 200 +[2023-10-07 20:28:49,490][root][INFO] - Avg. loss per last 100 batches: 0.090370 +[2023-10-07 20:28:50,189][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29186.2/29522=98.86% | mean: 0.01 | max: 5.16 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 6.01 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] safe harbor policy definition [SEP] ### +### [P_TEXT]: [CLS] legal protection from a lawsuit. regulators often apply safe harbor to some ### +### corporate actions as long as those actions are taken in good faith. 1. a regulation that protects ### +### individuals or corporations from the legal consequences of certain actions they undertake. [SEP] ### +### ======================================= h_v_q | Gates: 26525 ======================================= ### +### ('harbor', 0, 0) ('safe', 1, 2) ('policy', 2, 2014) ('definition', 3, 15) ('noun', 4, 21792) ### +### ('harbour', 5, 33) ('defined', 6, 264) ('policies', 7, 1606) ('bay', 8, 429) ('safety', 9, 76) ### +### ('port', 10, 496) ('.', 11, 9343) ('navy', 12, 738) ('sea', 13, 1099) ('center', 14, 2464) ### +### ('meaning', 15, 20) ('term', 16, 7801) ('strategy', 17, 3479) ('temple', 18, 672) ### +### ('dangerous', 19, 110) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('harbor', 0, 0) ('lawsuit', 1527, 1) ('safe', 1, 2) ('regulators', 23369, 3) ### +### ('protection', 939, 4) ('regulation', 175, 5) ('corporate', 1608, 6) ('apply', 9241, 7) ### +### ('protects', 5084, 8) ('regulator', 20776, 9) ('legal', 1443, 10) ('define', 1729, 11) ### +### ('litigation', 4848, 12) ('protect', 429, 13) ('actions', 1032, 14) ('definition', 3, 15) ### +### ('faith', 1564, 16) ('regulatory', 3527, 17) ('corporations', 12179, 18) ('action', 187, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('harbor', 0, 0) ('safe', 1, 2) ('definition', 3, 15) ('harbour', 5, 33) ('policy', 2, 2014) ### +### ('safety', 9, 76) ('meaning', 15, 20) ('defined', 6, 264) ('encompasses', 32, 22) ('bay', 8, 429) ### +### ('regulation', 175, 5) ('definitions', 20, 52) ('unsafe', 52, 25) ('port', 10, 496) ### +### ('policies', 7, 1606) ('dangerous', 19, 110) ('safer', 68, 38) ('corporation', 85, 34) ### +### ('navy', 12, 738) ('good', 82, 40) ### +############################################################################################################ +[2023-10-07 20:28:50,189][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:28:50,190][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:28:50,609][root][INFO] - Epoch: 4: Step: 201/1557, loss[v]=0.121278, lr=0.000017, acc@1[1]=238.5/256=0.931640625, acc@1[2]=248.0/256=0.96875 +[2023-10-07 20:30:07,017][root][INFO] - Train batch 300 +[2023-10-07 20:30:07,018][root][INFO] - Avg. loss per last 100 batches: 0.098187 +[2023-10-07 20:30:07,699][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29237.2/29522=99.04% | mean: 0.01 | max: 5.11 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 6.08 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] age requirements for fishing license in ny state [SEP] ### +### [P_TEXT]: [CLS] licenses to fish in ny are categorized based on factors such as : the applicant's ### +### age, physical ability and the longevity of the license. anglers younger than 16 years of age, ### +### native americans and resident landowners fishing on their property are exempt from buying a new ### +### york fishing permit. [SEP] ### +### ======================================= h_v_q | Gates: 27165 ======================================= ### +### ('fishing', 0, 1) ('age', 1, 6) ('ny', 2, 0) ('requirements', 3, 2520) ('license', 4, 7) ### +### ('york', 5, 5) ('state', 6, 21833) ('fish', 7, 2) ('pennsylvania', 8, 76) ('brooklyn', 9, 267) ### +### ('connecticut', 10, 282) ('wisconsin', 11, 6386) ('hunting', 12, 227) ('.', 13, 8225) ### +### ('jeremy', 14, 50) ('massachusetts', 15, 1925) ('manhattan', 16, 165) ('familiarity', 17, 25531) ### +### ('required', 18, 15916) ('california', 19, 2518) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('ny', 2, 0) ('fishing', 0, 1) ('fish', 7, 2) ('exempt', 5566, 3) ('licenses', 646, 4) ### +### ('york', 5, 5) ('age', 1, 6) ('license', 4, 7) ('resident', 1391, 8) ('categorized', 25733, 9) ### +### ('based', 5395, 10) ('native', 1130, 11) ('landowners', 15096, 12) ('longevity', 20802, 13) ### +### ('−', 102, 14) ('younger', 156, 15) ('angle', 9477, 16) ('permit', 3917, 17) ### +### ('applicant', 14236, 18) ('ˈ', 508, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('fishing', 0, 1) ('ny', 2, 0) ('age', 1, 6) ('license', 4, 7) ('york', 5, 5) ('fish', 7, 2) ### +### ('pennsylvania', 8, 76) ('requirements', 3, 2520) ('jeremy', 14, 50) ('salmon', 24, 32) ### +### ('older', 34, 36) ('brooklyn', 9, 267) ('hunting', 12, 227) ('connecticut', 10, 282) ### +### ('diving', 26, 77) ('−', 102, 14) ('manhattan', 16, 165) ('younger', 156, 15) ('anger', 47, 57) ### +### ('young', 56, 52) ### +############################################################################################################ +[2023-10-07 20:30:07,699][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:30:07,699][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:30:08,118][root][INFO] - Epoch: 4: Step: 301/1557, loss[v]=0.113851, lr=0.000017, acc@1[1]=243.0/256=0.94921875, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 20:31:23,783][root][INFO] - Train batch 400 +[2023-10-07 20:31:23,784][root][INFO] - Avg. loss per last 100 batches: 0.094855 +[2023-10-07 20:31:24,471][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29269.4/29522=99.14% | mean: 0.01 | max: 5.20 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 6.02 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what states is medical marijuana legal [SEP] ### +### [P_TEXT]: [CLS] in eight states, nevada, maine, colorado, washington, california, massachusetts, ### +### alaska, and oregon, the sale and possession of marijuana is legal for both medical and recreational ### +### use ; and washington dc has legalized personal use but not commercial sale. [SEP] ### +### ======================================= h_v_q | Gates: 27689 ======================================= ### +### ('marijuana', 0, 0) ('medical', 1, 11) ('states', 2, 5) ('legal', 3, 7) ('definition', 4, 2325) ### +### ('familiarity', 5, 26322) ('state', 6, 21) ('medicine', 7, 153) ('cannabis', 8, 18) ('is', 9, 1312) ### +### ('hampshire', 10, 4836) ('regions', 11, 86) ('encompasses', 12, 283) ('kingdom', 13, 63) ### +### ('hospital', 14, 176) ('.', 15, 3947) ('california', 16, 22) ('1945', 17, 2653) ### +### ('nations', 18, 232) ('alfred', 19, 173) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('marijuana', 0, 0) ('nevada', 45, 1) ('sale', 11888, 2) ('dc', 15080, 3) ('oregon', 65, 4) ### +### ('states', 2, 5) ('washington', 47, 6) ('legal', 3, 7) ('maine', 1866, 8) ('recreational', 945, 9) ### +### ('alaska', 1025, 10) ('medical', 1, 11) ('personal', 264, 12) ('commercial', 54, 13) ### +### ('possession', 9085, 14) ('ˈ', 271, 15) ('colorado', 158, 16) ('−', 111, 17) ('cannabis', 8, 18) ### +### ('massachusetts', 34, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('marijuana', 0, 0) ('states', 2, 5) ('medical', 1, 11) ('legal', 3, 7) ('state', 6, 21) ### +### ('cannabis', 8, 18) ('california', 16, 22) ('nevada', 45, 1) ('washington', 47, 6) ### +### ('oregon', 65, 4) ('massachusetts', 34, 19) ('regions', 11, 86) ('kingdom', 13, 63) ### +### ('medicine', 7, 153) ('commercial', 54, 13) ('−', 111, 17) ('colorado', 158, 16) ### +### ('personal', 264, 12) ('hospital', 14, 176) ('ˈ', 271, 15) ### +############################################################################################################ +[2023-10-07 20:31:24,472][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:31:24,472][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:31:24,877][root][INFO] - Epoch: 4: Step: 401/1557, loss[v]=0.099094, lr=0.000017, acc@1[1]=235.0/256=0.91796875, acc@1[2]=246.0/256=0.9609375 +[2023-10-07 20:32:41,297][root][INFO] - Train batch 500 +[2023-10-07 20:32:41,298][root][INFO] - Avg. loss per last 100 batches: 0.096865 +[2023-10-07 20:32:42,013][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29268.1/29522=99.14% | mean: 0.01 | max: 5.15 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.14 | max: 5.92 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] does beef contain protein [SEP] ### +### [P_TEXT]: [CLS] beef is a great source of protein. beef supplements let you avoid the fat and ### +### cholesterol from steak and hamburger, but still get the protein from red meat. beef protein is rich ### +### in vitamin a, c, b6, b12, thiamin, riboflavin, niacin, calcium, phosphorus, magnesium, sodium and ### +### potassium. onsume beef protein daily to build new muscle tissue. * you should always ingest protein ### +### from food sources, but the purity of beef protein powder gives you what you need, when you need it. ### +### *. always check calories per serving to make sure beef protein fits within your daily values. [SEP] ### +### ======================================= h_v_q | Gates: 27427 ======================================= ### +### ('beef', 0, 0) ('protein', 1, 1) ('contain', 2, 2522) ('does', 3, 15810) ('contains', 4, 1054) ### +### ('.', 5, 7970) ('familiarity', 6, 22177) ('proteins', 7, 4) ('did', 8, 8790) ('containing', 9, 846) ### +### ('doing', 10, 374) ('doesn', 11, 2781) (';', 12, 4107) ('"', 13, 13548) ('contained', 14, 17298) ### +### ('do', 15, 10408) ('or', 16, 23349) ('meat', 17, 3) ('steel', 18, 1009) ('include', 19, 479) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('beef', 0, 0) ('protein', 1, 1) ('##sume', 21155, 2) ('meat', 17, 3) ('proteins', 7, 4) ### +### ('steak', 245, 5) ('hamburger', 14792, 6) ('ˈ', 570, 7) ('supplements', 26745, 8) ('rich', 1914, 9) ### +### ('wingspan', 258, 10) ('muscle', 1566, 11) ('vitamin', 1280, 12) ('magnesium', 6636, 13) ### +### ('##ο', 191, 14) ('−', 284, 15) ('##α', 124, 16) ('ruins', 28, 17) ('sodium', 1586, 18) ### +### ('afraid', 184, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('beef', 0, 0) ('protein', 1, 1) ('proteins', 7, 4) ('meat', 17, 3) ('contain', 2, 2522) ### +### ('contains', 4, 1054) ('ruins', 28, 17) ('hating', 55, 22) ('doing', 10, 374) ('steak', 245, 5) ### +### ('stark', 33, 56) ('anger', 43, 38) ('simon', 42, 63) ('##α', 124, 16) ('##₂', 100, 25) ### +### ('hated', 102, 29) ('##ο', 191, 14) ('presenter', 98, 34) ('albert', 20, 203) ('wingspan', 258, 10) ### +############################################################################################################ +[2023-10-07 20:32:42,013][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:32:42,013][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:32:42,437][root][INFO] - Epoch: 4: Step: 501/1557, loss[v]=0.078121, lr=0.000017, acc@1[1]=241.5/256=0.943359375, acc@1[2]=251.5/256=0.982421875 +[2023-10-07 20:33:58,948][root][INFO] - Train batch 600 +[2023-10-07 20:33:58,949][root][INFO] - Avg. loss per last 100 batches: 0.096089 +[2023-10-07 20:33:59,635][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29294.4/29522=99.23% | mean: 0.01 | max: 5.05 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.14 | max: 6.01 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] when was the first earth day [SEP] ### +### [P_TEXT]: [CLS] april 22, 1970 a arbor day a was the first earth day. today, a common practice in ### +### celebration of earth day is still to plant new trees. cherry trees blooming in antelope valley ### +### california, as captured by our friend kerri willerford. thanks kerri! [SEP] ### +### ======================================= h_v_q | Gates: 27488 ======================================= ### +### ('earth', 0, 2) ('first', 1, 9) ('day', 2, 4) ('was', 3, 154) ('.', 4, 4449) ('1977', 5, 263) ### +### ('1979', 6, 462) ('august', 7, 237) ('sunday', 8, 550) ('september', 9, 119) ('1989', 10, 1799) ### +### ('inaugural', 11, 589) ('knew', 12, 666) ('1st', 13, 27) ('familiarity', 14, 24532) ### +### ('november', 15, 180) ('april', 16, 5) ('early', 17, 60) ('when', 18, 814) ('july', 19, 93) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('arbor', 8896, 0) ('cherry', 6750, 1) ('earth', 0, 2) ('kerr', 7537, 3) ('day', 2, 4) ### +### ('april', 16, 5) ('practice', 994, 6) ('trees', 4236, 7) ('celebration', 730, 8) ('first', 1, 9) ### +### ('bloom', 15640, 10) ('california', 163, 11) ('tree', 4418, 12) ('ant', 4356, 13) ### +### ('captured', 2261, 14) ('1970', 43, 15) ('valley', 800, 16) ('thanks', 19005, 17) ### +### ('##elo', 27798, 18) ('ˈ', 527, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('earth', 0, 2) ('day', 2, 4) ('first', 1, 9) ('april', 16, 5) ('was', 3, 154) ('1st', 13, 27) ### +### ('1970', 43, 15) ('early', 17, 60) ('june', 21, 42) ('september', 9, 119) ('july', 19, 93) ### +### ('date', 27, 54) ('august', 7, 237) ('1977', 5, 263) ('1969', 51, 34) ('november', 15, 180) ### +### ('postwar', 23, 149) ('second', 39, 69) ('initial', 26, 132) ('december', 25, 153) ### +############################################################################################################ +[2023-10-07 20:33:59,636][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:33:59,636][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:34:00,057][root][INFO] - Epoch: 4: Step: 601/1557, loss[v]=0.094779, lr=0.000016, acc@1[1]=238.5/256=0.931640625, acc@1[2]=248.5/256=0.970703125 +[2023-10-07 20:35:16,122][root][INFO] - Train batch 700 +[2023-10-07 20:35:16,123][root][INFO] - Avg. loss per last 100 batches: 0.098210 +[2023-10-07 20:35:16,828][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29227.7/29522=99.00% | mean: 0.01 | max: 4.96 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.14 | max: 5.81 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is tilth [SEP] ### +### [P_TEXT]: [CLS] wiktionary ( 0. 00 / 0 votes ) rate this definition : 1 tilth ( noun ) the state of ### +### being tilled, or prepared for a crop ; culture. the land is in good tilth and ready to plant. 2 ### +### tilth ( noun ) rich cultivated soil. what good tilth, he said, grabbing a handful of soil. [SEP] ### +### ======================================= h_v_q | Gates: 27432 ======================================= ### +### ('tilt', 0, 0) ('##h', 1, 9) ('is', 2, 461) ('tilted', 3, 22) ('tilting', 4, 20) ### +### ('definition', 5, 10) ('encompasses', 6, 17) ('.', 7, 7401) ('familiarity', 8, 24326) ### +### ('something', 9, 1774) ('alfred', 10, 351) ('noun', 11, 480) ('refers', 12, 3639) ('tip', 13, 75) ### +### ('ward', 14, 2964) ('##hi', 15, 1355) ('coach', 16, 5795) ('stands', 17, 4344) ('tower', 18, 1189) ### +### ('emma', 19, 2074) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('tilt', 0, 0) ('ready', 680, 1) ('soil', 440, 2) ('crop', 7600, 3) ('wi', 5704, 4) ### +### ('till', 4263, 5) ('handful', 14712, 6) ('definitions', 1979, 7) ('define', 12843, 8) ('##h', 1, 9) ### +### ('definition', 5, 10) ('##ry', 7055, 11) ('meaning', 2413, 12) ('cultivated', 8571, 13) ### +### ('ˈ', 1006, 14) ('##kti', 28125, 15) ('land', 1211, 16) ('encompasses', 6, 17) ('00', 16931, 18) ### +### ('plant', 4021, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('tilt', 0, 0) ('##h', 1, 9) ('definition', 5, 10) ('tilted', 3, 22) ('tilting', 4, 20) ### +### ('encompasses', 6, 17) ('is', 2, 461) ('tip', 13, 75) ('soil', 440, 2) ('ready', 680, 1) ### +### ('tipped', 52, 50) ('anger', 42, 66) ('−', 168, 33) ('hugh', 59, 58) ('gideon', 104, 47) ### +### ('being', 171, 38) ('##α', 182, 39) ('afraid', 95, 68) ('crops', 453, 29) ('cultural', 102, 61) ### +############################################################################################################ +[2023-10-07 20:35:16,829][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:35:16,829][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:35:17,252][root][INFO] - Epoch: 4: Step: 701/1557, loss[v]=0.063694, lr=0.000016, acc@1[1]=243.0/256=0.94921875, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 20:36:33,921][root][INFO] - Train batch 800 +[2023-10-07 20:36:33,922][root][INFO] - Avg. loss per last 100 batches: 0.100258 +[2023-10-07 20:36:34,622][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29167.2/29522=98.80% | mean: 0.01 | max: 5.38 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 6.00 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] temperature in el paso tx [SEP] ### +### [P_TEXT]: [CLS] days of hot weather in el paso. heat is no stranger to el paso. most summer ### +### afternoons are spent in the 90s fahrenheit with more than a handful climbing above that. typically ### +### 14 days a year see the temperature peak at 100 a°f or more. about half the days with 100 - degree ### +### weather occur in june with the rest sprinkled from may to september. [SEP] ### +### ======================================= h_v_q | Gates: 27256 ======================================= ### +### ('paso', 0, 0) ('temperature', 1, 2) ('el', 2, 5) ('texas', 3, 127) ('tx', 4, 283) ('##°', 5, 8) ### +### ('temperatures', 6, 39) ('familiarity', 7, 25053) ('.', 8, 8294) ('climate', 9, 21) ### +### ('colorado', 10, 516) ('arkansas', 11, 2438) ('louisiana', 12, 4591) ('missouri', 13, 11860) ### +### ('dallas', 14, 1024) ('al', 15, 484) ('illinois', 16, 6488) ('minnesota', 17, 13465) ### +### ('houston', 18, 955) ('pennsylvania', 19, 1392) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('paso', 0, 0) ('weather', 153, 1) ('temperature', 1, 2) ('summer', 34, 3) ('stranger', 3353, 4) ### +### ('el', 2, 5) ('hot', 33, 6) ('afternoon', 1179, 7) ('##°', 5, 8) ('##kled', 25874, 9) ### +### ('°f', 75, 10) ('handful', 15572, 11) ('june', 105, 12) ('peak', 1392, 13) ('degree', 258, 14) ### +### ('ˈ', 1026, 15) ('evening', 1095, 16) ('winter', 182, 17) ('afternoons', 26433, 18) ### +### ('heat', 61, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('paso', 0, 0) ('temperature', 1, 2) ('el', 2, 5) ('texas', 3, 127) ('##°', 5, 8) ### +### ('temperatures', 6, 39) ('climate', 9, 21) ('tx', 4, 283) ('summer', 34, 3) ('hot', 33, 6) ### +### ('weather', 153, 1) ('°f', 75, 10) ('colorado', 10, 516) ('heat', 61, 19) ('june', 105, 12) ### +### ('antonio', 37, 81) ('##₂', 115, 28) ('simon', 57, 72) ('winter', 182, 17) ('warm', 92, 44) ### +############################################################################################################ +[2023-10-07 20:36:34,623][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:36:34,623][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:36:35,047][root][INFO] - Epoch: 4: Step: 801/1557, loss[v]=0.078845, lr=0.000016, acc@1[1]=245.0/256=0.95703125, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 20:37:51,323][root][INFO] - Train batch 900 +[2023-10-07 20:37:51,324][root][INFO] - Avg. loss per last 100 batches: 0.092697 +[2023-10-07 20:37:52,005][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29234.2/29522=99.02% | mean: 0.01 | max: 4.91 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.17 | max: 6.02 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how far is mpls mn to boston ma [SEP] ### +### [P_TEXT]: [CLS] distance from minneapolis, mn to boston, ma. the total distance from minneapolis, ### +### mn to boston, ma is 1, 124 miles. this is equivalent to 1a808 kilometers or 976 nautical miles. ### +### your trip begins in minneapolis, minnesota. it ends in boston, massachusetts. your flight direction ### +### from minneapolis, mn to boston, ma is east ( 91 degrees from north ). [SEP] ### +### ======================================= h_v_q | Gates: 27908 ======================================= ### +### ('mp', 0, 1262) ('boston', 1, 2) ('##ls', 2, 10232) ('distance', 3, 3) ('far', 4, 5) ('mn', 5, 11) ### +### ('massachusetts', 6, 21) ('miles', 7, 4) ('minnesota', 8, 6) ('farther', 9, 15) ('ma', 10, 12) ### +### ('familiarity', 11, 19651) ('mps', 12, 14030) ('to', 13, 1961) ('hampshire', 14, 7049) ### +### ('distances', 15, 13) ('virginia', 16, 2819) ('pennsylvania', 17, 765) ('kent', 18, 1508) ### +### ('colorado', 19, 3175) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('minneapolis', 801, 0) ('nautical', 8983, 1) ('boston', 1, 2) ('distance', 3, 3) ('miles', 7, 4) ### +### ('far', 4, 5) ('minnesota', 8, 6) ('trip', 260, 7) ('equivalent', 14385, 8) ('flight', 677, 9) ### +### ('kilometers', 973, 10) ('mn', 5, 11) ('ma', 10, 12) ('distances', 15, 13) ('1a', 13294, 14) ### +### ('farther', 9, 15) ('total', 616, 16) ('ˈ', 109, 17) ('east', 130, 18) ('direction', 3094, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('boston', 1, 2) ('distance', 3, 3) ('far', 4, 5) ('miles', 7, 4) ('minnesota', 8, 6) ('mn', 5, 11) ### +### ('massachusetts', 6, 21) ('mp', 0, 1262) ('ma', 10, 12) ('farther', 9, 15) ('distances', 15, 13) ### +### ('mile', 34, 27) ('##ls', 2, 10232) ('distant', 36, 34) ('harvard', 29, 69) ('length', 21, 160) ### +### ('ˈ', 109, 17) ('minneapolis', 801, 0) ('height', 25, 167) ('trip', 260, 7) ### +############################################################################################################ +[2023-10-07 20:37:52,005][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:37:52,005][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:37:52,430][root][INFO] - Epoch: 4: Step: 901/1557, loss[v]=0.054970, lr=0.000016, acc@1[1]=245.5/256=0.958984375, acc@1[2]=252.0/256=0.984375 +[2023-10-07 20:39:09,326][root][INFO] - Train batch 1000 +[2023-10-07 20:39:09,327][root][INFO] - Avg. loss per last 100 batches: 0.098275 +[2023-10-07 20:39:10,033][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29270.2/29522=99.15% | mean: 0.01 | max: 5.05 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.17 | max: 5.96 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] why ask powerful coaching questions [SEP] ### +### [P_TEXT]: [CLS] what would a professional coach be without questions, especially powerful and ### +### empowering questions? every coach knows that along with the skill of listening, powerful ### +### questioning is the key skill to master to be brilliant at what you do. effective, empowering ### +### questions evoke inspiration, creativity, motivation, and self - discovery for your clients. [SEP] ### +### ======================================= h_v_q | Gates: 27593 ======================================= ### +### ('coaching', 0, 7) ('powerful', 1, 3) ('questions', 2, 4) ('why', 3, 1505) ('coach', 4, 0) ### +### ('ask', 5, 305) ('power', 6, 42) ('strong', 7, 36) ('question', 8, 20) ('because', 9, 1649) ### +### ('reasons', 10, 14794) ('answer', 11, 645) ('teaching', 12, 965) ('asked', 13, 99) ### +### ('training', 14, 222) ('familiarity', 15, 24552) ('.', 16, 7757) ('say', 17, 4002) ### +### ('reason', 18, 5178) ('coached', 19, 13) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('coach', 4, 0) ('##powering', 8421, 1) ('brilliant', 329, 2) ('powerful', 1, 3) ### +### ('questions', 2, 4) ('master', 2050, 5) ('coaches', 23, 6) ('coaching', 0, 7) ('effective', 28, 8) ### +### ('listening', 9414, 9) ('professional', 322, 10) ('skill', 3528, 11) ('questioning', 1420, 12) ### +### ('coached', 19, 13) ('inspiration', 4417, 14) ('em', 5149, 15) ('ev', 18628, 16) ('key', 260, 17) ### +### ('clients', 3921, 18) ('ˈ', 662, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('coaching', 0, 7) ('powerful', 1, 3) ('questions', 2, 4) ('coach', 4, 0) ('power', 6, 42) ### +### ('question', 8, 20) ('strong', 7, 36) ('ask', 5, 305) ('coached', 19, 13) ('why', 3, 1505) ### +### ('coaches', 23, 6) ('effective', 28, 8) ('asked', 13, 99) ('training', 14, 222) ('potent', 20, 77) ### +### ('answer', 11, 645) ('because', 9, 1649) ('anger', 42, 61) ('brilliant', 329, 2) ### +### ('sharply', 82, 33) ### +############################################################################################################ +[2023-10-07 20:39:10,034][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:39:10,034][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:39:10,439][root][INFO] - Epoch: 4: Step: 1001/1557, loss[v]=0.098554, lr=0.000016, acc@1[1]=241.0/256=0.94140625, acc@1[2]=249.5/256=0.974609375 +[2023-10-07 20:40:26,977][root][INFO] - Train batch 1100 +[2023-10-07 20:40:26,978][root][INFO] - Avg. loss per last 100 batches: 0.092795 +[2023-10-07 20:40:27,681][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29275.0/29522=99.16% | mean: 0.01 | max: 5.44 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 5.95 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is main source of vitamin a [SEP] ### +### [P_TEXT]: [CLS] there are two main sources of vitamin a, animal sources and plant sources. known as ### +### retinol, animal sources of this essential vitamin contain preformed vitamin a. vegetable sources ### +### contain carotenoids, such as beta - carotene, which is converted to retinol by the body. itamin a ### +### that comes from animal sources is well absorbed and used more efficiently by the body then other ### +### sources of the vitamin. the two main sources of vitamin a from animal products are beef liver and ### +### turkey giblets. other sources of this fat soluble vitamin include : 1 chicken liver. 2 cod liver ### +### oil. [SEP] ### +### ======================================= h_v_q | Gates: 27480 ======================================= ### +### ('vitamin', 0, 0) ('main', 1, 18) ('source', 2, 4) ('sources', 3, 3) ('primary', 4, 520) ### +### ('a', 5, 839) ('principal', 6, 816) ('is', 7, 8183) ('.', 8, 12600) ('encompasses', 9, 110) ### +### ('definition', 10, 1833) ('familiarity', 11, 28128) ('largest', 12, 1484) ('top', 13, 1088) ### +### ('major', 14, 968) ('relating', 15, 11062) ('refers', 16, 19013) ('something', 17, 4797) ### +### ('outlets', 18, 79) ('of', 19, 15765) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('vitamin', 0, 0) ('##formed', 21944, 1) ('animal', 2297, 2) ('sources', 3, 3) ('source', 2, 4) ### +### ('essential', 1698, 5) ('re', 10217, 6) ('liver', 14814, 7) ('##tino', 29374, 8) ### +### ('vegetable', 8087, 9) ('animals', 3673, 10) ('##ote', 19316, 11) ('beef', 4808, 12) ### +### ('vegetables', 4265, 13) ('absorbed', 6569, 14) ('nutrients', 2211, 15) ('−', 112, 16) ### +### ('ˈ', 165, 17) ('main', 1, 18) ('##ids', 25726, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('vitamin', 0, 0) ('source', 2, 4) ('main', 1, 18) ('sources', 3, 3) ('primary', 4, 520) ### +### ('a', 5, 839) ('encompasses', 9, 110) ('principal', 6, 816) ('outlets', 18, 79) ('−', 112, 16) ### +### ('jeremy', 37, 83) ('sharply', 79, 32) ('supplies', 40, 91) ('ˈ', 165, 17) ('transmitter', 30, 137) ### +### ('prominent', 27, 204) ('##₂', 156, 23) ('а', 62, 85) ('biggest', 36, 133) ('simon', 49, 106) ### +############################################################################################################ +[2023-10-07 20:40:27,682][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:40:27,682][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:40:28,104][root][INFO] - Epoch: 4: Step: 1101/1557, loss[v]=0.162869, lr=0.000016, acc@1[1]=232.0/256=0.90625, acc@1[2]=246.0/256=0.9609375 +[2023-10-07 20:41:44,659][root][INFO] - Train batch 1200 +[2023-10-07 20:41:44,660][root][INFO] - Avg. loss per last 100 batches: 0.102726 +[2023-10-07 20:41:45,383][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29223.5/29522=98.99% | mean: 0.01 | max: 5.70 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 6.12 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what elements are in the massive star [SEP] ### +### [P_TEXT]: [CLS] leaving main life sequence. when a massive star has expended all of its core ### +### hydrogen, this means that the gravitational forces pushing inward will overcome the fusion forces ### +### which had been pushing outward. the outer layers of the star will then compact inwards. [SEP] ### +### ======================================= h_v_q | Gates: 27840 ======================================= ### +### ('massive', 0, 11) ('star', 1, 3) ('elements', 2, 658) ('element', 3, 813) ('heavy', 4, 110) ### +### ('are', 5, 14944) ('large', 6, 56) ('huge', 7, 102) ('familiarity', 8, 25111) ('stars', 9, 13) ### +### ('giant', 10, 970) ('components', 11, 889) ('.', 12, 5716) ('include', 13, 1116) ### +### ('powerful', 14, 602) ('largest', 15, 277) ('julian', 16, 308) ('sun', 17, 158) ('##₂', 18, 28) ### +### ('starring', 19, 126) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('gravitational', 7539, 0) ('life', 792, 1) ('compact', 1139, 2) ('star', 1, 3) ### +### ('hydrogen', 2975, 4) ('sequence', 2742, 5) ('layers', 4838, 6) ('leaving', 1056, 7) ### +### ('overcome', 3301, 8) ('pushing', 739, 9) ('fusion', 3439, 10) ('massive', 0, 11) ### +### ('core', 1194, 12) ('stars', 9, 13) ('##ded', 9786, 14) ('inward', 28237, 15) ('main', 1691, 16) ### +### ('outer', 2651, 17) ('means', 6073, 18) ('forces', 44, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('star', 1, 3) ('massive', 0, 11) ('elements', 2, 658) ('stars', 9, 13) ('large', 6, 56) ### +### ('heavy', 4, 110) ('element', 3, 813) ('huge', 7, 102) ('##₂', 18, 28) ('forces', 44, 19) ### +### ('encompasses', 31, 29) ('starring', 19, 126) ('sun', 17, 158) ('big', 32, 83) ('##α', 69, 33) ### +### ('giant', 10, 970) ('simon', 45, 70) ('largest', 15, 277) ('gideon', 143, 24) ('−', 112, 27) ### +############################################################################################################ +[2023-10-07 20:41:45,384][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:41:45,384][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:41:45,810][root][INFO] - Epoch: 4: Step: 1201/1557, loss[v]=0.070322, lr=0.000016, acc@1[1]=242.5/256=0.947265625, acc@1[2]=249.5/256=0.974609375 +[2023-10-07 20:43:02,953][root][INFO] - Train batch 1300 +[2023-10-07 20:43:02,954][root][INFO] - Avg. loss per last 100 batches: 0.091698 +[2023-10-07 20:43:03,656][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29316.9/29522=99.31% | mean: 0.01 | max: 5.41 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.17 | max: 6.09 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is metamorphic rock used for? [SEP] ### +### [P_TEXT]: [CLS] < br / > < br / > < b > slate < / b > is another metamorphic rock that is used in ### +### buildings for floor and roofing tiles, and it was once used for blackboards. < br / > < br / > < b ### +### > quartzite < / b > is used as a source of silica ; other forms of metamorphic rocks are useful as ### +### building components, monuments, counter tops, and facings. [SEP] ### +### ======================================= h_v_q | Gates: 28509 ======================================= ### +### ('##morphic', 0, 3) ('meta', 1, 6) ('rock', 2, 7) ('used', 3, 38) ('familiarity', 4, 25339) ### +### ('employed', 5, 64) ('.', 6, 6973) ('##morphism', 7, 207) ('use', 8, 96) ('definition', 9, 63) ### +### ('ana', 10, 1002) ('alfred', 11, 241) ('cited', 12, 226) ('usage', 13, 86) ('relating', 14, 15170) ### +### ('uses', 15, 74) ('using', 16, 623) ('encompasses', 17, 19) ('useful', 18, 13) ('is', 19, 169) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('slate', 2739, 0) ('br', 9707, 1) ('b', 5656, 2) ('##morphic', 0, 3) ('quartz', 5441, 4) ### +### ('<', 15316, 5) ('meta', 1, 6) ('rock', 2, 7) ('buildings', 1518, 8) ('tiles', 16355, 9) ### +### ('tops', 6870, 10) ('facing', 3188, 11) ('##lica', 16586, 12) ('useful', 18, 13) ### +### ('##ite', 8860, 14) ('rocks', 94, 15) ('black', 3639, 16) ('##board', 10583, 17) ('roof', 2430, 18) ### +### ('encompasses', 17, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##morphic', 0, 3) ('meta', 1, 6) ('rock', 2, 7) ('used', 3, 38) ('employed', 5, 64) ### +### ('useful', 18, 13) ('encompasses', 17, 19) ('definition', 9, 63) ('use', 8, 96) ('uses', 15, 74) ### +### ('usage', 13, 86) ('##₂', 31, 39) ('−', 41, 30) ('rocks', 94, 15) ('ˈ', 55, 22) ### +### ('##morphism', 7, 207) ('##α', 54, 44) ('sharply', 106, 32) ('crashing', 146, 27) ### +### ('hating', 105, 35) ### +############################################################################################################ +[2023-10-07 20:43:03,657][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:43:03,657][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:43:04,061][root][INFO] - Epoch: 4: Step: 1301/1557, loss[v]=0.064492, lr=0.000016, acc@1[1]=245.0/256=0.95703125, acc@1[2]=252.0/256=0.984375 +[2023-10-07 20:44:21,922][root][INFO] - Train batch 1400 +[2023-10-07 20:44:21,925][root][INFO] - Avg. loss per last 100 batches: 0.095286 +[2023-10-07 20:44:22,631][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29228.4/29522=99.01% | mean: 0.01 | max: 5.40 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 6.19 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the postsynaptic density [SEP] ### +### [P_TEXT]: [CLS] the postsynaptic density ( psd ) is a protein dense specialization attached to the ### +### postsynaptic membrane. psds were originally identified by electron microscopy as an electron - ### +### dense region at the membrane of a postsynaptic neuron. he structure and composition of the psd have ### +### been the focus of numerous molecular studies of synaptic plasticity, a cellular model of learning ### +### and memory. psds are sized on the order of 250 to 500 nanometres in diameter and 25 to 50 ### +### nanometres in thickness, depending on the activity state of the synapse. [SEP] ### +### ======================================= h_v_q | Gates: 28093 ======================================= ### +### ('density', 0, 0) ('##ptic', 1, 2) ('post', 2, 18) ('posts', 3, 23) ('##yna', 4, 27) ### +### ('definition', 5, 39) ('is', 6, 254) ('familiarity', 7, 25670) ('encompasses', 8, 10) ### +### ('positions', 9, 2190) ('relating', 10, 12924) ('something', 11, 3290) ('.', 12, 11386) ### +### ('refers', 13, 2431) ('alfred', 14, 235) ('evidence', 15, 1095) ('stands', 16, 2696) ### +### ('electronic', 17, 4128) ('position', 18, 4967) ('corps', 19, 8731) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('density', 0, 0) ('ps', 924, 1) ('##ptic', 1, 2) ('##ds', 4588, 3) ('dense', 67, 4) ### +### ('##d', 10976, 5) ('syn', 8815, 6) ('diameter', 1213, 7) ('protein', 1721, 8) ### +### ('microscopy', 24822, 9) ('encompasses', 8, 10) ('thickness', 2232, 11) ('plastic', 2048, 12) ### +### ('depending', 19460, 13) ('nano', 1752, 14) ('molecular', 86, 15) ('ˈ', 138, 16) ### +### ('membrane', 172, 17) ('post', 2, 18) ('identified', 7056, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('density', 0, 0) ('##ptic', 1, 2) ('post', 2, 18) ('posts', 3, 23) ('##yna', 4, 27) ### +### ('definition', 5, 39) ('encompasses', 8, 10) ('is', 6, 254) ('dense', 67, 4) ('molecular', 86, 15) ### +### ('##₂', 45, 32) ('hating', 50, 41) ('ˈ', 138, 16) ('−', 70, 36) ('sharply', 66, 37) ### +### ('membrane', 172, 17) ('wingspan', 128, 24) ('hesitated', 102, 31) ('alfred', 14, 235) ### +### ('posted', 33, 95) ### +############################################################################################################ +[2023-10-07 20:44:22,631][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:44:22,631][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:44:23,055][root][INFO] - Epoch: 4: Step: 1401/1557, loss[v]=0.094048, lr=0.000016, acc@1[1]=241.5/256=0.943359375, acc@1[2]=247.5/256=0.966796875 +[2023-10-07 20:45:39,819][root][INFO] - Train batch 1500 +[2023-10-07 20:45:39,820][root][INFO] - Avg. loss per last 100 batches: 0.092892 +[2023-10-07 20:45:40,519][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29233.4/29522=99.02% | mean: 0.01 | max: 5.09 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.15 | max: 5.84 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what format is mla [SEP] ### +### [P_TEXT]: [CLS] mla style specifies guidelines for formatting manuscripts and using the english ### +### language in writing. mla style also provides writers with a system for referencing their sources ### +### through parenthetical citation in their essays and works cited pages. [SEP] ### +### ======================================= h_v_q | Gates: 27112 ======================================= ### +### ('mla', 0, 0) ('format', 1, 9) ('encompasses', 2, 25) ('.', 3, 2433) ('is', 4, 369) ### +### ('definition', 5, 109) ('formats', 6, 23) ('familiarity', 7, 24580) ('noun', 8, 21657) ### +### ('form', 9, 186) ('or', 10, 5012) ('style', 11, 2) ('series', 12, 644) ('refers', 13, 3036) ### +### ('language', 14, 10) ('relating', 15, 3702) ('mor', 16, 730) ('ba', 17, 465) ('"', 18, 3440) ### +### ('stands', 19, 1673) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('mla', 0, 0) ('manuscripts', 12217, 1) ('style', 11, 2) ('guidelines', 3038, 3) ### +### ('writing', 2454, 4) ('##hetic', 23967, 5) ('english', 297, 6) ('writers', 6436, 7) ### +### ('styles', 1125, 8) ('format', 1, 9) ('language', 14, 10) ('citation', 4775, 11) ### +### ('manuscript', 4955, 12) ('cited', 5901, 13) ('essays', 15896, 14) ('parent', 5638, 15) ### +### ('pages', 8431, 16) ('writer', 3082, 17) ('ˈ', 2097, 18) ('write', 15777, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('mla', 0, 0) ('format', 1, 9) ('encompasses', 2, 25) ('formats', 6, 23) ('style', 11, 2) ### +### ('definition', 5, 109) ('language', 14, 10) ('is', 4, 369) ('form', 9, 186) ('.', 3, 2433) ### +### ('genre', 22, 51) ('english', 297, 6) ('mp', 34, 75) ('describes', 40, 63) ('anger', 80, 57) ### +### ('hating', 277, 26) ('system', 55, 86) ('presenter', 167, 45) ('literature', 46, 102) ### +### ('albert', 24, 185) ### +############################################################################################################ +[2023-10-07 20:45:40,520][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:45:40,520][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:45:40,926][root][INFO] - Epoch: 4: Step: 1501/1557, loss[v]=0.111734, lr=0.000016, acc@1[1]=235.5/256=0.919921875, acc@1[2]=247.0/256=0.96484375 +[2023-10-07 20:46:24,152][root][INFO] - rank=1; last iteration 1557 +[2023-10-07 20:46:24,152][root][INFO] - rank=0; last iteration 1557 +[2023-10-07 20:46:24,152][root][INFO] - rank=3; last iteration 1557 +[2023-10-07 20:46:24,153][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 20:46:24,152][root][INFO] - rank=2; last iteration 1557 +[2023-10-07 20:46:24,153][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 20:46:24,153][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-07 20:46:24,153][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 20:46:24,153][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-07 20:46:24,153][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 20:46:24,153][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-07 20:46:24,153][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-07 20:46:24,160][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 20:46:24,160][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 20:46:24,161][root][INFO] - Epoch finished on 1 +[2023-10-07 20:46:24,161][root][INFO] - Epoch finished on 0 +[2023-10-07 20:46:24,161][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 20:46:24,161][root][INFO] - Epoch finished on 2 +[2023-10-07 20:46:24,161][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 20:46:24,162][root][INFO] - Epoch finished on 3 +[2023-10-07 20:46:35,110][root][INFO] - Saved checkpoint at ./vdr_4 +[2023-10-07 20:46:35,110][root][INFO] - Saved checkpoint at ./vdr_4 +[2023-10-07 20:46:35,111][root][INFO] - Av Loss per epoch=0.096018 +[2023-10-07 20:46:35,111][root][INFO] - Av Loss per epoch=0.096018 +[2023-10-07 20:46:35,112][root][INFO] - epoch total (1) correct predictions=374369 +[2023-10-07 20:46:35,112][root][INFO] - epoch total (1) correct predictions=374369 +[2023-10-07 20:46:35,112][root][INFO] - epoch total (2) correct predictions=387105 +[2023-10-07 20:46:35,112][root][INFO] - epoch total (2) correct predictions=387105 +[2023-10-07 20:46:35,112][root][INFO] - Saved checkpoint at ./vdr_4 +[2023-10-07 20:46:35,113][root][INFO] - Av Loss per epoch=0.096018 +[2023-10-07 20:46:35,114][root][INFO] - epoch total (1) correct predictions=374369 +[2023-10-07 20:46:35,114][root][INFO] - epoch total (2) correct predictions=387105 +[2023-10-07 20:46:35,113][root][INFO] - Saved checkpoint at ./vdr_4 +[2023-10-07 20:46:35,114][root][INFO] - Av Loss per epoch=0.096018 +[2023-10-07 20:46:35,115][root][INFO] - epoch total (1) correct predictions=374369 +[2023-10-07 20:46:35,115][root][INFO] - epoch total (2) correct predictions=387105 +[2023-10-07 20:46:35,116][root][INFO] - ***** Epoch 5 ***** +[2023-10-07 20:46:35,116][root][INFO] - ***** Epoch 5 ***** +[2023-10-07 20:46:35,118][root][INFO] - ***** Epoch 5 ***** +[2023-10-07 20:46:35,119][root][INFO] - ***** Epoch 5 ***** +[2023-10-07 20:46:35,122][root][INFO] - rank=0; Iteration start +[2023-10-07 20:46:35,122][root][INFO] - rank=3; Iteration start +[2023-10-07 20:46:35,122][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 20:46:35,123][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 20:46:35,123][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 20:46:35,123][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 20:46:35,124][root][INFO] - rank=2; Iteration start +[2023-10-07 20:46:35,124][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-07 20:46:35,125][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 20:46:35,125][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 20:46:35,125][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-07 20:46:35,125][root][INFO] - rank=1; Iteration start +[2023-10-07 20:46:35,125][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 20:46:35,125][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 20:46:35,126][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-07 20:46:35,126][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-07 20:46:36,115][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29236.4/29522=99.03% | mean: 0.01 | max: 4.97 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 5.98 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] where is the aegean sea located on the world map [SEP] ### +### [P_TEXT]: [CLS] submit questions - new articles. the aegean sea is an elongated bay of the ### +### mediterranean sea located between the mainland of greece and turkey. it covers about 214, 000 ### +### square kilometers ( 83, 000 square miles ) in area. east of crete the sea reaches its maximum depth ### +### of 3, 543 meters ( 11, 624 feet ). ubmit questions - new articles. the aegean sea is an elongated ### +### bay of the mediterranean sea located between the mainland of greece and turkey. it covers about ### +### 214, 000 square kilometers ( 83, 000 square miles ) in area. east of crete the sea reaches its ### +### maximum depth of 3, 543 meters ( 11, 624 feet ). [SEP] ### +### ======================================= h_v_q | Gates: 27368 ======================================= ### +### ('aegean', 0, 1) ('sea', 1, 3) ('located', 2, 33) ('world', 3, 15813) ('map', 4, 6529) ### +### ('familiarity', 5, 26996) ('situated', 6, 73) ('is', 7, 114) ('.', 8, 5552) ('india', 9, 3547) ### +### ('mediterranean', 10, 7) ('ocean', 11, 23) ('greece', 12, 2) ('relating', 13, 14910) ### +### ('states', 14, 12078) ('washington', 15, 7538) ('brazil', 16, 4495) ('connecticut', 17, 14570) ### +### ('global', 18, 4842) ('pacific', 19, 550) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('mainland', 5917, 0) ('aegean', 0, 1) ('greece', 12, 2) ('sea', 1, 3) ('bay', 168, 4) ### +### ('crete', 235, 5) ('depth', 9457, 6) ('mediterranean', 10, 7) ('elongated', 5499, 8) ### +### ('turkey', 858, 9) ('encompasses', 25, 10) ('submit', 12240, 11) ('miles', 354, 12) ### +### ('feet', 2540, 13) ('kilometers', 19878, 14) ('bays', 14908, 15) ('turkish', 749, 16) ### +### ('depths', 13828, 17) ('questions', 4851, 18) ('##bm', 22787, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('aegean', 0, 1) ('sea', 1, 3) ('located', 2, 33) ('greece', 12, 2) ('mediterranean', 10, 7) ### +### ('situated', 6, 73) ('ocean', 11, 23) ('encompasses', 25, 10) ('is', 7, 114) ('greek', 22, 27) ### +### ('bay', 168, 4) ('where', 34, 47) ('crete', 235, 5) ('east', 97, 20) ('seas', 73, 29) ### +### ('numerous', 95, 31) ('marine', 37, 148) ('miles', 354, 12) ('##ο', 156, 28) ('##₂', 77, 68) ### +############################################################################################################ +[2023-10-07 20:46:36,115][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:46:36,115][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:46:36,513][root][INFO] - Epoch: 5: Step: 1/1557, loss[v]=0.107257, lr=0.000016, acc@1[1]=241.0/256=0.94140625, acc@1[2]=251.5/256=0.982421875 +[2023-10-07 20:47:53,680][root][INFO] - Train batch 100 +[2023-10-07 20:47:53,681][root][INFO] - Avg. loss per last 100 batches: 0.096501 +[2023-10-07 20:47:54,398][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29345.2/29522=99.40% | mean: 0.01 | max: 5.14 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.02 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] where is 18417 zip [SEP] ### +### [P_TEXT]: [CLS] 18417 is a sparsely populated, rural zip code in equinunk, pennsylvania. the ### +### population is primarily white, older, and mostly married couples. the median age here is 49. there ### +### are 581 men and 540 women. the median age for men is 49 while for women the median age is 49. n the ### +### 18417 zip code, the registered sex offenders can be found at http : / / www. pameganslaw. state. ### +### pa. us /. please be sure to read all the disclaimers provided by the state. for more information ### +### check out the resources listed below : [SEP] ### +### ======================================= h_v_q | Gates: 28522 ======================================= ### +### ('zip', 0, 7) ('##7', 1, 5) ('1841', 2, 12) ('located', 3, 862) ('is', 4, 234) ('##8', 5, 10) ### +### ('familiarity', 6, 27023) ('seven', 7, 203) ('wherever', 8, 5207) ('1911', 9, 109) ('##9', 10, 17) ### +### ('##6', 11, 29) ('7', 12, 127) ('1942', 13, 208) ('1840', 14, 101) ('1941', 15, 172) ### +### ('downtown', 16, 712) ('1945', 17, 419) ('.', 18, 12059) ('south', 19, 3061) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##unk', 22103, 0) ('rural', 159, 1) ('pennsylvania', 46, 2) ('age', 5505, 3) ### +### ('sparsely', 15268, 4) ('##7', 1, 5) ('population', 1769, 6) ('zip', 0, 7) ('disc', 6694, 8) ### +### ('pam', 11109, 9) ('##8', 5, 10) ('pa', 401, 11) ('1841', 2, 12) ('here', 6395, 13) ### +### ('median', 23066, 14) ('populated', 8178, 15) ('offenders', 18726, 16) ('##9', 10, 17) ### +### ('sex', 5135, 18) ('married', 4786, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('zip', 0, 7) ('##7', 1, 5) ('1841', 2, 12) ('##8', 5, 10) ('##9', 10, 17) ('##6', 11, 29) ### +### ('pennsylvania', 46, 2) ('##1', 21, 37) ('is', 4, 234) ('1911', 9, 109) ('encompasses', 40, 22) ### +### ('1840', 14, 101) ('located', 3, 862) ('7', 12, 127) ('rural', 159, 1) ('seven', 7, 203) ### +### ('1941', 15, 172) ('##3', 33, 62) ('1843', 28, 79) ('1842', 36, 68) ### +############################################################################################################ +[2023-10-07 20:47:54,399][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:47:54,399][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:47:54,818][root][INFO] - Epoch: 5: Step: 101/1557, loss[v]=0.101361, lr=0.000016, acc@1[1]=241.0/256=0.94140625, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 20:49:11,233][root][INFO] - Train batch 200 +[2023-10-07 20:49:11,234][root][INFO] - Avg. loss per last 100 batches: 0.087572 +[2023-10-07 20:49:11,944][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29302.9/29522=99.26% | mean: 0.01 | max: 5.35 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 5.97 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what was the bear flag revolt a reaction to [SEP] ### +### [P_TEXT]: [CLS] during the bear flag revolt, from june to july 1846, a small group of american ### +### settlers in california rebelled against the mexican government and proclaimed california an ### +### independent republic. [SEP] ### +### ======================================= h_v_q | Gates: 27762 ======================================= ### +### ('flag', 0, 0) ('reaction', 1, 10529) ('bear', 2, 2) ('revolt', 3, 1) ('response', 4, 5536) ### +### ('was', 5, 3618) ('rebellion', 6, 5) ('.', 7, 6681) ('reactions', 8, 21452) ('rebel', 9, 29) ### +### ('familiarity', 10, 28190) ('protest', 11, 252) ('something', 12, 3144) ('flags', 13, 18) ### +### ('knew', 14, 257) ('relating', 15, 17218) ('ended', 16, 221) ('attention', 17, 4955) ### +### ('answer', 18, 9619) ('became', 19, 503) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('flag', 0, 0) ('revolt', 3, 1) ('bear', 2, 2) ('california', 642, 3) ('mexican', 2690, 4) ### +### ('rebellion', 6, 5) ('proclaimed', 1845, 6) ('settlers', 3778, 7) ('rebelled', 4748, 8) ### +### ('june', 1178, 9) ('uprising', 83, 10) ('july', 590, 11) ('mexico', 1325, 12) ### +### ('declared', 2385, 13) ('bears', 36, 14) ('ˈ', 57, 15) ('mutiny', 269, 16) ('against', 1356, 17) ### +### ('flags', 13, 18) ('wingspan', 78, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('flag', 0, 0) ('bear', 2, 2) ('revolt', 3, 1) ('rebellion', 6, 5) ('rebel', 9, 29) ### +### ('flags', 13, 18) ('reaction', 1, 10529) ('was', 5, 3618) ('response', 4, 5536) ('bears', 36, 14) ### +### ('government', 29, 23) ('protest', 11, 252) ('ˈ', 57, 15) ('uprising', 83, 10) ('##ο', 40, 31) ### +### ('##₂', 24, 65) ('wingspan', 78, 19) ('−', 59, 34) ('knew', 14, 257) ('anger', 41, 55) ### +############################################################################################################ +[2023-10-07 20:49:11,944][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:49:11,944][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:49:12,367][root][INFO] - Epoch: 5: Step: 201/1557, loss[v]=0.105675, lr=0.000016, acc@1[1]=237.5/256=0.927734375, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 20:50:28,500][root][INFO] - Train batch 300 +[2023-10-07 20:50:28,500][root][INFO] - Avg. loss per last 100 batches: 0.089900 +[2023-10-07 20:50:29,213][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29208.4/29522=98.94% | mean: 0.01 | max: 5.33 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.16 | max: 6.11 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] causes of deforestation nasa [SEP] ### +### [P_TEXT]: [CLS] the single biggest direct cause of tropical deforestation is conversion to cropland ### +### and pasture, mostly for subsistence, which is growing crops or raising livestock to meet daily ### +### needs. the conversion to agricultural land usually results from multiple direct factors. [SEP] ### +### ======================================= h_v_q | Gates: 27265 ======================================= ### +### ('##orestation', 0, 0) ('nasa', 1, 706) ('def', 2, 8) ('causes', 3, 10) ('cause', 4, 16) ### +### ('.', 5, 10403) ('caused', 6, 136) ('familiarity', 7, 27379) ('reasons', 8, 738) ('of', 9, 23215) ### +### ('relating', 10, 13949) ('sources', 11, 2482) ('alfred', 12, 146) ('factors', 13, 20) ### +### ('forest', 14, 118) ('development', 15, 5899) ('software', 16, 13940) ('richard', 17, 196) ### +### ('command', 18, 1471) ('albert', 19, 451) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##orestation', 0, 0) ('tropical', 3326, 1) ('subsistence', 7575, 2) ('pasture', 10533, 3) ### +### ('biggest', 4876, 4) ('crop', 4498, 5) ('livestock', 8245, 6) ('agricultural', 453, 7) ### +### ('def', 2, 8) ('crops', 607, 9) ('causes', 3, 10) ('conversion', 4919, 11) ('agriculture', 416, 12) ### +### ('direct', 10709, 13) ('wingspan', 28, 14) ('##land', 2031, 15) ('cause', 4, 16) ('land', 1008, 17) ### +### ('results', 2310, 18) ('multiple', 1021, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##orestation', 0, 0) ('def', 2, 8) ('causes', 3, 10) ('cause', 4, 16) ('nasa', 1, 706) ### +### ('caused', 6, 136) ('factors', 13, 20) ('wingspan', 28, 14) ('reasons', 8, 738) ('causing', 35, 24) ### +### ('forest', 14, 118) ('##₂', 36, 51) ('alfred', 12, 146) ('agricultural', 453, 7) ### +### ('rainforest', 51, 52) ('afraid', 92, 32) ('−', 83, 45) ('sharply', 95, 38) ('stab', 82, 53) ### +### ('hating', 187, 21) ### +############################################################################################################ +[2023-10-07 20:50:29,213][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:50:29,213][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:50:29,633][root][INFO] - Epoch: 5: Step: 301/1557, loss[v]=0.165034, lr=0.000016, acc@1[1]=235.5/256=0.919921875, acc@1[2]=243.5/256=0.951171875 +[2023-10-07 20:51:47,202][root][INFO] - Train batch 400 +[2023-10-07 20:51:47,203][root][INFO] - Avg. loss per last 100 batches: 0.093803 +[2023-10-07 20:51:47,887][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29274.9/29522=99.16% | mean: 0.01 | max: 4.88 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.18 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is enterprising [SEP] ### +### [P_TEXT]: [CLS] enterprising - marked by imagination, initiative, and readiness to undertake new ### +### projects ; an enterprising foreign policy ; an enterprising young man likely to go far. 1 ### +### adventuresome, adventurous - willing to undertake or seeking out new and daring enterprises ; ### +### adventurous pioneers ; the risks and gains of an adventuresome economy. [SEP] ### +### ======================================= h_v_q | Gates: 28045 ======================================= ### +### ('##ising', 0, 14) ('##pr', 1, 32) ('enter', 2, 3) ('##izing', 3, 106) ('entering', 4, 27) ### +### ('enters', 5, 29) ('entered', 6, 33) ('##isation', 7, 78) ('definition', 8, 10) ('is', 9, 501) ### +### ('encompasses', 10, 18) ('familiarity', 11, 26916) ('noun', 12, 19816) ('refers', 13, 5460) ### +### ('pr', 14, 178) ('.', 15, 7773) ('relating', 16, 12950) ('something', 17, 2353) ('exit', 18, 45) ### +### ('inside', 19, 479) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('adventurous', 16750, 0) ('imagination', 9320, 1) ('pioneers', 2530, 2) ('enter', 2, 3) ### +### ('initiative', 1000, 4) ('##ome', 21882, 5) ('daring', 4799, 6) ('define', 12645, 7) ### +### ('adventures', 14142, 8) ('projects', 13267, 9) ('definition', 8, 10) ('pioneer', 2018, 11) ### +### ('readiness', 1693, 12) ('far', 9653, 13) ('##ising', 0, 14) ('economy', 1097, 15) ### +### ('young', 5466, 16) ('man', 2948, 17) ('encompasses', 10, 18) ('enterprises', 2911, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('enter', 2, 3) ('##ising', 0, 14) ('##pr', 1, 32) ('entering', 4, 27) ('definition', 8, 10) ### +### ('enters', 5, 29) ('entered', 6, 33) ('encompasses', 10, 18) ('##izing', 3, 106) ### +### ('##isation', 7, 78) ('is', 9, 501) ('exit', 18, 45) ('entry', 25, 30) ('pr', 14, 178) ### +### ('escape', 20, 84) ('entrance', 30, 76) ('wingspan', 86, 44) ('seek', 55, 61) ('−', 132, 40) ### +### ('julian', 29, 144) ### +############################################################################################################ +[2023-10-07 20:51:47,888][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:51:47,888][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:51:48,310][root][INFO] - Epoch: 5: Step: 401/1557, loss[v]=0.088533, lr=0.000016, acc@1[1]=239.0/256=0.93359375, acc@1[2]=248.0/256=0.96875 +[2023-10-07 20:53:05,670][root][INFO] - Train batch 500 +[2023-10-07 20:53:05,671][root][INFO] - Avg. loss per last 100 batches: 0.087224 +[2023-10-07 20:53:06,397][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29185.8/29522=98.86% | mean: 0.01 | max: 5.14 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 5.97 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] when will the train refurbishment at disney world be done [SEP] ### +### [P_TEXT]: [CLS] the refurbishment begins with the closing of the walt disney world railroad on ### +### sept. 28, 2014, and it is scheduled to go through nov. 8, 2014. as of now, it is tentatively ### +### scheduled to reopen to guests on nov. 9, 2014, but please remember that dates are subject to ### +### change. [SEP] ### +### ======================================= h_v_q | Gates: 27512 ======================================= ### +### ('disney', 0, 1) ('refurbishment', 1, 5) ('train', 2, 18) ('world', 3, 59) ('done', 4, 2281) ### +### ('familiarity', 5, 21786) ('.', 6, 4855) ('railroad', 7, 0) ('renovation', 8, 15) ### +### ('september', 9, 104) ('sunday', 10, 7053) ('refurbished', 11, 16) ('date', 12, 52) ### +### ('april', 13, 709) ('doing', 14, 287) ('november', 15, 217) ('june', 16, 188) ### +### ('restoration', 17, 55) ('railway', 18, 7) ('december', 19, 436) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('railroad', 7, 0) ('disney', 0, 1) ('scheduled', 4783, 2) ('closing', 2250, 3) ('walt', 994, 4) ### +### ('refurbishment', 1, 5) ('nov', 18824, 6) ('railway', 18, 7) ('sept', 19062, 8) ### +### ('##open', 25025, 9) ('2014', 2142, 10) ('closed', 2305, 11) ('open', 2843, 12) ### +### ('railroads', 18993, 13) ('guests', 3482, 14) ('renovation', 8, 15) ('refurbished', 11, 16) ### +### ('railways', 2435, 17) ('train', 2, 18) ('dates', 2701, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('disney', 0, 1) ('refurbishment', 1, 5) ('train', 2, 18) ('world', 3, 59) ('railroad', 7, 0) ### +### ('renovation', 8, 15) ('railway', 18, 7) ('refurbished', 11, 16) ('date', 12, 52) ### +### ('september', 9, 104) ('restoration', 17, 55) ('when', 21, 82) ('rail', 42, 20) ('trains', 26, 90) ### +### ('june', 16, 188) ('november', 15, 217) ('doing', 14, 287) ('july', 32, 136) ### +### ('construction', 35, 122) ('−', 69, 38) ### +############################################################################################################ +[2023-10-07 20:53:06,398][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:53:06,398][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:53:06,803][root][INFO] - Epoch: 5: Step: 501/1557, loss[v]=0.063606, lr=0.000015, acc@1[1]=243.5/256=0.951171875, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 20:54:23,193][root][INFO] - Train batch 600 +[2023-10-07 20:54:23,193][root][INFO] - Avg. loss per last 100 batches: 0.089005 +[2023-10-07 20:54:23,899][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29292.8/29522=99.22% | mean: 0.01 | max: 5.38 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 5.85 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what kingdom is volvox classified in [SEP] ### +### [P_TEXT]: [CLS] description : volvox is a green algae and the culmination of the volvocine line. it ### +### is a hollow sphere made up of a single layer of 500 - 60, 000 biflagellate cells. the flagella beat ### +### so as to turn the volvox in circles and therefore propel the spheres in water. volvox is also ### +### polar. [SEP] ### +### ======================================= h_v_q | Gates: 28249 ======================================= ### +### ('volvo', 0, 0) ('classified', 1, 3558) ('kingdom', 2, 13331) ('##x', 3, 9) ### +### ('familiarity', 4, 26619) ('classification', 5, 3843) ('classify', 6, 9765) ('is', 7, 138) ### +### ('states', 8, 19559) ('relating', 9, 21134) ('division', 10, 2091) ('definition', 11, 71) ### +### ('.', 12, 6350) ('territory', 13, 14299) ('country', 14, 11765) ('julian', 15, 179) ### +### ('nation', 16, 8628) ('province', 17, 14593) ('alfred', 18, 116) ('wales', 19, 7480) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('volvo', 0, 0) ('algae', 16691, 1) ('sphere', 6382, 2) ('##cine', 26958, 3) ('circles', 4095, 4) ### +### ('spheres', 16747, 5) ('hollow', 261, 6) ('green', 353, 7) ('polar', 16367, 8) ('##x', 3, 9) ### +### ('flag', 2510, 10) ('##ella', 14479, 11) ('beat', 9458, 12) ('layers', 16276, 13) ('ˈ', 401, 14) ### +### ('encompasses', 22, 15) ('cells', 7879, 16) ('wingspan', 198, 17) ('line', 5272, 18) ### +### ('crashing', 253, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('volvo', 0, 0) ('##x', 3, 9) ('classified', 1, 3558) ('is', 7, 138) ('encompasses', 22, 15) ### +### ('definition', 11, 71) ('kingdom', 2, 13331) ('alfred', 18, 116) ('∈', 54, 30) ('hollow', 261, 6) ### +### ('julian', 15, 179) ('hating', 106, 21) ('green', 353, 7) ('wingspan', 198, 17) ('−', 88, 39) ### +### ('crashing', 253, 19) ('##₂', 136, 34) ('##大', 166, 28) ('fernando', 158, 31) ### +### ('classification', 5, 3843) ### +############################################################################################################ +[2023-10-07 20:54:23,899][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:54:23,899][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:54:24,320][root][INFO] - Epoch: 5: Step: 601/1557, loss[v]=0.086282, lr=0.000015, acc@1[1]=237.5/256=0.927734375, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 20:55:41,099][root][INFO] - Train batch 700 +[2023-10-07 20:55:41,100][root][INFO] - Avg. loss per last 100 batches: 0.087916 +[2023-10-07 20:55:41,795][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29334.6/29522=99.37% | mean: 0.01 | max: 5.12 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.11 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what provence is portes le valence [SEP] ### +### [P_TEXT]: [CLS] the city of portes - les - la¨s valence is a french city located south east of. ### +### france the city of - portes - les la¨s valence is located in the department of drome dra´me of the ### +### - french. region rhone rha´ne alpeshe city of portes - les - la¨s valence is located in the ### +### township of - portes - les la¨s valence part of the district. of valence [SEP] ### +### ======================================= h_v_q | Gates: 28356 ======================================= ### +### ('porte', 0, 0) ('provence', 1, 2089) ('vale', 2, 3) ('##nce', 3, 1) ('le', 4, 28) ### +### ('familiarity', 5, 23242) ('julian', 6, 42) ('##s', 7, 22) ('relating', 8, 16016) ('is', 9, 267) ### +### ('definition', 10, 3724) ('encompasses', 11, 31) ('noun', 12, 26840) ('##स', 13, 125) ### +### ('refers', 14, 23746) ('.', 15, 1184) ('##α', 16, 15) ('odd', 17, 52) ('emma', 18, 2335) ### +### ('##₂', 19, 44) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('porte', 0, 0) ('##nce', 3, 1) ('##¨', 23984, 2) ('vale', 2, 3) ('rhone', 5116, 4) ('les', 82, 5) ### +### ('la', 143, 6) ('france', 115, 7) ('french', 955, 8) ('dr', 8008, 9) ('crashing', 33, 10) ### +### ('city', 798, 11) ('department', 4211, 12) ('cities', 6471, 13) ('##me', 11199, 14) ('##α', 16, 15) ### +### ('##ome', 16114, 16) ('township', 9720, 17) ('##ο', 36, 18) ('ˈ', 84, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('porte', 0, 0) ('##nce', 3, 1) ('vale', 2, 3) ('le', 4, 28) ('julian', 6, 42) ('##s', 7, 22) ### +### ('provence', 1, 2089) ('encompasses', 11, 31) ('##α', 16, 15) ('crashing', 33, 10) ('les', 82, 5) ### +### ('##ο', 36, 18) ('france', 115, 7) ('is', 9, 267) ('la', 143, 6) ('##₂', 19, 44) ('odd', 17, 52) ### +### ('##ང', 45, 24) ('cyrillic', 29, 51) ('##स', 13, 125) ### +############################################################################################################ +[2023-10-07 20:55:41,795][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:55:41,795][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:55:42,218][root][INFO] - Epoch: 5: Step: 701/1557, loss[v]=0.132252, lr=0.000015, acc@1[1]=242.5/256=0.947265625, acc@1[2]=248.0/256=0.96875 +[2023-10-07 20:56:58,655][root][INFO] - Train batch 800 +[2023-10-07 20:56:58,657][root][INFO] - Avg. loss per last 100 batches: 0.092173 +[2023-10-07 20:56:59,344][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29297.3/29522=99.24% | mean: 0.01 | max: 5.29 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.17 | max: 5.93 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how were chinampas constructed [SEP] ### +### [P_TEXT]: [CLS] how were aztec chinampas constructed? ( 5 points ) raised plots for crops were ### +### built on dry land with irrigation channels between them raised plots for crops were built from mud ### +### in the lake bed with canals between them raised plots for crops were built on dry land with canals ### +### between them raised plots for crops were built from depositing dry earth into the river bed with ### +### canals between them [SEP] ### +### ======================================= h_v_q | Gates: 28052 ======================================= ### +### ('china', 0, 5) ('constructed', 1, 8) ('##as', 2, 17) ('##mp', 3, 22) ('were', 4, 125) ### +### ('built', 5, 7) ('construction', 6, 25) ('somehow', 7, 100) ('familiarity', 8, 25779) ### +### ('chinese', 9, 291) ('designed', 10, 1647) ('relating', 11, 9785) ('founded', 12, 124) ### +### ('eager', 13, 1433) ('created', 14, 156) ('manufactured', 15, 24) ('.', 16, 4329) ('made', 17, 123) ### +### ('jeremy', 18, 179) ('constructing', 19, 42) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('aztec', 13725, 0) ('crops', 1432, 1) ('irrigation', 762, 2) ('plots', 21861, 3) ### +### ('plot', 10010, 4) ('china', 0, 5) ('points', 7803, 6) ('built', 5, 7) ('constructed', 1, 8) ### +### ('crop', 9628, 9) ('canals', 23481, 10) ('dry', 4627, 11) ('raised', 993, 12) ('ˈ', 651, 13) ### +### ('canal', 5936, 14) ('mud', 1274, 15) ('land', 1142, 16) ('##as', 2, 17) ('wingspan', 139, 18) ### +### ('channel', 996, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('china', 0, 5) ('constructed', 1, 8) ('##as', 2, 17) ('##mp', 3, 22) ('built', 5, 7) ### +### ('construction', 6, 25) ('were', 4, 125) ('somehow', 7, 100) ('manufactured', 15, 24) ### +### ('chinese', 9, 291) ('founded', 12, 124) ('constructing', 19, 42) ('made', 17, 123) ### +### ('created', 14, 156) ('building', 22, 61) ('rebuilt', 30, 30) ('build', 32, 39) ('jeremy', 18, 179) ### +### ('##₂', 43, 35) ('construct', 39, 50) ### +############################################################################################################ +[2023-10-07 20:56:59,344][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:56:59,344][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:56:59,769][root][INFO] - Epoch: 5: Step: 801/1557, loss[v]=0.072606, lr=0.000015, acc@1[1]=237.0/256=0.92578125, acc@1[2]=247.0/256=0.96484375 +[2023-10-07 20:58:16,012][root][INFO] - Train batch 900 +[2023-10-07 20:58:16,012][root][INFO] - Avg. loss per last 100 batches: 0.092408 +[2023-10-07 20:58:16,746][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29186.0/29522=98.86% | mean: 0.01 | max: 4.95 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.14 | max: 5.59 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what does ccim stand for [SEP] ### +### [P_TEXT]: [CLS] accima refers to a professional designation offered by the ccim institute. the ccim ### +### institute is a highly regarded commercial real estate and investment property organization based in ### +### the united states. some consider the ccim designation to be roughly equivalent to an amda of ### +### commercial real estate. ccimas are recognized as being experts in the commercial and investment ### +### real estate industry. ccim stands for acertified commercial investment membera. the process of ### +### achieving the designation is a combination of theory / coursework, commercial real estate ### +### experience and examinations. [SEP] ### +### ======================================= h_v_q | Gates: 28116 ======================================= ### +### ('##im', 0, 5) ('cc', 1, 0) ('stands', 2, 48) ('stand', 3, 36) ('noun', 4, 18228) ### +### ('refers', 5, 192) ('familiarity', 6, 22068) ('.', 7, 3963) ('relating', 8, 3855) ('stood', 9, 172) ### +### ('definition', 10, 125) ('is', 11, 267) ('something', 12, 2410) ('or', 13, 12671) ('##ib', 14, 845) ### +### ('##em', 15, 1974) (';', 16, 2181) ('loosely', 17, 336) ('sam', 18, 217) ('sense', 19, 6888) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('cc', 1, 0) ('acc', 17290, 1) ('designation', 956, 2) ('commercial', 773, 3) ('institute', 205, 4) ### +### ('##im', 0, 5) ('##ima', 1731, 6) ('##fied', 4041, 7) ('##da', 8179, 8) ('investment', 5066, 9) ### +### ('offered', 5494, 10) ('encompasses', 31, 11) ('professional', 330, 12) ('designations', 7744, 13) ### +### ('highly', 3075, 14) ('industry', 2018, 15) ('estate', 663, 16) ('crashing', 218, 17) ### +### ('equivalent', 7497, 18) ('recognized', 2867, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('cc', 1, 0) ('##im', 0, 5) ('stands', 2, 48) ('stand', 3, 36) ('refers', 5, 192) ### +### ('encompasses', 31, 11) ('stood', 9, 172) ('definition', 10, 125) ('−', 37, 30) ### +### ('institute', 205, 4) ('##₂', 44, 32) ('is', 11, 267) ('meaning', 22, 93) ('julian', 20, 139) ### +### ('crashing', 218, 17) ('professional', 330, 12) ('hating', 64, 50) ('sam', 18, 217) ### +### ('commercial', 773, 3) ('##ο', 174, 33) ### +############################################################################################################ +[2023-10-07 20:58:16,746][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:58:16,746][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:58:17,176][root][INFO] - Epoch: 5: Step: 901/1557, loss[v]=0.081474, lr=0.000015, acc@1[1]=238.0/256=0.9296875, acc@1[2]=248.5/256=0.970703125 +[2023-10-07 20:59:33,053][root][INFO] - Train batch 1000 +[2023-10-07 20:59:33,053][root][INFO] - Avg. loss per last 100 batches: 0.092803 +[2023-10-07 20:59:33,735][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29251.4/29522=99.08% | mean: 0.01 | max: 5.27 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 5.93 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] meaning of nonage [SEP] ### +### [P_TEXT]: [CLS] webster dictionary ( 0. 00 / 0 votes ) rate this definition : nonage ( noun ) the ### +### ninth part of movable goods, formerly payable to the clergy on the death of persons in their ### +### parishes. nonage ( noun ) time of life before a person becomes of age ; legal immaturity ; ### +### minority. [SEP] ### +### ======================================= h_v_q | Gates: 27099 ======================================= ### +### ('non', 0, 2) ('##age', 1, 0) ('noun', 2, 674) ('definition', 3, 5) ('meaning', 4, 6) ### +### ('##ages', 5, 19) ('familiarity', 6, 24962) ('means', 7, 162) ('relating', 8, 9700) ### +### ('defined', 9, 58) ('julian', 10, 87) ('semi', 11, 119) ('term', 12, 771) ('sense', 13, 4696) ### +### ('refers', 14, 5331) ('lack', 15, 107) ('latin', 16, 705) ('language', 17, 7348) ('.', 18, 5166) ### +### ('or', 19, 15584) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##age', 1, 0) ('webster', 9169, 1) ('non', 0, 2) ('##vable', 23154, 3) ('define', 4573, 4) ### +### ('definition', 3, 5) ('meaning', 4, 6) ('definitions', 72, 7) ('crashing', 71, 8) ### +### ('parishes', 23110, 9) ('mo', 2788, 10) ('clergy', 8170, 11) ('ˈ', 542, 12) ('ninth', 7164, 13) ### +### ('wingspan', 143, 14) ('minority', 2448, 15) ('##ང', 356, 16) ('##ο', 178, 17) ### +### ('encompasses', 180, 18) ('##ages', 5, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('non', 0, 2) ('##age', 1, 0) ('definition', 3, 5) ('meaning', 4, 6) ('##ages', 5, 19) ### +### ('noun', 2, 674) ('defined', 9, 58) ('julian', 10, 87) ('means', 7, 162) ('mean', 23, 29) ### +### ('semi', 11, 119) ('lack', 15, 107) ('definitions', 72, 7) ('##₂', 35, 33) ('crashing', 71, 8) ### +### ('hating', 36, 61) ('never', 34, 62) ('crashed', 63, 24) ('−', 66, 22) ('ignoring', 38, 68) ### +############################################################################################################ +[2023-10-07 20:59:33,736][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 20:59:33,736][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 20:59:34,156][root][INFO] - Epoch: 5: Step: 1001/1557, loss[v]=0.124855, lr=0.000015, acc@1[1]=241.0/256=0.94140625, acc@1[2]=245.0/256=0.95703125 +[2023-10-07 21:00:52,306][root][INFO] - Train batch 1100 +[2023-10-07 21:00:52,307][root][INFO] - Avg. loss per last 100 batches: 0.088618 +[2023-10-07 21:00:53,010][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29205.9/29522=98.93% | mean: 0.01 | max: 4.68 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 5.76 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what two revolutions occurred in the northeast region? [SEP] ### +### [P_TEXT]: [CLS] answer to i will upvote what two revolutions occurred in the northeast region? ### +### industrial revolution and mexican revolution french and indian revolution and revolutionary war ### +### revolutionary war and industrial revolution log in with facebook or [SEP] ### +### ======================================= h_v_q | Gates: 27651 ======================================= ### +### ('revolutions', 0, 3) ('northeast', 1, 0) ('two', 2, 24) ('region', 3, 8) ('revolution', 4, 4) ### +### ('twin', 5, 45) ('northwest', 6, 10) ('familiarity', 7, 18016) ('happened', 8, 44) ### +### ('began', 9, 5427) ('southeast', 10, 41) ('occurred', 11, 48) ('ended', 12, 177) ('.', 13, 5862) ### +### ('relating', 14, 7477) ('northern', 15, 183) ('died', 16, 493) ('caused', 17, 214) ### +### ('three', 18, 81) ('north', 19, 126) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('northeast', 1, 0) ('revolutionary', 295, 1) ('mexican', 3375, 2) ('revolutions', 0, 3) ### +### ('revolution', 4, 4) ('industrial', 972, 5) ('crashing', 61, 6) ('ˈ', 90, 7) ('region', 3, 8) ### +### ('french', 1325, 9) ('northwest', 6, 10) ('wingspan', 855, 11) ('indian', 925, 12) ### +### ('mexico', 526, 13) ('##ο', 426, 14) ('crashed', 63, 15) ('cyrillic', 310, 16) ('##ང', 797, 17) ### +### ('−', 160, 18) ('hating', 70, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('revolutions', 0, 3) ('northeast', 1, 0) ('region', 3, 8) ('revolution', 4, 4) ('two', 2, 24) ### +### ('twin', 5, 45) ('northwest', 6, 10) ('happened', 8, 44) ('southeast', 10, 41) ('occurred', 11, 48) ### +### ('regions', 20, 31) ('crashing', 61, 6) ('three', 18, 81) ('##₂', 35, 33) ('ˈ', 90, 7) ### +### ('crashed', 63, 15) ('ended', 12, 177) ('twins', 22, 96) ('north', 19, 126) ('hating', 70, 19) ### +############################################################################################################ +[2023-10-07 21:00:53,010][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:00:53,010][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:00:53,415][root][INFO] - Epoch: 5: Step: 1101/1557, loss[v]=0.075806, lr=0.000015, acc@1[1]=245.0/256=0.95703125, acc@1[2]=251.5/256=0.982421875 +[2023-10-07 21:02:10,884][root][INFO] - Train batch 1200 +[2023-10-07 21:02:10,885][root][INFO] - Avg. loss per last 100 batches: 0.088579 +[2023-10-07 21:02:11,593][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29251.1/29522=99.08% | mean: 0.01 | max: 5.36 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 5.90 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what causes young bean plants to rot at the root system? [SEP] ### +### [P_TEXT]: [CLS] beans may exhibit seed rot, damping - off, and root rot symptoms due to many other ### +### causes. sclerotinia sclerotiorum which causes sclerotinia wilt or white mold also causes a stem rot ### +### under certain conditions. the disease frequently occurs after a period of warm, humid weather. ### +### [SEP] ### +### ======================================= h_v_q | Gates: 27414 ======================================= ### +### ('rot', 0, 1) ('bean', 1, 26) ('root', 2, 12) ('young', 3, 10874) ('plants', 4, 700) ### +### ('system', 5, 20093) ('caused', 6, 244) ('causes', 7, 15) ('cause', 8, 95) ('.', 9, 13713) ### +### ('familiarity', 10, 25331) ('plant', 11, 1285) ('at', 12, 24800) ('teenage', 13, 1242) ### +### ('beans', 14, 0) ('relating', 15, 7546) ('systems', 16, 7906) ('julian', 17, 262) ### +### ('network', 18, 16714) ('to', 19, 24997) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('beans', 14, 0) ('rot', 0, 1) ('stem', 2399, 2) ('mold', 3585, 3) ('##inia', 28204, 4) ### +### ('white', 1087, 5) ('symptoms', 1387, 6) ('sc', 7976, 7) ('wingspan', 668, 8) ('stems', 5027, 9) ### +### ('ˈ', 166, 10) ('weather', 4915, 11) ('root', 2, 12) ('crashing', 28, 13) ('seed', 2388, 14) ### +### ('causes', 7, 15) ('wil', 10196, 16) ('damp', 2948, 17) ('humid', 15420, 18) ('warm', 2661, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('rot', 0, 1) ('bean', 1, 26) ('root', 2, 12) ('causes', 7, 15) ('beans', 14, 0) ('caused', 6, 244) ### +### ('cause', 8, 95) ('plants', 4, 700) ('crashing', 28, 13) ('ruins', 20, 28) ('##₂', 21, 36) ### +### ('roots', 27, 39) ('##大', 35, 30) ('young', 3, 10874) ('−', 42, 25) ('hating', 47, 23) ### +### ('anger', 29, 58) ('altogether', 24, 72) ('hugh', 40, 35) ('presenter', 30, 60) ### +############################################################################################################ +[2023-10-07 21:02:11,593][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:02:11,593][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:02:12,016][root][INFO] - Epoch: 5: Step: 1201/1557, loss[v]=0.109285, lr=0.000015, acc@1[1]=239.5/256=0.935546875, acc@1[2]=246.0/256=0.9609375 +[2023-10-07 21:03:29,562][root][INFO] - Train batch 1300 +[2023-10-07 21:03:29,563][root][INFO] - Avg. loss per last 100 batches: 0.093406 +[2023-10-07 21:03:30,290][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29181.9/29522=98.85% | mean: 0.01 | max: 5.28 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.13 | max: 5.92 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how long should it take to get federal refund when i file electronically? [SEP] ### +### [P_TEXT]: [CLS] filing a return through the mail is, of course, the slower of the two processes. ### +### for those who mail their federal return, the irs will remit a tax refund within six weeks after ### +### they have been received the completed return. if you file electronically, you should receive your ### +### tax refund within three weeks. the irs allows two options for receiving a federal tax refund : ### +### paper check sent via the mail, or an electronic funds transfer ( also called a direct deposit ). ### +### the direct deposit method, not surprisingly, is always the quicker option. [SEP] ### +### ======================================= h_v_q | Gates: 27635 ======================================= ### +### ('ref', 0, 3) ('federal', 1, 1) ('file', 2, 4) ('electronically', 3, 44) ('##und', 4, 23) ### +### ('electronic', 5, 18) ('should', 6, 105) ('minutes', 7, 123) ('get', 8, 244) ('weeks', 9, 8) ### +### ('.', 10, 7888) ('computer', 11, 3126) ('got', 12, 1552) ('30', 13, 466) ('take', 14, 5658) ### +### ('national', 15, 295) ('familiarity', 16, 18625) ('days', 17, 398) ('gets', 18, 169) ### +### ('months', 19, 1253) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('irs', 25086, 0) ('federal', 1, 1) ('return', 1771, 2) ('ref', 0, 3) ('file', 2, 4) ### +### ('paper', 805, 5) ('via', 4891, 6) ('mail', 3430, 7) ('weeks', 9, 8) ('check', 1550, 9) ### +### ('tax', 248, 10) ('filing', 79, 11) ('receiving', 1470, 12) ('ˈ', 404, 13) ('long', 23, 14) ### +### ('transfer', 1505, 15) ('receive', 131, 16) ('through', 1534, 17) ('electronic', 5, 18) ### +### ('crashing', 144, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ref', 0, 3) ('federal', 1, 1) ('file', 2, 4) ('##und', 4, 23) ('electronic', 5, 18) ### +### ('electronically', 3, 44) ('weeks', 9, 8) ('should', 6, 105) ('minutes', 7, 123) ('long', 23, 14) ### +### ('get', 8, 244) ('files', 25, 22) ('digital', 22, 110) ('gets', 18, 169) ('when', 24, 119) ### +### ('national', 15, 295) ('30', 13, 466) ('filing', 79, 11) ('getting', 20, 309) ('days', 17, 398) ### +############################################################################################################ +[2023-10-07 21:03:30,291][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:03:30,291][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:03:30,717][root][INFO] - Epoch: 5: Step: 1301/1557, loss[v]=0.052468, lr=0.000015, acc@1[1]=244.0/256=0.953125, acc@1[2]=252.0/256=0.984375 +[2023-10-07 21:04:48,744][root][INFO] - Train batch 1400 +[2023-10-07 21:04:48,744][root][INFO] - Avg. loss per last 100 batches: 0.089262 +[2023-10-07 21:04:49,474][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29138.7/29522=98.70% | mean: 0.01 | max: 4.97 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.13 | max: 5.75 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a cleanroom [SEP] ### +### [P_TEXT]: [CLS] a cleanroom is an area or enclosure in which the air has to be cleaned to a defined ### +### particle limit, usually in particles per cubic foot. in the united states most cleanrooms are ### +### discussed in terms of classification limits. [SEP] ### +### ======================================= h_v_q | Gates: 26903 ======================================= ### +### ('##room', 0, 0) ('clean', 1, 1) ('encompasses', 2, 8) ('definition', 3, 14) ('is', 4, 320) ### +### ('refers', 5, 2918) ('familiarity', 6, 25489) ('noun', 7, 16373) ('.', 8, 10744) ### +### ('cleaning', 9, 23) ('room', 10, 12) ('a', 11, 837) ('##rooms', 12, 6) ('relating', 13, 12860) ### +### ('defined', 14, 31) ('term', 15, 297) ('julian', 16, 140) ('something', 17, 2935) ('simon', 18, 60) ### +### ('raw', 19, 1601) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##room', 0, 0) ('clean', 1, 1) ('enclosure', 5262, 2) ('particles', 3604, 3) ### +### ('classification', 2261, 4) ('cubic', 12107, 5) ('##rooms', 12, 6) ('limits', 5125, 7) ### +### ('encompasses', 2, 8) ('particle', 8186, 9) ('ˈ', 299, 10) ('crashing', 29, 11) ('room', 10, 12) ### +### ('foot', 5610, 13) ('definition', 3, 14) ('define', 5752, 15) ('limit', 6338, 16) ('air', 5021, 17) ### +### ('feet', 7032, 18) ('rooms', 312, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##room', 0, 0) ('clean', 1, 1) ('encompasses', 2, 8) ('definition', 3, 14) ('##rooms', 12, 6) ### +### ('room', 10, 12) ('cleaning', 9, 23) ('defined', 14, 31) ('is', 4, 320) ('crashing', 29, 11) ### +### ('simon', 18, 60) ('julian', 16, 140) ('hating', 28, 32) ('−', 32, 28) ('cleaned', 61, 22) ### +### ('meaning', 62, 21) ('term', 15, 297) ('anger', 31, 51) ('refers', 5, 2918) ('ruins', 76, 35) ### +############################################################################################################ +[2023-10-07 21:04:49,474][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:04:49,474][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:04:49,879][root][INFO] - Epoch: 5: Step: 1401/1557, loss[v]=0.107737, lr=0.000015, acc@1[1]=243.0/256=0.94921875, acc@1[2]=247.5/256=0.966796875 +[2023-10-07 21:06:08,072][root][INFO] - Train batch 1500 +[2023-10-07 21:06:08,073][root][INFO] - Avg. loss per last 100 batches: 0.088872 +[2023-10-07 21:06:08,779][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29166.3/29522=98.80% | mean: 0.01 | max: 5.19 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.05 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what does symbolic language do in a text [SEP] ### +### [P_TEXT]: [CLS] symbolic language. formalized ( artificial ) language ( such as mathematics ) which ### +### uses symbols with specific meanings, in order to avoid ambiguities and inadequacies of natural ### +### languages such as english. ymbolic language. formalized ( artificial ) language ( such as ### +### mathematics ) which uses symbols with specific meanings, in order to avoid ambiguities and ### +### inadequacies of natural languages such as english. [SEP] ### +### ======================================= h_v_q | Gates: 27921 ======================================= ### +### ('symbolic', 0, 1) ('language', 1, 3) ('text', 2, 3763) ('do', 3, 8092) ('doing', 4, 821) ### +### ('.', 5, 5191) ('languages', 6, 6) ('familiarity', 7, 23450) ('ceremonial', 8, 71) ### +### ('symbol', 9, 11) ('something', 10, 874) ('relating', 11, 9079) ('brazil', 12, 1676) ### +### ('did', 13, 10652) ('is', 14, 502) ('a', 15, 9220) ('julian', 16, 95) ('ト', 17, 299) ### +### ('emotional', 18, 923) ('ability', 19, 2219) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('artificial', 268, 0) ('symbolic', 0, 1) ('formal', 252, 2) ('language', 1, 3) ('symbols', 33, 4) ### +### ('##ized', 7869, 5) ('languages', 6, 6) ('y', 7835, 7) ('define', 18762, 8) ('uses', 1544, 9) ### +### ('avoid', 5284, 10) ('symbol', 9, 11) ('definition', 109, 12) ('##mbo', 23038, 13) ### +### ('natural', 1889, 14) ('##qua', 17898, 15) ('ˈ', 847, 16) ('meanings', 19013, 17) ### +### ('crashing', 108, 18) ('encompasses', 36, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('symbolic', 0, 1) ('language', 1, 3) ('languages', 6, 6) ('symbol', 9, 11) ('ceremonial', 8, 71) ### +### ('symbols', 33, 4) ('text', 2, 3763) ('encompasses', 36, 19) ('artificial', 268, 0) ### +### ('formal', 252, 2) ('grammar', 22, 61) ('julian', 16, 95) ('doing', 4, 821) ('definition', 109, 12) ### +### ('crashing', 108, 18) ('dialect', 79, 27) ('−', 128, 22) ('##₂', 88, 30) ('mathematics', 162, 21) ### +### ('words', 57, 55) ### +############################################################################################################ +[2023-10-07 21:06:08,779][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:06:08,779][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:06:09,203][root][INFO] - Epoch: 5: Step: 1501/1557, loss[v]=0.083446, lr=0.000015, acc@1[1]=243.5/256=0.951171875, acc@1[2]=252.0/256=0.984375 +[2023-10-07 21:06:53,126][root][INFO] - rank=0; last iteration 1557 +[2023-10-07 21:06:53,127][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 21:06:53,127][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-07 21:06:53,128][root][INFO] - rank=1; last iteration 1557 +[2023-10-07 21:06:53,129][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 21:06:53,129][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-07 21:06:53,129][root][INFO] - rank=2; last iteration 1557 +[2023-10-07 21:06:53,129][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 21:06:53,129][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-07 21:06:53,134][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 21:06:53,135][root][INFO] - Epoch finished on 0 +[2023-10-07 21:06:53,136][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 21:06:53,137][root][INFO] - Epoch finished on 1 +[2023-10-07 21:06:53,138][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 21:06:53,138][root][INFO] - Epoch finished on 2 +[2023-10-07 21:06:53,152][root][INFO] - rank=3; last iteration 1557 +[2023-10-07 21:06:53,153][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 21:06:53,153][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-07 21:06:53,161][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 21:06:53,161][root][INFO] - Epoch finished on 3 +[2023-10-07 21:07:32,719][root][INFO] - Saved checkpoint at ./vdr_5 +[2023-10-07 21:07:32,720][root][INFO] - Av Loss per epoch=0.090429 +[2023-10-07 21:07:32,720][root][INFO] - epoch total (1) correct predictions=375089 +[2023-10-07 21:07:32,720][root][INFO] - epoch total (2) correct predictions=387895 +[2023-10-07 21:07:32,720][root][INFO] - Saved checkpoint at ./vdr_5 +[2023-10-07 21:07:32,721][root][INFO] - Av Loss per epoch=0.090429 +[2023-10-07 21:07:32,720][root][INFO] - Saved checkpoint at ./vdr_5 +[2023-10-07 21:07:32,721][root][INFO] - epoch total (1) correct predictions=375089 +[2023-10-07 21:07:32,722][root][INFO] - epoch total (2) correct predictions=387895 +[2023-10-07 21:07:32,722][root][INFO] - Av Loss per epoch=0.090429 +[2023-10-07 21:07:32,721][root][INFO] - Saved checkpoint at ./vdr_5 +[2023-10-07 21:07:32,722][root][INFO] - epoch total (1) correct predictions=375089 +[2023-10-07 21:07:32,722][root][INFO] - Av Loss per epoch=0.090429 +[2023-10-07 21:07:32,722][root][INFO] - epoch total (2) correct predictions=387895 +[2023-10-07 21:07:32,722][root][INFO] - epoch total (1) correct predictions=375089 +[2023-10-07 21:07:32,722][root][INFO] - epoch total (2) correct predictions=387895 +[2023-10-07 21:07:32,724][root][INFO] - ***** Epoch 6 ***** +[2023-10-07 21:07:32,726][root][INFO] - ***** Epoch 6 ***** +[2023-10-07 21:07:32,728][root][INFO] - ***** Epoch 6 ***** +[2023-10-07 21:07:32,730][root][INFO] - rank=0; Iteration start +[2023-10-07 21:07:32,731][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 21:07:32,731][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 21:07:32,729][root][INFO] - ***** Epoch 6 ***** +[2023-10-07 21:07:32,733][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-07 21:07:32,733][root][INFO] - rank=3; Iteration start +[2023-10-07 21:07:32,733][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 21:07:32,733][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 21:07:32,735][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-07 21:07:32,736][root][INFO] - rank=2; Iteration start +[2023-10-07 21:07:32,736][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 21:07:32,737][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 21:07:32,737][root][INFO] - rank=1; Iteration start +[2023-10-07 21:07:32,737][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 21:07:32,738][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 21:07:32,739][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-07 21:07:32,739][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-07 21:07:33,737][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29189.8/29522=98.87% | mean: 0.01 | max: 4.93 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.4/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.17 | max: 5.91 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is physalis [SEP] ### +### [P_TEXT]: [CLS] physalis ( / efaeªseleªs /, sometimes / faeªeseeªleªs /, from physalis = bladder ) ### +### is a genus of flowering plants in the nightshade family ( solanaceae ), which grow in warm ### +### temperate and subtropical regions of the world. most of the species, of which there may be 75 - 90, ### +### are indigenous to the new world. ot all physalis species bear edible fruit. select species are ### +### cultivated for their edible fruit, however ; the typical physalis fruit is similar to a firm tomato ### +### in texture, and like strawberries or pineapple in flavor, with a mild acidity. [SEP] ### +### ======================================= h_v_q | Gates: 27293 ======================================= ### +### ('##alis', 0, 10) ('##ys', 1, 19) ('ph', 2, 6) ('is', 3, 172) ('definition', 4, 51) ### +### ('familiarity', 5, 26388) ('noun', 6, 11889) ('encompasses', 7, 8) ('refers', 8, 2082) ### +### ('julian', 9, 94) ('term', 10, 15106) ('relating', 11, 16614) ('something', 12, 8221) ### +### ('altogether', 13, 74) ('##ale', 14, 1009) ('.', 15, 11177) ('##al', 16, 431) ('##y', 17, 1427) ### +### ('stands', 18, 3203) ('defined', 19, 3226) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('bladder', 26418, 0) ('##ceae', 2025, 1) ('##ª', 27237, 2) ('flowering', 461, 3) ### +### ('##fa', 14612, 4) ('##had', 23639, 5) ('ph', 2, 6) ('fae', 17316, 7) ('encompasses', 7, 8) ### +### ('genus', 190, 9) ('##alis', 0, 10) ('##ese', 5008, 11) ('sol', 4167, 12) ### +### ('subtropical', 10003, 13) ('fruit', 8168, 14) ('##sel', 18345, 15) ('nights', 9829, 16) ### +### ('warm', 2355, 17) ('indigenous', 2498, 18) ('##ys', 1, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##alis', 0, 10) ('ph', 2, 6) ('##ys', 1, 19) ('encompasses', 7, 8) ('definition', 4, 51) ### +### ('is', 3, 172) ('julian', 9, 94) ('altogether', 13, 74) ('crashing', 21, 29) ('##₂', 26, 45) ### +### ('−', 40, 41) ('genus', 190, 9) ('wingspan', 53, 37) ('ruins', 32, 76) ('flowering', 461, 3) ### +### ('##大', 50, 55) ('crashed', 52, 52) ('##al', 16, 431) ('refers', 8, 2082) ('ˈ', 141, 27) ### +############################################################################################################ +[2023-10-07 21:07:33,737][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:07:33,737][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:07:34,134][root][INFO] - Epoch: 6: Step: 1/1557, loss[v]=0.110724, lr=0.000015, acc@1[1]=239.5/256=0.935546875, acc@1[2]=246.0/256=0.9609375 +[2023-10-07 21:08:50,720][root][INFO] - Train batch 100 +[2023-10-07 21:08:50,721][root][INFO] - Avg. loss per last 100 batches: 0.085258 +[2023-10-07 21:08:51,413][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29211.4/29522=98.95% | mean: 0.01 | max: 5.42 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 6.09 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] is reckless driving a criminal misdemeanor in washington [SEP] ### +### [P_TEXT]: [CLS] criminal defense : reckless driving. reckless driving is a gross misdemeanor crime ### +### in washington state, it carries a maximum penalty of 365 days in jail and a $ 5, 000 fine. the ### +### elements of a reckless driving charges involve the driving of a vehicle with the willful and wanton ### +### disregard for the safety of persons or property. a person who is convicted of reckless driving in ### +### the state of washington will likely have their driveras license suspended for 30 days by the ### +### department of licensing. if such a suspension is imposed, the driver will need to purchase sr - 22 ### +### insurance ( high risk driving insurance ) for a three year period. [SEP] ### +### ======================================= h_v_q | Gates: 28063 ======================================= ### +### ('reckless', 0, 0) ('washington', 1, 1) ('mis', 2, 36) ('##de', 3, 73) ('driving', 4, 10) ### +### ('criminal', 5, 5) ('##me', 6, 366) ('familiarity', 7, 26464) ('altogether', 8, 71) ### +### ('julian', 9, 112) ('careless', 10, 45) ('relating', 11, 9983) ('is', 12, 384) ### +### ('california', 13, 3733) ('wild', 14, 110) ('##anor', 15, 4588) ('.', 16, 10763) ('drive', 17, 70) ### +### ('selfish', 18, 293) ('stupid', 19, 78) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('reckless', 0, 0) ('washington', 1, 1) ('sr', 9965, 2) ('suspension', 3193, 3) ### +### ('penalty', 3065, 4) ('criminal', 5, 5) ('suspended', 9540, 6) ('jail', 10580, 7) ('ˈ', 344, 8) ### +### ('fine', 1866, 9) ('driving', 4, 10) ('encompasses', 72, 11) ('crime', 22, 12) ### +### ('convicted', 3942, 13) ('vehicle', 208, 14) ('fined', 20517, 15) ('−', 55, 16) ('driver', 99, 17) ### +### ('gross', 473, 18) ('fines', 12831, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('reckless', 0, 0) ('washington', 1, 1) ('driving', 4, 10) ('criminal', 5, 5) ('mis', 2, 36) ### +### ('##de', 3, 73) ('##me', 6, 366) ('careless', 10, 45) ('altogether', 8, 71) ('julian', 9, 112) ### +### ('crime', 22, 12) ('drive', 17, 70) ('wild', 14, 110) ('stupid', 19, 78) ('crashing', 37, 26) ### +### ('encompasses', 72, 11) ('##₂', 36, 30) ('−', 55, 16) ('is', 12, 384) ('foolish', 25, 116) ### +############################################################################################################ +[2023-10-07 21:08:51,413][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:08:51,413][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:08:51,836][root][INFO] - Epoch: 6: Step: 101/1557, loss[v]=0.074719, lr=0.000015, acc@1[1]=243.5/256=0.951171875, acc@1[2]=247.5/256=0.966796875 +[2023-10-07 21:10:08,911][root][INFO] - Train batch 200 +[2023-10-07 21:10:08,912][root][INFO] - Avg. loss per last 100 batches: 0.088674 +[2023-10-07 21:10:09,642][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29226.4/29522=99.00% | mean: 0.01 | max: 4.84 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 5.95 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] who is st bernadette [SEP] ### +### [P_TEXT]: [CLS] bernadette soubirous biography. bernadette of lourdes is the french saint whose ### +### body is said to have been miraculously preserved. in 1858, 14 - year - old bernadette soubirous had ### +### a vision of the virgin mary in a grotto near lourdes. [SEP] ### +### ======================================= h_v_q | Gates: 28320 ======================================= ### +### ('bern', 0, 0) ('##tte', 1, 10) ('st', 2, 80) ('##ade', 3, 22) ('familiarity', 4, 25125) ### +### ('who', 5, 39) ('is', 6, 333) ('saint', 7, 4) ('altogether', 8, 93) ('whose', 9, 44) ### +### ('relating', 10, 14506) ('julian', 11, 120) ('identity', 12, 2807) ('was', 13, 680) ### +### ('caroline', 14, 102) ('simon', 15, 166) ('august', 16, 1962) ('born', 17, 15361) ('.', 18, 8465) ### +### ('encompasses', 19, 15) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('bern', 0, 0) ('miraculous', 21222, 1) ('##rdes', 27900, 2) ('preserved', 5939, 3) ('saint', 7, 4) ### +### ('vision', 1849, 5) ('lou', 9388, 6) ('french', 3486, 7) ('miracles', 11054, 8) ('virgin', 971, 9) ### +### ('##tte', 1, 10) ('old', 3355, 11) ('mary', 558, 12) ('##otto', 28713, 13) ('biography', 8548, 14) ### +### ('encompasses', 19, 15) ('##iro', 25573, 16) ('ˈ', 54, 17) ('body', 1575, 18) ('france', 355, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('bern', 0, 0) ('##tte', 1, 10) ('##ade', 3, 22) ('st', 2, 80) ('who', 5, 39) ('saint', 7, 4) ### +### ('whose', 9, 44) ('encompasses', 19, 15) ('altogether', 8, 93) ('crashing', 21, 29) ### +### ('julian', 11, 120) ('is', 6, 333) ('caroline', 14, 102) ('##₂', 25, 37) ('ˈ', 54, 17) ### +### ('##ttes', 38, 27) ('−', 45, 28) ('##ང', 50, 23) ('##大', 44, 41) ('##α', 57, 33) ### +############################################################################################################ +[2023-10-07 21:10:09,642][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:10:09,642][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:10:10,061][root][INFO] - Epoch: 6: Step: 201/1557, loss[v]=0.059807, lr=0.000015, acc@1[1]=241.0/256=0.94140625, acc@1[2]=254.0/256=0.9921875 +[2023-10-07 21:11:26,259][root][INFO] - Train batch 300 +[2023-10-07 21:11:26,260][root][INFO] - Avg. loss per last 100 batches: 0.087153 +[2023-10-07 21:11:26,982][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29223.0/29522=98.99% | mean: 0.01 | max: 5.12 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.13 | max: 5.96 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what was aunt bee's last name [SEP] ### +### [P_TEXT]: [CLS] is aunt bee from andy griffith show? yes, beatrice aunt bee taylor ( played by the ### +### late frances elizabeth bavier, december 14, 1902 - - december 6, 1989. ) was part of the cast of ### +### the andy griffith show. [SEP] ### +### ======================================= h_v_q | Gates: 27527 ======================================= ### +### ('bee', 0, 0) ('aunt', 1, 2) ('name', 2, 18291) ('surname', 3, 14962) ('last', 4, 4981) ### +### ('familiarity', 5, 20522) ('was', 6, 51) ('uncle', 7, 41) ('names', 8, 12903) ### +### ('something', 9, 9490) ('.', 10, 4837) ('mom', 11, 35) ('noun', 12, 25750) ('final', 13, 2605) ### +### ('nickname', 14, 367) ('mother', 15, 25) ('sister', 16, 92) ('bees', 17, 9) ('relating', 18, 18896) ### +### ('pseudonym', 19, 493) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('bee', 0, 0) ('griffith', 11859, 1) ('aunt', 1, 2) ('beatrice', 6131, 3) ('##vier', 13458, 4) ### +### ('andy', 1895, 5) ('taylor', 1499, 6) ('cast', 8056, 7) ('show', 3508, 8) ('bees', 17, 9) ### +### ('frances', 5253, 10) ('december', 1756, 11) ('yes', 3228, 12) ('ba', 3285, 13) ('plays', 9076, 14) ### +### ('elizabeth', 139, 15) ('played', 826, 16) ('cyrillic', 485, 17) ('shows', 9638, 18) ### +### ('##ང', 193, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('bee', 0, 0) ('aunt', 1, 2) ('was', 6, 51) ('uncle', 7, 41) ('bees', 17, 9) ('mom', 11, 35) ### +### ('mother', 15, 25) ('grandmother', 31, 26) ('altogether', 20, 66) ('sister', 16, 92) ### +### ('last', 4, 4981) ('julian', 25, 74) ('−', 64, 24) ('remained', 43, 42) ('##₂', 54, 48) ### +### ('sharply', 103, 21) ('elizabeth', 139, 15) ('##大', 91, 33) ('gideon', 84, 38) ('knew', 42, 76) ### +############################################################################################################ +[2023-10-07 21:11:26,983][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:11:26,983][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:11:27,406][root][INFO] - Epoch: 6: Step: 301/1557, loss[v]=0.084155, lr=0.000015, acc@1[1]=242.5/256=0.947265625, acc@1[2]=248.0/256=0.96875 +[2023-10-07 21:12:43,630][root][INFO] - Train batch 400 +[2023-10-07 21:12:43,631][root][INFO] - Avg. loss per last 100 batches: 0.084884 +[2023-10-07 21:12:44,323][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29235.0/29522=99.03% | mean: 0.01 | max: 4.92 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.14 | max: 5.89 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the origin of a caldera volcano [SEP] ### +### [P_TEXT]: [CLS] origin of the yellowstone caldera. caldera, which lies at the heart of yellowstone ### +### national park, is the tip of a huge volcano, which is in turn, bound to a hot spot. this is a fixed ### +### point within the earth's crust, where hot magma rises to the top. this accumulates below the ### +### earth's surface, thereby forming an underground magma chamber. aldera, which lies at the heart of ### +### yellowstone national park, is the tip of a huge volcano, which is in turn, bound to a hot spot. ### +### this is a fixed point within the earth's crust, where hot magma rises to the top. [SEP] ### +### ======================================= h_v_q | Gates: 27157 ======================================= ### +### ('calder', 0, 1) ('volcano', 1, 2) ('origin', 2, 12) ('##a', 3, 26) ('origins', 4, 47) ### +### ('familiarity', 5, 25060) ('.', 6, 5328) ('relating', 7, 16571) ('originated', 8, 68) ### +### ('is', 9, 327) ('source', 10, 331) ('volcanic', 11, 11) ('a', 12, 9364) ('encompasses', 13, 17) ### +### ('association', 14, 7807) ('simon', 15, 71) ('julian', 16, 62) ('volcanoes', 17, 8) ### +### ('sources', 18, 1920) ('alexander', 19, 288) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('yellowstone', 20884, 0) ('calder', 0, 1) ('volcano', 1, 2) ('tip', 4695, 3) ('magma', 150, 4) ### +### ('hot', 919, 5) ('underground', 3546, 6) ('bound', 6766, 7) ('volcanoes', 17, 8) ('crust', 9369, 9) ### +### ('fixed', 4037, 10) ('volcanic', 11, 11) ('origin', 2, 12) ('ˈ', 274, 13) ('lies', 1309, 14) ### +### ('huge', 6739, 15) ('wingspan', 433, 16) ('encompasses', 13, 17) ('park', 370, 18) ('##ང', 175, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('calder', 0, 1) ('volcano', 1, 2) ('origin', 2, 12) ('##a', 3, 26) ('origins', 4, 47) ### +### ('volcanic', 11, 11) ('volcanoes', 17, 8) ('encompasses', 13, 17) ('originated', 8, 68) ### +### ('##₂', 28, 24) ('julian', 16, 62) ('ruins', 27, 33) ('simon', 15, 71) ('##α', 40, 32) ### +### ('magma', 150, 4) ('crashing', 48, 22) ('dante', 24, 59) ('hugh', 36, 45) ('fernando', 31, 61) ### +### ('hating', 72, 34) ### +############################################################################################################ +[2023-10-07 21:12:44,323][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:12:44,323][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:12:44,755][root][INFO] - Epoch: 6: Step: 401/1557, loss[v]=0.060699, lr=0.000014, acc@1[1]=244.0/256=0.953125, acc@1[2]=252.0/256=0.984375 +[2023-10-07 21:14:01,750][root][INFO] - Train batch 500 +[2023-10-07 21:14:01,751][root][INFO] - Avg. loss per last 100 batches: 0.085919 +[2023-10-07 21:14:02,485][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29201.1/29522=98.91% | mean: 0.01 | max: 5.54 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.12 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how many inches is the samsung s7 edge screen [SEP] ### +### [P_TEXT]: [CLS] actual size of samsung galaxy s7 edge. 1 depth : 7. 7mm ( 0. 3inch ) 2 height : ### +### 150. 9mm ( 5. 94inch ) 3 weight : 157g ( 5. 54oz ) width : 72. 6mm ( 2. 1 86inch ) screen - size : ### +### 5. 5inch ( 139. 2 7mm ) resolution : 1440 x 2560. [SEP] ### +### ======================================= h_v_q | Gates: 27478 ======================================= ### +### ('inches', 0, 491) ('edge', 1, 1) ('samsung', 2, 2) ('##7', 3, 6) ('screen', 4, 5) ('s', 5, 53) ### +### ('seven', 6, 140) ('numerous', 7, 96) ('.', 8, 6454) ('familiarity', 9, 26546) ('number', 10, 465) ### +### ('7', 11, 77) ('1000', 12, 1984) ('many', 13, 836) ('000', 14, 20539) ('##8', 15, 16) ### +### ('adam', 16, 280) ('1977', 17, 126) ('screens', 18, 36) ('months', 19, 17703) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('galaxy', 129, 0) ('edge', 1, 1) ('samsung', 2, 2) ('size', 111, 3) ('depth', 3370, 4) ### +### ('screen', 4, 5) ('##7', 3, 6) ('actual', 2195, 7) ('resolution', 1263, 8) ('edges', 33, 9) ### +### ('width', 443, 10) ('height', 165, 11) ('sizes', 1779, 12) ('##9', 36, 13) ('ˈ', 445, 14) ### +### ('galaxies', 10705, 15) ('##8', 15, 16) ('##ch', 12246, 17) ('weight', 564, 18) ('##ང', 214, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('edge', 1, 1) ('samsung', 2, 2) ('##7', 3, 6) ('screen', 4, 5) ('inches', 0, 491) ('s', 5, 53) ### +### ('##8', 15, 16) ('seven', 6, 140) ('numerous', 7, 96) ('7', 11, 77) ('screens', 18, 36) ### +### ('edges', 33, 9) ('##9', 36, 13) ('galaxy', 129, 0) ('##6', 32, 26) ('1977', 17, 126) ### +### ('size', 111, 3) ('diameter', 34, 42) ('number', 10, 465) ('length', 47, 29) ### +############################################################################################################ +[2023-10-07 21:14:02,485][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:14:02,485][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:14:02,904][root][INFO] - Epoch: 6: Step: 501/1557, loss[v]=0.040982, lr=0.000014, acc@1[1]=247.5/256=0.966796875, acc@1[2]=252.5/256=0.986328125 +[2023-10-07 21:15:19,898][root][INFO] - Train batch 600 +[2023-10-07 21:15:19,899][root][INFO] - Avg. loss per last 100 batches: 0.083329 +[2023-10-07 21:15:20,622][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29098.2/29522=98.56% | mean: 0.01 | max: 5.01 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.16 | max: 5.87 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is telerik [SEP] ### +### [P_TEXT]: [CLS] telerik is a bulgarian company offering software tools for web, mobile, desktop ### +### application development, tools and subscription services for cross - platform application ### +### development. [SEP] ### +### ======================================= h_v_q | Gates: 27957 ======================================= ### +### ('##erik', 0, 0) ('tel', 1, 3) ('is', 2, 82) ('definition', 3, 69) ('encompasses', 4, 5) ### +### ('refers', 5, 5265) ('familiarity', 6, 25509) ('noun', 7, 10981) ('gil', 8, 225) ('.', 9, 3212) ### +### ('stands', 10, 8721) ('relating', 11, 12117) ('defined', 12, 4076) ('term', 13, 16238) ### +### ('or', 14, 12913) ('digital', 15, 680) ('exists', 16, 12111) ('something', 17, 2708) ### +### ('dod', 18, 4422) ('smith', 19, 2448) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##erik', 0, 0) ('bulgarian', 9863, 1) ('tools', 1511, 2) ('tel', 1, 3) ('bulgaria', 6641, 4) ### +### ('encompasses', 4, 5) ('subscription', 12067, 6) ('software', 639, 7) ('desktop', 22728, 8) ### +### ('company', 1542, 9) ('ˈ', 706, 10) ('companies', 1648, 11) ('crashing', 85, 12) ### +### ('services', 4327, 13) ('web', 1448, 14) ('platform', 657, 15) ('##α', 381, 16) ('##ང', 114, 17) ### +### ('cross', 3466, 18) ('application', 4172, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##erik', 0, 0) ('tel', 1, 3) ('encompasses', 4, 5) ('is', 2, 82) ('definition', 3, 69) ### +### ('gil', 8, 225) ('julian', 41, 41) ('crashing', 85, 12) ('##ང', 114, 17) ('hating', 65, 40) ### +### ('##₂', 142, 28) ('what', 72, 64) ('software', 639, 7) ('afraid', 189, 27) ('hugh', 136, 37) ### +### ('##ο', 223, 26) ('anger', 107, 51) ('refers', 5, 5265) ('odd', 59, 93) ('##α', 381, 16) ### +############################################################################################################ +[2023-10-07 21:15:20,623][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:15:20,623][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:15:21,048][root][INFO] - Epoch: 6: Step: 601/1557, loss[v]=0.071752, lr=0.000014, acc@1[1]=242.0/256=0.9453125, acc@1[2]=252.5/256=0.986328125 +[2023-10-07 21:16:36,773][root][INFO] - Train batch 700 +[2023-10-07 21:16:36,774][root][INFO] - Avg. loss per last 100 batches: 0.081711 +[2023-10-07 21:16:37,498][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29210.8/29522=98.95% | mean: 0.01 | max: 5.41 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 5.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.09 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] who voices superman [SEP] ### +### [P_TEXT]: [CLS] those are the live action actors. several actors have played superman via voice ### +### performances, starting with bud collyer. bud played superman on the radio from 1940 - 1951. he also ### +### played superman in the 1940s fleischer cartoons and the 1960s new adventures of superman cartoons. ### +### danny dark played superman in the super friends cartoons. [SEP] ### +### ======================================= h_v_q | Gates: 27297 ======================================= ### +### ('superman', 0, 0) ('voices', 1, 26) ('voice', 2, 2) ('whose', 3, 173) ('.', 4, 5024) ### +### ('voiced', 5, 178) ('who', 6, 22) ('familiarity', 7, 22959) ('portrayed', 8, 57) ('batman', 9, 11) ### +### ('audio', 10, 381) ('sound', 11, 1056) ('authority', 12, 2204) ('vocals', 13, 1617) ### +### ('spoken', 14, 340) ('relating', 15, 17544) ('broadcast', 16, 635) ('sounds', 17, 5884) ### +### ('clark', 18, 217) ('whispers', 19, 4987) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('superman', 0, 0) ('bud', 10880, 1) ('voice', 2, 2) ('actors', 1002, 3) ('cartoons', 1694, 4) ### +### ('actor', 139, 5) ('dark', 1420, 6) ('friends', 2839, 7) ('col', 12068, 8) ('via', 1763, 9) ### +### ('danny', 1824, 10) ('batman', 9, 11) ('performances', 1144, 12) ('radio', 23, 13) ### +### ('live', 1168, 14) ('action', 549, 15) ('ˈ', 1210, 16) ('cyrillic', 1164, 17) ('##ང', 226, 18) ### +### ('##ly', 5101, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('superman', 0, 0) ('voice', 2, 2) ('voices', 1, 26) ('who', 6, 22) ('batman', 9, 11) ### +### ('whose', 3, 173) ('voiced', 5, 178) ('portrayed', 8, 57) ('radio', 23, 13) ('super', 38, 20) ### +### ('julian', 31, 46) ('actor', 139, 5) ('dc', 28, 76) ('.', 4, 5024) ('audio', 10, 381) ### +### ('##ང', 226, 18) ('crashing', 211, 21) ('actors', 1002, 3) ('clark', 18, 217) ('played', 207, 33) ### +############################################################################################################ +[2023-10-07 21:16:37,498][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:16:37,498][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:16:37,921][root][INFO] - Epoch: 6: Step: 701/1557, loss[v]=0.097092, lr=0.000014, acc@1[1]=242.0/256=0.9453125, acc@1[2]=248.0/256=0.96875 +[2023-10-07 21:17:53,909][root][INFO] - Train batch 800 +[2023-10-07 21:17:53,910][root][INFO] - Avg. loss per last 100 batches: 0.084406 +[2023-10-07 21:17:54,591][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29305.7/29522=99.27% | mean: 0.01 | max: 5.36 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.04 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what county is fort mill sc [SEP] ### +### [P_TEXT]: [CLS] fort mill, south carolina. fort mill, also known as fort mill township, is a ### +### suburban town in york county in the u. s. state of south carolina, and a suburb of the city of ### +### charlotte, north carolina, and is near rock hill. [SEP] ### +### ======================================= h_v_q | Gates: 27511 ======================================= ### +### ('fort', 0, 0) ('mill', 1, 1) ('county', 2, 7) ('carolina', 3, 2) ('virginia', 4, 401) ### +### ('sc', 5, 15) ('illinois', 6, 4052) ('mississippi', 7, 235) ('familiarity', 8, 20227) ### +### ('colorado', 9, 9660) ('pennsylvania', 10, 2470) ('mills', 11, 19) ('ohio', 12, 3991) ### +### ('california', 13, 3682) ('louisiana', 14, 1981) ('missouri', 15, 4521) ('counties', 16, 18) ### +### ('relating', 17, 18049) ('wisconsin', 18, 4879) ('texas', 19, 1466) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('fort', 0, 0) ('mill', 1, 1) ('carolina', 3, 2) ('charlotte', 1549, 3) ('york', 147, 4) ### +### ('suburb', 4681, 5) ('township', 49, 6) ('county', 2, 7) ('nc', 608, 8) ('suburban', 2180, 9) ### +### ('rock', 311, 10) ('encompasses', 30, 11) ('south', 142, 12) ('suburbs', 1014, 13) ('ˈ', 446, 14) ### +### ('sc', 5, 15) ('town', 77, 16) ('where', 459, 17) ('counties', 16, 18) ('mills', 11, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('fort', 0, 0) ('mill', 1, 1) ('county', 2, 7) ('carolina', 3, 2) ('sc', 5, 15) ('mills', 11, 19) ### +### ('counties', 16, 18) ('encompasses', 30, 11) ('township', 49, 6) ('virginia', 4, 401) ### +### ('mississippi', 7, 235) ('york', 147, 4) ('town', 77, 16) ('is', 21, 140) ('afraid', 62, 29) ### +### ('south', 142, 12) ('forts', 51, 57) ('julian', 68, 39) ('rural', 71, 48) ('farm', 22, 214) ### +############################################################################################################ +[2023-10-07 21:17:54,592][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:17:54,592][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:17:55,014][root][INFO] - Epoch: 6: Step: 801/1557, loss[v]=0.080648, lr=0.000014, acc@1[1]=237.0/256=0.92578125, acc@1[2]=251.5/256=0.982421875 +[2023-10-07 21:19:12,079][root][INFO] - Train batch 900 +[2023-10-07 21:19:12,079][root][INFO] - Avg. loss per last 100 batches: 0.082559 +[2023-10-07 21:19:12,782][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29307.6/29522=99.27% | mean: 0.01 | max: 5.30 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.02 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] when an oligopoly exists, how many producers dominate the market? [SEP] ### +### [P_TEXT]: [CLS] brainly has millions of high quality answers, all of them carefully moderated by ### +### our most trusted community members, but certified answers are the finest of the finest. the ### +### oligopoly is known to have a one producer dominating the market. this results in a few suppliers / ### +### sellers in the market, and thus can cause a high increase in the price of the products that are ### +### being sold in its respective community. [SEP] ### +### ======================================= h_v_q | Gates: 28498 ======================================= ### +### ('ol', 0, 16) ('producers', 1, 18) ('exists', 2, 10854) ('market', 3, 4) ('##igo', 4, 8) ### +### ('##pol', 5, 28) ('##y', 6, 43) ('dominate', 7, 31) ('dominated', 8, 49) ('producer', 9, 6) ### +### ('.', 10, 4404) ('existed', 11, 13373) ('numerous', 12, 57) ('familiarity', 13, 25379) ### +### ('five', 14, 8381) ('exist', 15, 20060) ('1000', 16, 6155) ('000', 17, 9126) ('seven', 18, 8592) ### +### ('six', 19, 8719) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('brain', 157, 0) ('certified', 17490, 1) ('finest', 20666, 2) ('##ly', 429, 3) ('market', 3, 4) ### +### ('answers', 14155, 5) ('producer', 9, 6) ('moderate', 14193, 7) ('##igo', 4, 8) ### +### ('quality', 1727, 9) ('trusted', 13561, 10) ('suppliers', 679, 11) ('ˈ', 930, 12) ### +### ('community', 964, 13) ('carefully', 4906, 14) ('encompasses', 60, 15) ('ol', 0, 16) ### +### ('price', 3745, 17) ('producers', 1, 18) ('answer', 263, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('market', 3, 4) ('ol', 0, 16) ('producers', 1, 18) ('##igo', 4, 8) ('##pol', 5, 28) ### +### ('producer', 9, 6) ('dominate', 7, 31) ('##y', 6, 43) ('dominated', 8, 49) ('numerous', 12, 57) ### +### ('dominates', 22, 47) ('encompasses', 60, 15) ('dominant', 21, 161) ('brain', 157, 0) ### +### ('many', 42, 100) ('million', 23, 249) ('julian', 55, 76) ('number', 20, 353) ('exists', 2, 10854) ### +### ('production', 37, 160) ### +############################################################################################################ +[2023-10-07 21:19:12,782][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:19:12,783][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:19:13,207][root][INFO] - Epoch: 6: Step: 901/1557, loss[v]=0.100270, lr=0.000014, acc@1[1]=238.0/256=0.9296875, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 21:20:29,352][root][INFO] - Train batch 1000 +[2023-10-07 21:20:29,353][root][INFO] - Avg. loss per last 100 batches: 0.078694 +[2023-10-07 21:20:30,080][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29293.1/29522=99.22% | mean: 0.01 | max: 5.15 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.16 | max: 6.29 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] cost of regrading a crawl space [SEP] ### +### [P_TEXT]: [CLS] crawl space experts say sealing a crawl space usually costs about $ 5, 500 on ### +### average, but the costs can easily range from $ 1, 500 to $ 15, 000 depending on the issues and the ### +### size of the home. [SEP] ### +### ======================================= h_v_q | Gates: 27872 ======================================= ### +### ('crawl', 0, 0) ('space', 1, 1) ('$', 2, 6) ('##rad', 3, 28011) ('cost', 4, 4) ('reg', 5, 26002) ### +### ('familiarity', 6, 20865) ('crawled', 7, 21) ('.', 8, 4834) ('luke', 9, 392) ('##ing', 10, 16427) ### +### ('relating', 11, 22698) ('costs', 12, 2) ('climb', 13, 426) ('walk', 14, 328) ('price', 15, 7) ### +### ('plural', 16, 8433) ('martin', 17, 1043) ('crawling', 18, 48) ('spaces', 19, 13) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('crawl', 0, 0) ('space', 1, 1) ('costs', 12, 2) ('sealing', 10042, 3) ('cost', 4, 4) ### +### ('depending', 9216, 5) ('$', 2, 6) ('price', 15, 7) ('ˈ', 411, 8) ('500', 1289, 9) ### +### ('seal', 5331, 10) ('300', 638, 11) ('wingspan', 118, 12) ('spaces', 19, 13) ('##ང', 35, 14) ### +### ('sealed', 5969, 15) ('average', 133, 16) ('hesitated', 54, 17) ('##ο', 96, 18) ('hating', 89, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('crawl', 0, 0) ('space', 1, 1) ('$', 2, 6) ('cost', 4, 4) ('crawled', 7, 21) ('costs', 12, 2) ### +### ('price', 15, 7) ('spaces', 19, 13) ('crawling', 18, 48) ('##ང', 35, 14) ('hesitated', 54, 17) ### +### ('##₂', 26, 46) ('julian', 23, 67) ('simon', 21, 72) ('ruins', 28, 47) ('stumbled', 48, 24) ### +### ('hugh', 32, 40) ('altogether', 27, 60) ('wingspan', 118, 12) ('afraid', 55, 28) ### +############################################################################################################ +[2023-10-07 21:20:30,081][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:20:30,081][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:20:30,510][root][INFO] - Epoch: 6: Step: 1001/1557, loss[v]=0.095158, lr=0.000014, acc@1[1]=243.5/256=0.951171875, acc@1[2]=249.5/256=0.974609375 +[2023-10-07 21:21:46,309][root][INFO] - Train batch 1100 +[2023-10-07 21:21:46,310][root][INFO] - Avg. loss per last 100 batches: 0.086798 +[2023-10-07 21:21:47,031][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29258.2/29522=99.11% | mean: 0.01 | max: 5.33 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.16 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is italian meatball made of [SEP] ### +### [P_TEXT]: [CLS] select the meatball ingredients. 1 ground meat : choose from ground beef, veal, ### +### pork, italian sausage, lamb, bison, turkey, and / or chicken. 2 some recipes call for a mix of two ### +### or more ground meats. 3 lean ground meat works well and makes the meatballs more healthful. [SEP] ### +### ======================================= h_v_q | Gates: 27468 ======================================= ### +### ('##ball', 0, 2) ('meat', 1, 3) ('italian', 2, 7) ('made', 3, 1767) ('italy', 4, 11) ### +### ('is', 5, 20065) ('familiarity', 6, 21194) ('.', 7, 5460) ('produced', 8, 15158) ### +### ('relating', 9, 22111) ('created', 10, 4634) ('definition', 11, 4475) ('manufactured', 12, 580) ### +### ('refers', 13, 19665) ('milan', 14, 1606) ('make', 15, 342) ('encompasses', 16, 356) ### +### ('albert', 17, 602) ('stands', 18, 10471) ('##balls', 19, 1) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('ground', 8982, 0) ('##balls', 19, 1) ('##ball', 0, 2) ('meat', 1, 3) ('bison', 16960, 4) ### +### ('ingredients', 1033, 5) ('lean', 3389, 6) ('italian', 2, 7) ('select', 3854, 8) ### +### ('sausage', 1969, 9) ('beef', 254, 10) ('italy', 4, 11) ('ˈ', 326, 12) ('turkey', 679, 13) ### +### ('choose', 15068, 14) ('recipes', 20924, 15) ('lamb', 4828, 16) ('pork', 5360, 17) ### +### ('crashing', 58, 18) ('mix', 2928, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##ball', 0, 2) ('meat', 1, 3) ('italian', 2, 7) ('italy', 4, 11) ('##balls', 19, 1) ### +### ('made', 3, 1767) ('crashing', 58, 18) ('gideon', 55, 25) ('julian', 28, 85) ('##ο', 64, 28) ### +### ('##ང', 74, 22) ('ball', 22, 143) ('simon', 37, 72) ('hating', 78, 20) ('makes', 27, 117) ### +### ('stumbled', 79, 29) ('hugh', 65, 51) ('presenter', 70, 50) ('##₂', 93, 35) ('julia', 101, 23) ### +############################################################################################################ +[2023-10-07 21:21:47,032][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:21:47,032][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:21:47,452][root][INFO] - Epoch: 6: Step: 1101/1557, loss[v]=0.107659, lr=0.000014, acc@1[1]=238.5/256=0.931640625, acc@1[2]=245.0/256=0.95703125 +[2023-10-07 21:23:04,091][root][INFO] - Train batch 1200 +[2023-10-07 21:23:04,092][root][INFO] - Avg. loss per last 100 batches: 0.087994 +[2023-10-07 21:23:04,776][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29233.9/29522=99.02% | mean: 0.01 | max: 5.22 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 5.93 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what comes with gears of war ultimate edition [SEP] ### +### [P_TEXT]: [CLS] gears of war : ultimate edition comes with all the gears of war games, microsoft ### +### has announced. nice! if you buy gears of war : ultimate edition, or the xbox one gears of war : ### +### ultimate edition bundle, you unlock the entire gears collection via xbox one backward ### +### compatibility. [SEP] ### +### ======================================= h_v_q | Gates: 27889 ======================================= ### +### ('war', 0, 2) ('ultimate', 1, 1) ('gears', 2, 0) ('edition', 3, 3) ('comes', 4, 101) ### +### ('editions', 5, 21) ('.', 6, 12468) ('came', 7, 139) ('familiarity', 8, 19964) ('battle', 9, 80) ### +### ('version', 10, 36) ('gear', 11, 54) ('of', 12, 6088) ('what', 13, 1974) ('julian', 14, 95) ### +### ('anton', 15, 223) ('with', 16, 159) ('plural', 17, 1903) ('albert', 18, 296) ('military', 19, 123) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('gears', 2, 0) ('ultimate', 1, 1) ('war', 0, 2) ('edition', 3, 3) ('xbox', 21648, 4) ### +### ('games', 2521, 5) ('nice', 1613, 6) ('ˈ', 496, 7) ('backward', 4680, 8) ('game', 628, 9) ### +### ('wingspan', 292, 10) ('##ང', 73, 11) ('buy', 1040, 12) ('microsoft', 13210, 13) ### +### ('bundle', 2931, 14) ('crashing', 124, 15) ('##ο', 228, 16) ('##α', 142, 17) ('hesitated', 196, 18) ### +### ('unlock', 20626, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('gears', 2, 0) ('war', 0, 2) ('ultimate', 1, 1) ('edition', 3, 3) ('comes', 4, 101) ### +### ('editions', 5, 21) ('version', 10, 36) ('gear', 11, 54) ('battle', 9, 80) ('came', 7, 139) ### +### ('wars', 20, 41) ('julian', 14, 95) ('gideon', 27, 23) ('warfare', 21, 51) ('altogether', 25, 49) ### +### ('combat', 23, 67) ('hugh', 33, 47) ('with', 16, 159) ('##ང', 73, 11) ('military', 19, 123) ### +############################################################################################################ +[2023-10-07 21:23:04,776][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:23:04,776][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:23:05,195][root][INFO] - Epoch: 6: Step: 1201/1557, loss[v]=0.062274, lr=0.000014, acc@1[1]=246.5/256=0.962890625, acc@1[2]=252.0/256=0.984375 +[2023-10-07 21:24:22,238][root][INFO] - Train batch 1300 +[2023-10-07 21:24:22,239][root][INFO] - Avg. loss per last 100 batches: 0.079657 +[2023-10-07 21:24:22,948][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29107.2/29522=98.60% | mean: 0.01 | max: 5.50 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.16 | max: 6.40 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] agitated definition [SEP] ### +### [P_TEXT]: [CLS] adjective. verb. the definition of agitated is someone who is very distressed or ### +### troubled. being so nervous about something that you are constantly pacing is an example of ### +### agitated. agitated means to have shaken up or stirred - up something. djective. verb. the ### +### definition of agitated is someone who is very distressed or troubled. being so nervous about ### +### something that you are constantly pacing is an example of agitated. [SEP] ### +### ======================================= h_v_q | Gates: 27636 ======================================= ### +### ('agitated', 0, 0) ('agitation', 1, 10) ('definition', 2, 7) ('or', 3, 11549) ('.', 4, 9555) ### +### (';', 5, 7872) ('angry', 6, 73) ('annoyed', 7, 81) ('frustrated', 8, 70) ('excited', 9, 75) ### +### ('upset', 10, 96) ('noun', 11, 4126) ('irritated', 12, 207) ('movement', 13, 636) ### +### ('nervous', 14, 13) ('disturbed', 15, 123) ('disturbance', 16, 430) ('defined', 17, 66) ### +### ('anger', 18, 29) ('anxious', 19, 43) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('agitated', 0, 0) ('pacing', 129, 1) ('dj', 11450, 2) ('troubled', 368, 3) ('distressed', 39, 4) ### +### ('stirred', 21, 5) ('shaken', 4442, 6) ('definition', 2, 7) ('definitions', 102, 8) ### +### ('define', 6381, 9) ('agitation', 1, 10) ('constantly', 2812, 11) ('meaning', 51, 12) ### +### ('nervous', 14, 13) ('afraid', 1415, 14) ('adjective', 16837, 15) ('ˈ', 5820, 16) ### +### ('stirring', 81, 17) ('worried', 32, 18) ('somewhere', 1984, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('agitated', 0, 0) ('agitation', 1, 10) ('definition', 2, 7) ('angry', 6, 73) ('nervous', 14, 13) ### +### ('frustrated', 8, 70) ('annoyed', 7, 81) ('stirred', 21, 5) ('excited', 9, 75) ### +### ('distressed', 39, 4) ('anger', 18, 29) ('upset', 10, 96) ('anxious', 19, 43) ('worried', 32, 18) ### +### ('defined', 17, 66) ('meaning', 51, 12) ('pacing', 129, 1) ('disturbed', 15, 123) ### +### ('definitions', 102, 8) ('irritated', 12, 207) ### +############################################################################################################ +[2023-10-07 21:24:22,948][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:24:22,948][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:24:23,373][root][INFO] - Epoch: 6: Step: 1301/1557, loss[v]=0.055310, lr=0.000014, acc@1[1]=245.5/256=0.958984375, acc@1[2]=254.0/256=0.9921875 +[2023-10-07 21:25:39,309][root][INFO] - Train batch 1400 +[2023-10-07 21:25:39,309][root][INFO] - Avg. loss per last 100 batches: 0.081272 +[2023-10-07 21:25:40,027][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29169.6/29522=98.81% | mean: 0.01 | max: 5.60 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.35 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what does janky mean [SEP] ### +### [P_TEXT]: [CLS] obsidian entertainment is the definition of janky. glitchy, buggy, weird issues in ### +### the game, something goes wrong in a game randomly. it is the collision of multiple'yes / ### +### no'commands in the game that results in weird anomalies throughout. [SEP] ### +### ======================================= h_v_q | Gates: 26902 ======================================= ### +### ('##ky', 0, 2) ('jan', 1, 3) ('noun', 2, 26048) ('definition', 3, 7) ('familiarity', 4, 26596) ### +### ('means', 5, 266) ('something', 6, 796) ('meaning', 7, 20) ('sense', 8, 4487) ('##sky', 9, 74) ### +### ('is', 10, 438) ('refers', 11, 13187) ('.', 12, 9710) ('##ny', 13, 131) ('latin', 14, 2944) ### +### ('mean', 15, 54) ('relating', 16, 18559) ('term', 17, 1765) (';', 18, 7214) ('murray', 19, 6043) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('obsidian', 6625, 0) ('weird', 699, 1) ('##ky', 0, 2) ('jan', 1, 3) ('entertainment', 2044, 4) ### +### ('definitions', 414, 5) ('bug', 16157, 6) ('definition', 3, 7) ('define', 8108, 8) ('##gy', 125, 9) ### +### ('commands', 16114, 10) ('##chy', 10459, 11) ('wrong', 2864, 12) ('game', 30, 13) ('ˈ', 652, 14) ### +### ('crashing', 34, 15) ('encompasses', 47, 16) ('odd', 291, 17) ('strange', 570, 18) ### +### ('command', 833, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##ky', 0, 2) ('jan', 1, 3) ('definition', 3, 7) ('meaning', 7, 20) ('##sky', 9, 74) ### +### ('means', 5, 266) ('mean', 15, 54) ('game', 30, 13) ('crashing', 34, 15) ('##ny', 13, 131) ### +### ('encompasses', 47, 16) ('something', 6, 796) ('is', 10, 438) ('crashed', 39, 41) ('##gy', 125, 9) ### +### ('defined', 36, 59) ('##ང', 61, 31) ('stumbled', 73, 25) ('weird', 699, 1) ('jean', 22, 142) ### +############################################################################################################ +[2023-10-07 21:25:40,027][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:25:40,027][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:25:40,450][root][INFO] - Epoch: 6: Step: 1401/1557, loss[v]=0.084784, lr=0.000014, acc@1[1]=238.5/256=0.931640625, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 21:26:56,637][root][INFO] - Train batch 1500 +[2023-10-07 21:26:56,637][root][INFO] - Avg. loss per last 100 batches: 0.088141 +[2023-10-07 21:26:57,349][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29131.6/29522=98.68% | mean: 0.01 | max: 5.20 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.4/29522=100.00% | mean: 0.14 | max: 6.00 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the kanasas city club [SEP] ### +### [P_TEXT]: [CLS] from wikipedia, the free encyclopedia. the kansas city club, founded in 1882 and ### +### located in the library district of downtown kansas city, missouri, usa, is the oldest existing ### +### gentlemen's club in missouri. n 2002, a developer bought the kansas city club's 1922 building and ### +### turned it into loft apartments and a banquet hall, renaming it the clubhouse on baltimore.. since ### +### 2010, the club has lent space to washington university in st. louis's olin school of business local ### +### executive mba program. [SEP] ### +### ======================================= h_v_q | Gates: 27428 ======================================= ### +### ('kan', 0, 6549) ('club', 1, 0) ('##asa', 2, 24307) ('city', 3, 7) ('clubs', 4, 10) ### +### ('##s', 5, 18986) ('is', 6, 423) ('familiarity', 7, 25273) ('district', 8, 14) ('village', 9, 239) ### +### ('urban', 10, 81) ('town', 11, 91) ('relating', 12, 18133) ('local', 13, 84) ('.', 14, 2676) ### +### ('association', 15, 959) ('what', 16, 597) ('refers', 17, 18692) ('encompasses', 18, 31) ### +### ('street', 19, 164) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('club', 1, 0) ('missouri', 348, 1) ('baltimore', 1628, 2) ('kansas', 552, 3) ### +### ('gentlemen', 11659, 4) ('oldest', 2178, 5) ('downtown', 122, 6) ('city', 3, 7) ('loft', 11222, 8) ### +### ('clubhouse', 520, 9) ('clubs', 4, 10) ('banquet', 11212, 11) ('founded', 44, 12) ### +### ('cities', 29, 13) ('district', 8, 14) ('1882', 8677, 15) ('gentleman', 14041, 16) ('ˈ', 1598, 17) ### +### ('stab', 206, 18) ('where', 4169, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('club', 1, 0) ('city', 3, 7) ('clubs', 4, 10) ('district', 8, 14) ('kan', 0, 6549) ### +### ('urban', 10, 81) ('encompasses', 18, 31) ('cities', 29, 13) ('town', 11, 91) ('local', 13, 84) ### +### ('is', 6, 423) ('founded', 44, 12) ('village', 9, 239) ('downtown', 122, 6) ('missouri', 348, 1) ### +### ('street', 19, 164) ('julian', 50, 38) ('nightclub', 46, 68) ('kansas', 552, 3) ### +### ('crashing', 95, 28) ### +############################################################################################################ +[2023-10-07 21:26:57,350][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:26:57,350][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:26:57,751][root][INFO] - Epoch: 6: Step: 1501/1557, loss[v]=0.079288, lr=0.000014, acc@1[1]=239.0/256=0.93359375, acc@1[2]=254.0/256=0.9921875 +[2023-10-07 21:27:41,608][root][INFO] - rank=1; last iteration 1557 +[2023-10-07 21:27:41,608][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 21:27:41,608][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-07 21:27:41,614][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 21:27:41,614][root][INFO] - Epoch finished on 1 +[2023-10-07 21:27:41,633][root][INFO] - rank=3; last iteration 1557 +[2023-10-07 21:27:41,633][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 21:27:41,633][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-07 21:27:41,635][root][INFO] - rank=2; last iteration 1557 +[2023-10-07 21:27:41,635][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 21:27:41,635][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-07 21:27:41,637][root][INFO] - rank=0; last iteration 1557 +[2023-10-07 21:27:41,637][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 21:27:41,637][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-07 21:27:41,641][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 21:27:41,641][root][INFO] - Epoch finished on 3 +[2023-10-07 21:27:41,643][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 21:27:41,644][root][INFO] - Epoch finished on 2 +[2023-10-07 21:27:41,644][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 21:27:41,645][root][INFO] - Epoch finished on 0 +[2023-10-07 21:28:22,333][root][INFO] - Saved checkpoint at ./vdr_6 +[2023-10-07 21:28:22,334][root][INFO] - Av Loss per epoch=0.084486 +[2023-10-07 21:28:22,334][root][INFO] - epoch total (1) correct predictions=375934 +[2023-10-07 21:28:22,334][root][INFO] - Saved checkpoint at ./vdr_6 +[2023-10-07 21:28:22,335][root][INFO] - epoch total (2) correct predictions=388660 +[2023-10-07 21:28:22,335][root][INFO] - Av Loss per epoch=0.084486 +[2023-10-07 21:28:22,335][root][INFO] - epoch total (1) correct predictions=375934 +[2023-10-07 21:28:22,335][root][INFO] - epoch total (2) correct predictions=388660 +[2023-10-07 21:28:22,335][root][INFO] - Saved checkpoint at ./vdr_6 +[2023-10-07 21:28:22,336][root][INFO] - Av Loss per epoch=0.084486 +[2023-10-07 21:28:22,336][root][INFO] - epoch total (1) correct predictions=375934 +[2023-10-07 21:28:22,336][root][INFO] - epoch total (2) correct predictions=388660 +[2023-10-07 21:28:22,338][root][INFO] - ***** Epoch 7 ***** +[2023-10-07 21:28:22,338][root][INFO] - Saved checkpoint at ./vdr_6 +[2023-10-07 21:28:22,340][root][INFO] - Av Loss per epoch=0.084486 +[2023-10-07 21:28:22,340][root][INFO] - epoch total (1) correct predictions=375934 +[2023-10-07 21:28:22,341][root][INFO] - epoch total (2) correct predictions=388660 +[2023-10-07 21:28:22,340][root][INFO] - ***** Epoch 7 ***** +[2023-10-07 21:28:22,345][root][INFO] - rank=3; Iteration start +[2023-10-07 21:28:22,345][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 21:28:22,345][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 21:28:22,346][root][INFO] - rank=2; Iteration start +[2023-10-07 21:28:22,346][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 21:28:22,347][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 21:28:22,347][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-07 21:28:22,346][root][INFO] - ***** Epoch 7 ***** +[2023-10-07 21:28:22,348][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-07 21:28:22,349][root][INFO] - ***** Epoch 7 ***** +[2023-10-07 21:28:22,356][root][INFO] - rank=0; Iteration start +[2023-10-07 21:28:22,356][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 21:28:22,356][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 21:28:22,359][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-07 21:28:22,360][root][INFO] - rank=1; Iteration start +[2023-10-07 21:28:22,360][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 21:28:22,361][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 21:28:22,363][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-07 21:28:23,457][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29226.8/29522=99.00% | mean: 0.01 | max: 5.28 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.10 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] is castor oil ok to use on lawn for moles? [SEP] ### +### [P_TEXT]: [CLS] it isn't a lethal trap or poison, but a remarkably simple formula containing a¦ ### +### castor oil! in fact, in a michigan state university study, a castor oil - based repellent was ### +### effective in 26 out of 27 tests. product details. apply to lawn areas where mole activity is ### +### present. [SEP] ### +### ======================================= h_v_q | Gates: 28144 ======================================= ### +### ('cast', 0, 0) ('lawn', 1, 9) ('##or', 2, 13) ('mole', 3, 3) ('oil', 4, 5) ### +### ('familiarity', 5, 25058) ('use', 6, 10189) ('ok', 7, 8170) ('##s', 8, 22398) ('used', 9, 12515) ### +### ('oklahoma', 10, 8470) ('relating', 11, 19244) ('.', 12, 12404) ('uses', 13, 1249) ### +### ('graham', 14, 1613) ('##ors', 15, 111) ('for', 16, 24317) ('useful', 17, 283) ('beside', 18, 124) ### +### ('altogether', 19, 142) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('cast', 0, 0) ('##¦', 29306, 1) ('lethal', 5008, 2) ('mole', 3, 3) ('poison', 4603, 4) ### +### ('oil', 4, 5) ('formula', 6549, 6) ('effective', 158, 7) ('trap', 4091, 8) ('lawn', 1, 9) ### +### ('rep', 6413, 10) ('michigan', 85, 11) ('ˈ', 582, 12) ('##or', 2, 13) ('remarkably', 11320, 14) ### +### ('simple', 2304, 15) ('##ང', 51, 16) ('wingspan', 137, 17) ('containing', 2466, 18) ### +### ('hesitated', 89, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('cast', 0, 0) ('mole', 3, 3) ('lawn', 1, 9) ('oil', 4, 5) ('##or', 2, 13) ('crashing', 26, 20) ### +### ('##ors', 15, 111) ('##ང', 51, 16) ('michigan', 85, 11) ('ruins', 31, 53) ('effective', 158, 7) ### +### ('anger', 29, 63) ('hating', 45, 37) ('cyrillic', 42, 38) ('dangerous', 37, 54) ('beside', 18, 124) ### +### ('sharply', 68, 26) ('hugh', 44, 48) ('hesitated', 89, 19) ('angrily', 52, 45) ### +############################################################################################################ +[2023-10-07 21:28:23,457][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:28:23,457][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:28:23,846][root][INFO] - Epoch: 7: Step: 1/1557, loss[v]=0.076537, lr=0.000014, acc@1[1]=242.0/256=0.9453125, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 21:29:40,953][root][INFO] - Train batch 100 +[2023-10-07 21:29:40,953][root][INFO] - Avg. loss per last 100 batches: 0.083954 +[2023-10-07 21:29:41,674][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29084.6/29522=98.52% | mean: 0.01 | max: 5.22 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 5.93 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is meant by american imperialism [SEP] ### +### [P_TEXT]: [CLS] imperialism is a type of advocacy of empire. its name originated from the latin ### +### word imperium, which means to rule over large territories. imperialism is a policy of extending a ### +### country's power and influence through colonization, use of military force, or other means. ### +### imperialism has greatly shaped the contemporary world. he term imperialism has been applied to ### +### western ( and japanese ) political and economic dominance especially in asia and africa in the 19th ### +### and 20th centuries. its precise meaning continues to be debated by scholars. [SEP] ### +### ======================================= h_v_q | Gates: 26548 ======================================= ### +### ('imperialism', 0, 0) ('american', 1, 14224) ('means', 2, 28) ('noun', 3, 23863) ### +### ('america', 4, 6691) ('definition', 5, 22) ('meant', 6, 206) ('meaning', 7, 16) ('sense', 8, 3940) ### +### ('.', 9, 9359) ('or', 10, 15625) ('language', 11, 1596) ('familiarity', 12, 21022) ('is', 13, 196) ### +### ('something', 14, 3116) ('latin', 15, 5) ('americans', 16, 11937) ('refers', 17, 8015) ### +### ('term', 18, 71) ('phrase', 19, 539) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('imperialism', 0, 0) ('advocacy', 5908, 1) ('empire', 25, 2) ('imp', 12917, 3) ### +### ('encompasses', 35, 4) ('latin', 15, 5) ('colonization', 411, 6) ('originated', 1410, 7) ### +### ('crashing', 569, 8) ('ˈ', 2781, 9) ('advocate', 10758, 10) ('##eri', 19595, 11) ### +### ('policy', 684, 12) ('wingspan', 1453, 13) ('asia', 2489, 14) ('greatly', 2507, 15) ### +### ('meaning', 7, 16) ('define', 11581, 17) ('##ང', 783, 18) ('contemporary', 1525, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('imperialism', 0, 0) ('means', 2, 28) ('definition', 5, 22) ('meaning', 7, 16) ('latin', 15, 5) ### +### ('empire', 25, 2) ('meant', 6, 206) ('term', 18, 71) ('encompasses', 35, 4) ('fascism', 31, 23) ### +### ('is', 13, 196) ('mean', 29, 60) ('racism', 24, 111) ('capitalism', 34, 48) ('american', 1, 14224) ### +### ('influence', 48, 39) ('word', 79, 27) ('nationalism', 68, 36) ('words', 27, 139) ### +### ('africa', 56, 73) ### +############################################################################################################ +[2023-10-07 21:29:41,674][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:29:41,674][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:29:42,096][root][INFO] - Epoch: 7: Step: 101/1557, loss[v]=0.051691, lr=0.000014, acc@1[1]=246.5/256=0.962890625, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 21:30:59,130][root][INFO] - Train batch 200 +[2023-10-07 21:30:59,131][root][INFO] - Avg. loss per last 100 batches: 0.077311 +[2023-10-07 21:30:59,834][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29147.7/29522=98.73% | mean: 0.01 | max: 5.51 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.20 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] does the sun have a solid surface [SEP] ### +### [P_TEXT]: [CLS] click on image for larger version. the photosphere is the visible surface of the ### +### sun that we are most familiar with. since the sun is a ball of gas, this is not a solid surface but ### +### is actually a layer about 100 km thick ( very, very, thin compared to the 700, 000 km radius of the ### +### sun ). ince the sun is a ball of gas, this is not a solid surface but is actually a layer about 100 ### +### km thick ( very, very, thin compared to the 700, 000 km radius of the sun ). [SEP] ### +### ======================================= h_v_q | Gates: 26895 ======================================= ### +### ('sun', 0, 0) ('solid', 1, 14) ('surface', 2, 10) ('having', 3, 4350) ('.', 4, 10391) ### +### ('have', 5, 21290) ('has', 6, 9336) ('familiarity', 7, 13121) ('relating', 8, 13727) ### +### ('had', 9, 13060) ('surfaces', 10, 85) ('does', 11, 12841) ('answer', 12, 13311) ('firm', 13, 2474) ### +### ('a', 14, 11989) ('russell', 15, 317) ('refers', 16, 11701) ('is', 17, 234) ### +### ('mathematics', 18, 12059) ('plural', 19, 4377) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('sun', 0, 0) ('##pher', 17336, 1) ('visible', 1595, 2) ('inc', 15026, 3) ('photos', 11344, 4) ### +### ('gas', 245, 5) ('ball', 4347, 6) ('thin', 333, 7) ('encompasses', 90, 8) ('wingspan', 64, 9) ### +### ('surface', 2, 10) ('##ང', 75, 11) ('layers', 7346, 12) ('ˈ', 428, 13) ('solid', 1, 14) ### +### ('radius', 10102, 15) ('crashing', 43, 16) ('suns', 143, 17) ('afraid', 154, 18) ### +### ('hesitated', 108, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('sun', 0, 0) ('solid', 1, 14) ('surface', 2, 10) ('surfaces', 10, 85) ('ruins', 24, 41) ### +### ('crashing', 43, 16) ('##₂', 27, 30) ('wingspan', 64, 9) ('hating', 41, 23) ('cyrillic', 37, 33) ### +### ('##ང', 75, 11) ('encompasses', 90, 8) ('gideon', 61, 22) ('julian', 25, 65) ('ছ', 72, 24) ### +### ('−', 59, 35) ('presenter', 40, 46) ('julia', 36, 52) ('hesitated', 108, 19) ('bother', 68, 38) ### +############################################################################################################ +[2023-10-07 21:30:59,834][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:30:59,834][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:31:00,257][root][INFO] - Epoch: 7: Step: 201/1557, loss[v]=0.065396, lr=0.000014, acc@1[1]=242.0/256=0.9453125, acc@1[2]=253.5/256=0.990234375 +[2023-10-07 21:32:16,814][root][INFO] - Train batch 300 +[2023-10-07 21:32:16,815][root][INFO] - Avg. loss per last 100 batches: 0.078016 +[2023-10-07 21:32:17,542][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29221.1/29522=98.98% | mean: 0.01 | max: 5.22 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.36 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] white knuckle meaning [SEP] ### +### [P_TEXT]: [CLS] white knuckle meaning | white knuckle definition. the meaning of white knuckle is : ### +### the definition of white knuckle is : ( adjective ) something that causes a lot of excitement, ### +### tension, or stress. example sentences : i had a white knuckle ride when i taught my young son to ### +### drive. the roller coaster provided some white knuckle moments for the scared riders. media : [SEP] ### +### ======================================= h_v_q | Gates: 27632 ======================================= ### +### ('white', 0, 0) ('kn', 1, 2) ('##uck', 2, 6) ('##le', 3, 13) ('meaning', 4, 4) ('definition', 5, 9) ### +### ('noun', 6, 13891) ('familiarity', 7, 23749) ('relating', 8, 16499) ('sense', 9, 2389) ### +### ('means', 10, 50) ('.', 11, 6264) ('plural', 12, 2925) ('latin', 13, 2971) ('something', 14, 245) ### +### ('black', 15, 158) ('whites', 16, 16) ('whitish', 17, 126) ('language', 18, 1857) ### +### ('crashing', 19, 19) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('white', 0, 0) ('scared', 4289, 1) ('kn', 1, 2) ('definitions', 282, 3) ('meaning', 4, 4) ### +### ('coaster', 18236, 5) ('##uck', 2, 6) ('afraid', 100, 7) ('define', 13612, 8) ('definition', 5, 9) ### +### ('ride', 11849, 10) ('stress', 9485, 11) ('meanings', 504, 12) ('##le', 3, 13) ('ˈ', 98, 14) ### +### ('roller', 12839, 15) ('whites', 16, 16) ('|', 17673, 17) ('drive', 4244, 18) ('crashing', 19, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('white', 0, 0) ('kn', 1, 2) ('##uck', 2, 6) ('##le', 3, 13) ('meaning', 4, 4) ('definition', 5, 9) ### +### ('means', 10, 50) ('whites', 16, 16) ('crashing', 19, 19) ('wingspan', 36, 21) ('julian', 20, 44) ### +### ('##ང', 40, 22) ('afraid', 100, 7) ('−', 45, 24) ('##₂', 21, 72) ('mean', 65, 23) ### +### ('whitish', 17, 126) ('angrily', 37, 43) ('cyrillic', 64, 28) ('gideon', 46, 34) ### +############################################################################################################ +[2023-10-07 21:32:17,543][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:32:17,543][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:32:17,967][root][INFO] - Epoch: 7: Step: 301/1557, loss[v]=0.101827, lr=0.000013, acc@1[1]=235.5/256=0.919921875, acc@1[2]=246.5/256=0.962890625 +[2023-10-07 21:33:33,300][root][INFO] - Train batch 400 +[2023-10-07 21:33:33,301][root][INFO] - Avg. loss per last 100 batches: 0.081112 +[2023-10-07 21:33:33,990][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29155.7/29522=98.76% | mean: 0.01 | max: 5.11 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 5.87 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] where is costa mesa california on the map [SEP] ### +### [P_TEXT]: [CLS] detailed map of costa mesa and near places. welcome to the costa mesa google ### +### satellite map! this place is situated in orange county, california, united states, its geographical ### +### coordinates are 33a° 38'28 north, 117a° 55'4 west and its original name ( with diacritics ) is ### +### costa mesa. [SEP] ### +### ======================================= h_v_q | Gates: 27012 ======================================= ### +### ('mesa', 0, 1) ('costa', 1, 0) ('map', 2, 5) ('california', 3, 6) ('located', 4, 108) ### +### ('is', 5, 1489) ('maps', 6, 10) ('.', 7, 2305) ('united', 8, 430) ('familiarity', 9, 21147) ### +### ('brazil', 10, 1619) ('mapping', 11, 54) ('hollow', 12, 5123) ('situated', 13, 22) ### +### ('florida', 14, 1027) ('where', 15, 27) ('district', 16, 131) ('region', 17, 2737) ### +### ('relating', 18, 16694) ('pennsylvania', 19, 1758) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('costa', 1, 0) ('mesa', 0, 1) ('orange', 2468, 2) ('satellite', 1312, 3) ('coordinates', 13979, 4) ### +### ('map', 2, 5) ('california', 3, 6) ('geographical', 77, 7) ('welcome', 7691, 8) ('google', 8223, 9) ### +### ('maps', 6, 10) ('places', 32, 11) ('county', 73, 12) ('place', 28, 13) ('crashing', 62, 14) ### +### ('satellites', 5474, 15) ('near', 474, 16) ('##ང', 234, 17) ('ˈ', 438, 18) ('detailed', 2297, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('costa', 1, 0) ('mesa', 0, 1) ('map', 2, 5) ('california', 3, 6) ('maps', 6, 10) ### +### ('located', 4, 108) ('situated', 13, 22) ('where', 15, 27) ('mapping', 11, 54) ('places', 32, 11) ### +### ('place', 28, 13) ('geographical', 77, 7) ('north', 29, 43) ('west', 47, 21) ('crashing', 62, 14) ### +### ('district', 16, 131) ('county', 73, 12) ('united', 8, 430) ('tucson', 59, 69) ('states', 22, 174) ### +############################################################################################################ +[2023-10-07 21:33:33,991][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:33:33,991][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:33:34,407][root][INFO] - Epoch: 7: Step: 401/1557, loss[v]=0.077567, lr=0.000013, acc@1[1]=243.0/256=0.94921875, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 21:34:50,703][root][INFO] - Train batch 500 +[2023-10-07 21:34:50,703][root][INFO] - Avg. loss per last 100 batches: 0.083757 +[2023-10-07 21:34:51,386][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29163.2/29522=98.78% | mean: 0.01 | max: 4.93 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.7/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.18 | max: 6.00 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] states where civil war battles were fought [SEP] ### +### [P_TEXT]: [CLS] kentucky and west virginia, 1861a1865. a map of the parts of kentucky and west ### +### virginia, where some early battles of the american civil war were fought. the map shows terrain ### +### features, rivers, railroads, and the cities and towns of cincinnati, philippi, rich mountain, ### +### prestonburg, and mi... [SEP] ### +### ======================================= h_v_q | Gates: 27219 ======================================= ### +### ('war', 0, 4) ('civil', 1, 19) ('states', 2, 10017) ('fought', 3, 11) ('battles', 4, 5) ### +### ('where', 5, 27) ('battle', 6, 20) ('fighting', 7, 30) ('familiarity', 8, 26047) ### +### ('brazil', 9, 3487) ('virginia', 10, 7) ('state', 11, 2721) ('were', 12, 77) ### +### ('relating', 13, 15040) ('ohio', 14, 76) ('regions', 15, 329) ('played', 16, 840) ### +### ('california', 17, 5347) ('hampshire', 18, 12800) ('plural', 19, 7390) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('cincinnati', 586, 0) ('map', 2801, 1) ('kentucky', 173, 2) ('maps', 1564, 3) ('war', 0, 4) ### +### ('battles', 4, 5) ('west', 1155, 6) ('virginia', 10, 7) ('preston', 439, 8) ('##a1', 11392, 9) ### +### ('features', 338, 10) ('fought', 3, 11) ('rivers', 754, 12) ('terrain', 4887, 13) ### +### ('early', 1496, 14) ('rich', 2432, 15) ('towns', 1298, 16) ('cities', 82, 17) ### +### ('philipp', 14518, 18) ('civil', 1, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('war', 0, 4) ('civil', 1, 19) ('battles', 4, 5) ('fought', 3, 11) ('where', 5, 27) ### +### ('battle', 6, 20) ('virginia', 10, 7) ('fighting', 7, 30) ('were', 12, 77) ('ohio', 14, 76) ### +### ('fight', 24, 42) ('kentucky', 173, 2) ('wars', 36, 51) ('states', 2, 10017) ('military', 23, 104) ### +### ('cities', 82, 17) ('##ང', 88, 31) ('ˈ', 104, 29) ('julian', 46, 81) ('presenter', 45, 87) ### +############################################################################################################ +[2023-10-07 21:34:51,386][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:34:51,386][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:34:51,805][root][INFO] - Epoch: 7: Step: 501/1557, loss[v]=0.050206, lr=0.000013, acc@1[1]=243.5/256=0.951171875, acc@1[2]=254.5/256=0.994140625 +[2023-10-07 21:36:10,465][root][INFO] - Train batch 600 +[2023-10-07 21:36:10,466][root][INFO] - Avg. loss per last 100 batches: 0.077697 +[2023-10-07 21:36:11,191][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29098.1/29522=98.56% | mean: 0.01 | max: 5.07 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 5.95 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what falls under afmc [SEP] ### +### [P_TEXT]: [CLS] edwards afb falls under afmc, which employs more than 80, 000 people and has a ### +### budget of $ 60 billion. pawlikowski attached the new unit award streamer to the tps guidon after ### +### the award citation was read. [SEP] ### +### ======================================= h_v_q | Gates: 27796 ======================================= ### +### ('af', 0, 4) ('##mc', 1, 10) ('falls', 2, 49) ('under', 3, 28) ('familiarity', 4, 26758) ### +### ('falling', 5, 99) ('relating', 6, 19789) ('fell', 7, 279) ('beneath', 8, 1347) ### +### ('stands', 9, 10873) ('.', 10, 11705) ('hospital', 11, 3878) ('something', 12, 13005) ### +### ('fall', 13, 140) ('or', 14, 20789) ('refers', 15, 17229) ('what', 16, 8003) ('academy', 17, 1837) ### +### ('plural', 18, 13430) ('figures', 19, 5962) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('edwards', 951, 0) ('afb', 3219, 1) ('guido', 11910, 2) ('##owski', 24462, 3) ('af', 0, 4) ### +### ('paw', 10636, 5) ('award', 3796, 6) ('employs', 3641, 7) ('##ps', 13780, 8) ('budget', 181, 9) ### +### ('##mc', 1, 10) ('awards', 1279, 11) ('citation', 3746, 12) ('unit', 388, 13) ('stream', 447, 14) ### +### ('ˈ', 202, 15) ('attached', 2915, 16) ('billion', 15769, 17) ('##ང', 61, 18) ('crashing', 114, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('af', 0, 4) ('##mc', 1, 10) ('falls', 2, 49) ('under', 3, 28) ('falling', 5, 99) ('fall', 13, 140) ### +### ('gideon', 35, 27) ('##ང', 61, 18) ('fell', 7, 279) ('budget', 181, 9) ('edwards', 951, 0) ### +### ('stumbled', 60, 39) ('crashing', 114, 19) ('##α', 81, 31) ('beside', 28, 83) ('anger', 31, 79) ### +### ('ˈ', 202, 15) ('julian', 79, 45) ('hugh', 63, 58) ('−', 145, 24) ### +############################################################################################################ +[2023-10-07 21:36:11,192][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:36:11,192][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:36:11,613][root][INFO] - Epoch: 7: Step: 601/1557, loss[v]=0.107119, lr=0.000013, acc@1[1]=244.0/256=0.953125, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 21:37:29,334][root][INFO] - Train batch 700 +[2023-10-07 21:37:29,335][root][INFO] - Avg. loss per last 100 batches: 0.076680 +[2023-10-07 21:37:30,063][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29249.9/29522=99.08% | mean: 0.01 | max: 5.50 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.31 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] anthill meaning [SEP] ### +### [P_TEXT]: [CLS] definition of anthill. : a mound of debris thrown up by ants or termites in digging ### +### their nest. [SEP] ### +### ======================================= h_v_q | Gates: 27316 ======================================= ### +### ('##hill', 0, 2) ('ant', 1, 0) ('meaning', 2, 13) ('noun', 3, 23231) ('hill', 4, 21) ### +### ('definition', 5, 10) ('means', 6, 279) ('ants', 7, 1) ('sense', 8, 8548) ('familiarity', 9, 26991) ### +### ('.', 10, 3566) ('something', 11, 2742) ('meant', 12, 147) ('relating', 13, 19937) ### +### ('defined', 14, 39) ('refers', 15, 9326) ('latin', 16, 10129) ('##field', 17, 1302) ### +### ('plural', 18, 10179) ('language', 19, 15147) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('ant', 1, 0) ('ants', 7, 1) ('##hill', 0, 2) ('nest', 805, 3) ('debris', 4446, 4) ### +### ('thrown', 1770, 5) ('define', 7360, 6) ('definitions', 1001, 7) ('mound', 225, 8) ### +### ('digging', 6982, 9) ('definition', 5, 10) ('##ites', 21307, 11) ('mounds', 11396, 12) ### +### ('meaning', 2, 13) ('ˈ', 195, 14) ('crashing', 156, 15) ('dig', 10453, 16) ('##ང', 93, 17) ### +### ('nesting', 10703, 18) ('nests', 10032, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ant', 1, 0) ('##hill', 0, 2) ('meaning', 2, 13) ('definition', 5, 10) ('ants', 7, 1) ### +### ('hill', 4, 21) ('means', 6, 279) ('defined', 14, 39) ('term', 22, 27) ('mean', 21, 70) ### +### ('##α', 41, 25) ('meant', 12, 147) ('mound', 225, 8) ('##ང', 93, 17) ('presenter', 49, 37) ### +### ('−', 97, 24) ('lama', 27, 79) ('crashing', 156, 15) ('wingspan', 95, 31) ('gideon', 94, 33) ### +############################################################################################################ +[2023-10-07 21:37:30,063][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:37:30,063][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:37:30,473][root][INFO] - Epoch: 7: Step: 701/1557, loss[v]=0.059735, lr=0.000013, acc@1[1]=240.5/256=0.939453125, acc@1[2]=247.5/256=0.966796875 +[2023-10-07 21:38:47,233][root][INFO] - Train batch 800 +[2023-10-07 21:38:47,233][root][INFO] - Avg. loss per last 100 batches: 0.082997 +[2023-10-07 21:38:47,949][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29126.0/29522=98.66% | mean: 0.01 | max: 5.21 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.46 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a vellum [SEP] ### +### [P_TEXT]: [CLS] vellum ( derived from the latin word vitulinum meaning made from calf, leading to ### +### old french va©lin, calfskin ) often refers to a parchment made from calf skin, as opposed to that ### +### from other animals. it is prepared for writing or printing on, to produce single pages, scrolls, ### +### codices or books. [SEP] ### +### ======================================= h_v_q | Gates: 27675 ======================================= ### +### ('##llum', 0, 1) ('ve', 1, 6) ('definition', 2, 21) ('is', 3, 679) ('noun', 4, 13876) ### +### ('encompasses', 5, 17) ('familiarity', 6, 27200) ('or', 7, 17137) ('something', 8, 2936) ### +### ('a', 9, 13163) ('.', 10, 8910) ('plural', 11, 2776) ('relating', 12, 6184) (';', 13, 5840) ### +### ('refers', 14, 63) ('presenter', 15, 64) ('lama', 16, 100) ('baker', 17, 1815) ('va', 18, 32) ### +### ('val', 19, 296) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('calf', 20798, 0) ('##llum', 0, 1) ('parchment', 3618, 2) ('calves', 10277, 3) ('##lin', 2685, 4) ### +### ('vi', 22, 5) ('ve', 1, 6) ('##skin', 18350, 7) ('latin', 142, 8) ('##ices', 26763, 9) ### +### ('made', 1294, 10) ('meaning', 507, 11) ('skin', 1715, 12) ('animals', 3381, 13) ('cod', 7468, 14) ### +### ('french', 1710, 15) ('define', 8752, 16) ('encompasses', 5, 17) ('derived', 1309, 18) ### +### ('##tu', 13819, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##llum', 0, 1) ('ve', 1, 6) ('definition', 2, 21) ('encompasses', 5, 17) ('vi', 22, 5) ### +### ('is', 3, 679) ('va', 18, 32) ('refers', 14, 63) ('presenter', 15, 64) ('−', 41, 28) ### +### ('lama', 16, 100) ('printing', 54, 24) ('latin', 142, 8) ('crashing', 55, 26) ('##ང', 64, 31) ### +### ('##α', 94, 29) ('screenwriter', 48, 55) ('what', 37, 73) ('##ο', 185, 20) ('writing', 170, 30) ### +############################################################################################################ +[2023-10-07 21:38:47,950][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:38:47,950][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:38:48,354][root][INFO] - Epoch: 7: Step: 801/1557, loss[v]=0.091771, lr=0.000013, acc@1[1]=240.0/256=0.9375, acc@1[2]=253.5/256=0.990234375 +[2023-10-07 21:40:05,182][root][INFO] - Train batch 900 +[2023-10-07 21:40:05,183][root][INFO] - Avg. loss per last 100 batches: 1.770160 +[2023-10-07 21:40:05,905][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29161.6/29522=98.78% | mean: 0.01 | max: 5.63 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.17 | max: 6.19 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] blue boiler peas where to buy [SEP] ### +### [P_TEXT]: [CLS] the difference between the dried peas is that it's a whole pea and it has its skin ### +### on, where, say, a green split pea is obviously split and the skin's been taken off, he says. he ### +### says drought and quality issues with mainland peas has also forced the company to buy from farms in ### +### new zealand and tasmania. coles has recently delisted the company's dried peas but they can still ### +### be found in independent supermarkets such as iga or franklins, hammon says. staple winter ### +### ingredient is in short supply, writes leesha mckenny. they go by a few names - blue boilers, green ### +### field peas or just dried peas - but they all have one thing in common. they are scarce in nsw. just ### +### as the season for mushy peas - popularly served as pie floaters - comes into its own, many shop ### +### shelves are bare. [SEP] ### +### ======================================= h_v_q | Gates: 27471 ======================================= ### +### ('boiler', 0, 134) ('peas', 1, 0) ('blue', 2, 127) ('buy', 3, 53) ('where', 4, 163) ('.', 5, 5100) ### +### ('familiarity', 6, 27171) ('buying', 7, 165) ('purchase', 8, 453) ('to', 9, 9668) ### +### ('located', 10, 584) ('bought', 11, 117) ('relating', 12, 19061) ('place', 13, 5164) ### +### ('minnesota', 14, 7925) ('brazil', 15, 8369) ('places', 16, 1383) ('hampshire', 17, 14422) ### +### ('situated', 18, 3587) ('onto', 19, 66) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('peas', 1, 0) ('mainland', 10979, 1) ('cole', 1344, 2) ('nsw', 17501, 3) ('split', 2440, 4) ### +### ('float', 2492, 5) ('pea', 416, 6) ('winter', 10856, 7) ('tasmania', 11123, 8) ### +### ('drought', 17636, 9) ('bare', 132, 10) ('pie', 809, 11) ('zealand', 3856, 12) ('lee', 1611, 13) ### +### ('shelves', 8824, 14) ('ham', 10381, 15) ('mu', 4661, 16) ('crashing', 373, 17) ### +### ('supermarkets', 14816, 18) ('ingredient', 19407, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('peas', 1, 0) ('boiler', 0, 134) ('blue', 2, 127) ('buy', 3, 53) ('where', 4, 163) ### +### ('buying', 7, 165) ('bought', 11, 117) ('purchase', 8, 453) ('onto', 19, 66) ('green', 50, 25) ### +### ('located', 10, 584) ('bare', 132, 10) ('−', 96, 20) ('somewhere', 31, 131) ('boilers', 65, 57) ### +### ('presenter', 47, 85) ('australia', 38, 115) ('.', 5, 5100) ('shop', 139, 27) ('simon', 61, 103) ### +############################################################################################################ +[2023-10-07 21:40:05,905][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:40:05,905][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:40:06,325][root][INFO] - Epoch: 7: Step: 901/1557, loss[v]=0.138403, lr=0.000013, acc@1[1]=233.5/256=0.912109375, acc@1[2]=247.5/256=0.966796875 +[2023-10-07 21:41:22,751][root][INFO] - Train batch 1000 +[2023-10-07 21:41:22,752][root][INFO] - Avg. loss per last 100 batches: 0.896467 +[2023-10-07 21:41:23,476][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29078.9/29522=98.50% | mean: 0.01 | max: 5.17 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 5.95 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] land ordinance definition [SEP] ### +### [P_TEXT]: [CLS] land ordinance of 1785. the land ordinance of 1785 was adopted by the united states ### +### congress of the confederation on may 20, 1785. it set up a standardized system whereby settlers ### +### could purchase title to farmland in the undeveloped west. congress at the time did not have the ### +### power to raise revenue by direct taxation, so land sales provided an important revenue stream. ### +### [SEP] ### +### ======================================= h_v_q | Gates: 25436 ======================================= ### +### ('ordinance', 0, 0) ('land', 1, 1) ('definition', 2, 109) ('defined', 3, 1111) ('.', 4, 7106) ### +### ('noun', 5, 28618) ('act', 6, 819) ('law', 7, 416) ('lands', 8, 26) ('property', 9, 219) ### +### ('familiarity', 10, 28075) ('or', 11, 11821) ('soil', 12, 99) ('##land', 13, 75) ### +### ('something', 14, 3050) (';', 15, 4448) ('relating', 16, 17999) ('sense', 17, 17232) ### +### ('term', 18, 11717) ('temple', 19, 1889) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('ordinance', 0, 0) ('land', 1, 1) ('confederation', 13318, 2) ('farmland', 1264, 3) ### +### ('adopted', 1639, 4) ('standardized', 2384, 5) ('revenue', 8137, 6) ('taxation', 11368, 7) ### +### ('undeveloped', 17478, 8) ('1985', 6600, 9) ('settlers', 7144, 10) ('congress', 8038, 11) ### +### ('direct', 5156, 12) ('stream', 313, 13) ('##ο', 786, 14) ('west', 1762, 15) ('1785', 13672, 16) ### +### ('crashing', 180, 17) ('title', 1269, 18) ('raise', 8095, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ordinance', 0, 0) ('land', 1, 1) ('definition', 2, 109) ('lands', 8, 26) ('soil', 12, 99) ### +### ('##land', 13, 75) ('property', 9, 219) ('law', 7, 416) ('defined', 3, 1111) ('act', 6, 819) ### +### ('legislation', 30, 66) ('crashing', 180, 17) ('julian', 91, 59) ('stream', 313, 13) ### +### ('rural', 155, 35) ('.', 4, 7106) ('−', 245, 27) ('agriculture', 177, 41) ('definitions', 48, 196) ### +### ('ward', 22, 519) ### +############################################################################################################ +[2023-10-07 21:41:23,476][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:41:23,477][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:41:23,901][root][INFO] - Epoch: 7: Step: 1001/1557, loss[v]=0.116730, lr=0.000013, acc@1[1]=239.0/256=0.93359375, acc@1[2]=245.0/256=0.95703125 +[2023-10-07 21:42:39,794][root][INFO] - Train batch 1100 +[2023-10-07 21:42:39,794][root][INFO] - Avg. loss per last 100 batches: 2.095675 +[2023-10-07 21:42:40,476][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29158.7/29522=98.77% | mean: 0.01 | max: 5.32 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.16 | max: 5.95 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the population in bronx, ny [SEP] ### +### [P_TEXT]: [CLS] over 50, 000 more people are employed in the bronx this year than five years ago, ### +### according to the bureau of labor statistics. in the 1970s alone, the bronx population declined to ### +### 1. 169 million, from 1. 472 million, largely as a result of white flight. from 2007 to 2014, 30, ### +### 000 more whites left the bronx. but from 2014 to 2015, fewer than 1, 000 did. [SEP] ### +### ======================================= h_v_q | Gates: 26737 ======================================= ### +### ('bronx', 0, 0) ('population', 1, 2) ('york', 2, 336) ('ny', 3, 45) ('manhattan', 4, 95) ### +### ('familiarity', 5, 24910) ('.', 6, 6765) ('is', 7, 4042) ('sad', 8, 3638) ('brooklyn', 9, 28) ### +### ('populations', 10, 25) ('washington', 11, 7766) ('relating', 12, 23048) ('total', 13, 1510) ### +### ('number', 14, 36) ('volume', 15, 7527) ('people', 16, 20) ('ethnic', 17, 1157) ('urban', 18, 1099) ### +### ('plural', 19, 8571) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('bronx', 0, 0) ('whites', 13540, 1) ('population', 1, 2) ('statistics', 880, 3) ### +### ('employed', 2734, 4) ('million', 279, 5) ('leaving', 85, 6) ('##ང', 302, 7) ('employs', 3410, 8) ### +### ('2015', 1185, 9) ('2014', 4358, 10) ('crashing', 216, 11) ('labor', 4797, 12) ('300', 61, 13) ### +### ('alone', 1814, 14) ('30', 175, 15) ('ˈ', 1210, 16) ('flight', 3450, 17) ('##ο', 536, 18) ### +### ('white', 1807, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('bronx', 0, 0) ('population', 1, 2) ('ny', 3, 45) ('york', 2, 336) ('manhattan', 4, 95) ### +### ('brooklyn', 9, 28) ('populations', 10, 25) ('people', 16, 20) ('number', 14, 36) ('300', 61, 13) ### +### ('leaving', 85, 6) ('julian', 32, 71) ('stark', 67, 50) ('residents', 26, 128) ('30', 175, 15) ### +### ('crashing', 216, 11) ('million', 279, 5) ('−', 174, 22) ('∈', 56, 100) ('##ང', 302, 7) ### +############################################################################################################ +[2023-10-07 21:42:40,476][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:42:40,476][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:42:40,897][root][INFO] - Epoch: 7: Step: 1101/1557, loss[v]=0.097942, lr=0.000013, acc@1[1]=238.5/256=0.931640625, acc@1[2]=247.0/256=0.96484375 +[2023-10-07 21:43:57,532][root][INFO] - Train batch 1200 +[2023-10-07 21:43:57,533][root][INFO] - Avg. loss per last 100 batches: 0.077919 +[2023-10-07 21:43:58,255][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29176.5/29522=98.83% | mean: 0.01 | max: 5.15 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.16 | max: 6.17 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] internal temp for swordfish steaks [SEP] ### +### [P_TEXT]: [CLS] making the world better, one answer at a time. the best way to cook swordfish is to ### +### dredge it in lime juice, coat it with blackened fish seasoning, and sear it in a pan on high heat ### +### for a minute on both sides. then transfer it to an oven at 350 degrees until the internal ### +### temperature reaches 125 degrees ( medium rare ). ccording to new usda guidelines established in ### +### 2011, pork is fully cooked when its internal temperature is 145 degrees fahrenheit. prior to this ### +### revision, the cooking rec a¦ ommendation for fully cooked pork was that the internal temperature ### +### should be 160 degree fahrenheit. [SEP] ### +### ======================================= h_v_q | Gates: 28335 ======================================= ### +### ('##fish', 0, 0) ('steak', 1, 2028) ('sword', 2, 1) ('internal', 3, 23) ('te', 4, 2419) ### +### ('##°', 5, 142) ('familiarity', 6, 26524) ('temperature', 7, 2) ('.', 8, 15723) ('fish', 9, 4) ### +### ('inches', 10, 1779) ('for', 11, 14360) ('altitude', 12, 227) ('december', 13, 14989) ### +### ('##s', 14, 21980) ('february', 15, 9372) ('30', 16, 104) ('november', 17, 12044) ### +### ('volume', 18, 10598) ('speed', 19, 327) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('##fish', 0, 0) ('sword', 2, 1) ('temperature', 7, 2) ('##¦', 29334, 3) ('fish', 9, 4) ### +### ('pork', 18365, 5) ('lime', 14804, 6) ('transfer', 8167, 7) ('blackened', 12371, 8) ### +### ('crashing', 210, 9) ('cc', 10928, 10) ('cooked', 2144, 11) ('juice', 10435, 12) ### +### ('##ord', 12608, 13) ('##ο', 425, 14) ('hating', 119, 15) ('−', 29, 16) ('ˈ', 164, 17) ### +### ('rare', 2787, 18) ('swords', 51, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##fish', 0, 0) ('sword', 2, 1) ('internal', 3, 23) ('temperature', 7, 2) ('fish', 9, 4) ### +### ('##°', 5, 142) ('steak', 1, 2028) ('te', 4, 2419) ('−', 29, 16) ('altitude', 12, 227) ### +### ('swords', 51, 19) ('30', 16, 104) ('julian', 65, 47) ('hating', 119, 15) ('minutes', 60, 68) ### +### ('minute', 43, 103) ('dagger', 81, 49) ('screenwriter', 96, 26) ('stark', 58, 87) ('anger', 64, 72) ### +############################################################################################################ +[2023-10-07 21:43:58,256][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:43:58,256][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:43:58,677][root][INFO] - Epoch: 7: Step: 1201/1557, loss[v]=0.045030, lr=0.000013, acc@1[1]=246.0/256=0.9609375, acc@1[2]=255.5/256=0.998046875 +[2023-10-07 21:45:15,000][root][INFO] - Train batch 1300 +[2023-10-07 21:45:15,001][root][INFO] - Avg. loss per last 100 batches: 0.078277 +[2023-10-07 21:45:15,692][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29210.2/29522=98.94% | mean: 0.01 | max: 5.42 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.01 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is american independence day [SEP] ### +### [P_TEXT]: [CLS] in 1964, the name of the july 4 holiday was changed to republic day. in rwanda, ### +### july 4 is an official holiday known as liberation day, commemorating the end of the 1994 rwandan ### +### genocide in which the u. s. government also played a role. denmark also celebrates american ### +### independence on july 4. he second day of july, 1776, will be the most memorable epoch in the ### +### history of america. i am apt to believe that it will be celebrated by succeeding generations as the ### +### great anniversary festival. it ought to be commemorated as the day of deliverance, by solemn acts ### +### of devotion to god almighty. [SEP] ### +### ======================================= h_v_q | Gates: 27037 ======================================= ### +### ('independence', 0, 13) ('american', 1, 29) ('day', 2, 5) ('america', 3, 76) ('americans', 4, 106) ### +### ('familiarity', 5, 25682) ('is', 6, 856) ('definition', 7, 220) ('encompasses', 8, 64) ### +### ('date', 9, 100) ('relating', 10, 20911) ('.', 11, 7134) ('refers', 12, 4629) ('sunday', 13, 1535) ### +### ('september', 14, 457) ('square', 15, 10556) ('julian', 16, 52) ('days', 17, 99) ### +### ('independent', 18, 684) ('freedom', 19, 95) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('anniversary', 1600, 0) ('rwanda', 23372, 1) ('liberation', 351, 2) ('genocide', 15528, 3) ### +### ('denmark', 2212, 4) ('day', 2, 5) ('holiday', 657, 6) ('july', 49, 7) ('4', 19156, 8) ### +### ('epoch', 3176, 9) ('commemorated', 211, 10) ('deliver', 15243, 11) ('celebrated', 119, 12) ### +### ('independence', 0, 13) ('−', 36, 14) ('festival', 169, 15) ('republic', 635, 16) ### +### ('crashing', 177, 17) ('##ο', 168, 18) ('memorable', 10190, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('independence', 0, 13) ('day', 2, 5) ('american', 1, 29) ('america', 3, 76) ('americans', 4, 106) ### +### ('encompasses', 8, 64) ('date', 9, 100) ('july', 49, 7) ('−', 36, 14) ('definition', 7, 220) ### +### ('julian', 16, 52) ('is', 6, 856) ('hating', 40, 32) ('freedom', 19, 95) ('days', 17, 99) ### +### ('celebrated', 119, 12) ('june', 53, 26) ('presenter', 42, 46) ('commemorated', 211, 10) ### +### ('liberation', 351, 2) ### +############################################################################################################ +[2023-10-07 21:45:15,692][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:45:15,692][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:45:16,114][root][INFO] - Epoch: 7: Step: 1301/1557, loss[v]=0.048644, lr=0.000013, acc@1[1]=245.5/256=0.958984375, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 21:46:32,465][root][INFO] - Train batch 1400 +[2023-10-07 21:46:32,465][root][INFO] - Avg. loss per last 100 batches: 0.081328 +[2023-10-07 21:46:33,187][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29125.7/29522=98.66% | mean: 0.01 | max: 5.05 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.17 | max: 6.07 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] where is the catalytic converter located? [SEP] ### +### [P_TEXT]: [CLS] in a typical car, the catalytic converter is part of the exhaust system and under ### +### the car, between the engine and the muffler. often manufacturers locate the catalytic converter ### +### under the passenger seat. on long trips, passengers sometimes feel the heat it produces at their ### +### feet. continue reading. [SEP] ### +### ======================================= h_v_q | Gates: 27741 ======================================= ### +### ('catalytic', 0, 0) ('located', 1, 392) ('convert', 2, 2) ('##er', 3, 13) ('familiarity', 4, 27501) ### +### ('.', 5, 17226) ('situated', 6, 1731) ('united', 7, 23079) ('hampshire', 8, 19293) ('is', 9, 1040) ### +### ('encompasses', 10, 5) ('found', 11, 722) ('england', 12, 15097) ('where', 13, 213) ### +### ('##ers', 14, 139) ('headquartered', 15, 3220) ('conversion', 16, 48) ('south', 17, 16644) ### +### ('india', 18, 1409) ('founded', 19, 7253) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('catalytic', 0, 0) ('exhaust', 13136, 1) ('convert', 2, 2) ('##ffle', 18427, 3) ('car', 2080, 4) ### +### ('encompasses', 10, 5) ('trips', 12919, 6) ('passengers', 2992, 7) ('##ο', 132, 8) ### +### ('typical', 4558, 9) ('passenger', 137, 10) ('ˈ', 115, 11) ('crashing', 206, 12) ('##er', 3, 13) ### +### ('##ང', 159, 14) ('heat', 2394, 15) ('mu', 5137, 16) ('seat', 960, 17) ('cars', 6977, 18) ### +### ('engine', 414, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('catalytic', 0, 0) ('convert', 2, 2) ('##er', 3, 13) ('located', 1, 392) ('encompasses', 10, 5) ### +### ('locate', 21, 31) ('conversion', 16, 48) ('##ers', 14, 139) ('##₂', 37, 41) ('where', 13, 213) ### +### ('##α', 65, 20) ('catalyst', 43, 74) ('hugh', 51, 58) ('converted', 29, 131) ('−', 77, 26) ### +### ('ˈ', 115, 11) ('##ο', 132, 8) ('julian', 46, 88) ('screenwriter', 103, 24) ('passenger', 137, 10) ### +############################################################################################################ +[2023-10-07 21:46:33,187][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:46:33,187][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:46:33,615][root][INFO] - Epoch: 7: Step: 1401/1557, loss[v]=0.089824, lr=0.000013, acc@1[1]=239.5/256=0.935546875, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 21:47:50,230][root][INFO] - Train batch 1500 +[2023-10-07 21:47:50,231][root][INFO] - Avg. loss per last 100 batches: 2.070949 +[2023-10-07 21:47:50,925][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29086.3/29522=98.52% | mean: 0.01 | max: 5.38 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.17 | max: 5.91 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] when do blue n gold baby macaws wean [SEP] ### +### [P_TEXT]: [CLS] weaning is the period of time taken for the baby macaws to start eating on their ### +### own, after being fed by us or by their parents. this is a very gradual process, which can take ### +### anything from a few weeks up to months depending on the baby and the species. the blue - throated ### +### macaws generally wean the earliest from around 16 - 18 weeks. the blue and gold and scarlet macaws ### +### generally wean around 20 weeks. eaning is the period of time taken for the baby macaws to start ### +### eating on their own, after being fed by us or by their parents. [SEP] ### +### ======================================= h_v_q | Gates: 28030 ======================================= ### +### ('blue', 0, 8) ('gold', 1, 34) ('baby', 2, 1) ('mac', 3, 9) ('##aw', 4, 51) ('we', 5, 2) ### +### ('##an', 6, 185) ('n', 7, 22960) ('familiarity', 8, 25872) ('when', 9, 79) ('timing', 10, 129) ### +### ('do', 11, 2335) ('2017', 12, 20587) ('sunday', 13, 15471) ('early', 14, 71) ('.', 15, 16865) ### +### ('september', 16, 2836) ('november', 17, 10139) ('##स', 18, 143) ('october', 19, 4029) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('depending', 9404, 0) ('baby', 2, 1) ('we', 5, 2) ('eating', 14238, 3) ('weeks', 43, 4) ### +### ('##ng', 17962, 5) ('long', 3632, 6) ('period', 24, 7) ('blue', 0, 8) ('mac', 3, 9) ### +### ('ea', 15460, 10) ('throat', 11424, 11) ('##α', 147, 12) ('crashing', 209, 13) ('babies', 36, 14) ### +### ('##ο', 186, 15) ('ˈ', 493, 16) ('scarlet', 2546, 17) ('eat', 368, 18) ('time', 29, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('blue', 0, 8) ('baby', 2, 1) ('gold', 1, 34) ('mac', 3, 9) ('we', 5, 2) ('##aw', 4, 51) ### +### ('when', 9, 79) ('##an', 6, 185) ('timing', 10, 129) ('period', 24, 7) ('early', 14, 71) ### +### ('age', 23, 27) ('time', 29, 19) ('weeks', 43, 4) ('babies', 36, 14) ('after', 22, 67) ### +### ('late', 26, 102) ('##स', 18, 143) ('##₂', 70, 47) ('purple', 37, 181) ### +############################################################################################################ +[2023-10-07 21:47:50,926][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:47:50,926][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:47:51,347][root][INFO] - Epoch: 7: Step: 1501/1557, loss[v]=0.155199, lr=0.000013, acc@1[1]=239.5/256=0.935546875, acc@1[2]=250.5/256=0.978515625 +[2023-10-07 21:48:34,630][root][INFO] - rank=3; last iteration 1557 +[2023-10-07 21:48:34,630][root][INFO] - rank=1; last iteration 1557 +[2023-10-07 21:48:34,631][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 21:48:34,631][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 21:48:34,631][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-07 21:48:34,631][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-07 21:48:34,631][root][INFO] - rank=2; last iteration 1557 +[2023-10-07 21:48:34,631][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 21:48:34,631][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-07 21:48:34,636][root][INFO] - rank=0; last iteration 1557 +[2023-10-07 21:48:34,637][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 21:48:34,637][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-07 21:48:34,638][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 21:48:34,639][root][INFO] - Epoch finished on 3 +[2023-10-07 21:48:34,639][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 21:48:34,639][root][INFO] - Epoch finished on 1 +[2023-10-07 21:48:34,639][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 21:48:34,640][root][INFO] - Epoch finished on 2 +[2023-10-07 21:48:34,643][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 21:48:34,643][root][INFO] - Epoch finished on 0 +[2023-10-07 21:49:34,441][root][INFO] - Saved checkpoint at ./vdr_7 +[2023-10-07 21:49:34,465][root][INFO] - Av Loss per epoch=0.498163 +[2023-10-07 21:49:34,466][root][INFO] - epoch total (1) correct predictions=376663 +[2023-10-07 21:49:34,466][root][INFO] - epoch total (2) correct predictions=389067 +[2023-10-07 21:49:34,470][root][INFO] - ***** Epoch 8 ***** +[2023-10-07 21:49:34,477][root][INFO] - rank=3; Iteration start +[2023-10-07 21:49:34,477][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 21:49:34,477][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 21:49:34,443][root][INFO] - Saved checkpoint at ./vdr_7 +[2023-10-07 21:49:34,478][root][INFO] - Av Loss per epoch=0.498163 +[2023-10-07 21:49:34,478][root][INFO] - epoch total (1) correct predictions=376663 +[2023-10-07 21:49:34,478][root][INFO] - epoch total (2) correct predictions=389067 +[2023-10-07 21:49:34,479][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-07 21:49:34,479][root][INFO] - Saved checkpoint at ./vdr_7 +[2023-10-07 21:49:34,481][root][INFO] - Av Loss per epoch=0.498163 +[2023-10-07 21:49:34,481][root][INFO] - epoch total (1) correct predictions=376663 +[2023-10-07 21:49:34,481][root][INFO] - epoch total (2) correct predictions=389067 +[2023-10-07 21:49:34,484][root][INFO] - ***** Epoch 8 ***** +[2023-10-07 21:49:34,485][root][INFO] - ***** Epoch 8 ***** +[2023-10-07 21:49:34,443][root][INFO] - Saved checkpoint at ./vdr_7 +[2023-10-07 21:49:34,490][root][INFO] - Av Loss per epoch=0.498163 +[2023-10-07 21:49:34,490][root][INFO] - epoch total (1) correct predictions=376663 +[2023-10-07 21:49:34,490][root][INFO] - epoch total (2) correct predictions=389067 +[2023-10-07 21:49:34,491][root][INFO] - rank=1; Iteration start +[2023-10-07 21:49:34,492][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 21:49:34,492][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 21:49:34,492][root][INFO] - rank=2; Iteration start +[2023-10-07 21:49:34,492][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 21:49:34,492][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 21:49:34,494][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-07 21:49:34,494][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-07 21:49:34,495][root][INFO] - ***** Epoch 8 ***** +[2023-10-07 21:49:34,502][root][INFO] - rank=0; Iteration start +[2023-10-07 21:49:34,503][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 21:49:34,503][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 21:49:34,505][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-07 21:49:35,478][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29173.8/29522=98.82% | mean: 0.01 | max: 5.20 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.17 | max: 5.94 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what county whitehall, ny [SEP] ### +### [P_TEXT]: [CLS] whitehall, ny. sponsored topics. whitehall, new york is the name of a village and a ### +### town in washington county, new york. both are on the vermont border at the south end of lake ### +### champlain and lie between lake champlain and lake george. [SEP] ### +### ======================================= h_v_q | Gates: 28066 ======================================= ### +### ('whitehall', 0, 0) ('county', 1, 12) ('ny', 2, 3) ('york', 3, 4) ('.', 4, 883) ### +### ('manhattan', 5, 120) ('familiarity', 6, 25963) ('mirror', 7, 202) ('counties', 8, 35) ### +### ('washington', 9, 11) ('parish', 10, 280) ('district', 11, 95) ('pop', 12, 466) ### +### ('virginia', 13, 1773) ('whereas', 14, 322) ('brooklyn', 15, 128) ('wisconsin', 16, 1729) ### +### ('society', 17, 4554) ('london', 18, 1544) ('something', 19, 5738) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('whitehall', 0, 0) ('champ', 17663, 1) ('vermont', 82, 2) ('ny', 2, 3) ('york', 3, 4) ### +### ('sponsored', 5795, 5) ('village', 617, 6) ('border', 3907, 7) ('george', 605, 8) ### +### ('##lain', 25212, 9) ('where', 1074, 10) ('washington', 9, 11) ('county', 1, 12) ('lake', 1075, 13) ### +### ('encompasses', 1940, 14) ('crashing', 498, 15) ('topics', 8211, 16) ('town', 73, 17) ### +### ('villages', 1490, 18) ('stab', 1391, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('whitehall', 0, 0) ('county', 1, 12) ('ny', 2, 3) ('york', 3, 4) ('washington', 9, 11) ### +### ('manhattan', 5, 120) ('counties', 8, 35) ('vermont', 82, 2) ('district', 11, 95) ('town', 73, 17) ### +### ('russell', 21, 83) ('mirror', 7, 202) ('.', 4, 883) ('brooklyn', 15, 128) ('pond', 53, 52) ### +### ('bureaucracy', 22, 117) ('rural', 20, 141) ('connecticut', 38, 106) ('parish', 10, 280) ### +### ('village', 617, 6) ### +############################################################################################################ +[2023-10-07 21:49:35,479][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:49:35,479][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:49:35,869][root][INFO] - Epoch: 8: Step: 1/1557, loss[v]=0.094221, lr=0.000013, acc@1[1]=240.5/256=0.939453125, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 21:50:52,756][root][INFO] - Train batch 100 +[2023-10-07 21:50:52,758][root][INFO] - Avg. loss per last 100 batches: 0.075732 +[2023-10-07 21:50:53,460][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29116.5/29522=98.63% | mean: 0.01 | max: 5.23 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.14 | max: 6.00 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] original equipment manufacturer means [SEP] ### +### [P_TEXT]: [CLS] the term oem stands for original equipment manufacturer. what does that really ### +### mean? it means that the part was made by a company that is a subcontractor to a vehicle ### +### manufacturer. it does not mean the part was made by the manufacturer. some examples ; 1 most fuel ### +### parts on a vw are made by bosch. 2 this means that bosch is the oem for vw regardless of where you ### +### buy the bosch part it is still oem. t means that the part was made by a company that is a ### +### subcontractor to a vehicle manufacturer. it does not mean the part was made by the manufacturer. ### +### some examples ; 1 most fuel parts on a vw are made by bosch. 2 this means that bosch is the oem for ### +### vw regardless of where you buy the bosch part it is still oem. [SEP] ### +### ======================================= h_v_q | Gates: 26243 ======================================= ### +### ('equipment', 0, 7) ('original', 1, 10) ('manufacturer', 2, 8) ('means', 3, 5) ### +### ('originally', 4, 162) ('noun', 5, 20591) ('definition', 6, 15) ('meaning', 7, 9) ('gear', 8, 99) ### +### ('company', 9, 32) ('familiarity', 10, 25463) ('.', 11, 11925) ('machinery', 12, 101) ### +### ('latin', 13, 5115) ('manufacturers', 14, 34) ('firm', 15, 698) ('refers', 16, 955) ### +### ('mean', 17, 13) ('something', 18, 1296) ('aircraft', 19, 1429) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('bosch', 22945, 0) ('o', 6511, 1) ('##em', 17560, 2) ('part', 517, 3) ('made', 1291, 4) ### +### ('means', 3, 5) ('parts', 2550, 6) ('equipment', 0, 7) ('manufacturer', 2, 8) ('meaning', 7, 9) ### +### ('original', 1, 10) ('define', 7061, 11) ('vehicle', 372, 12) ('mean', 17, 13) ('ˈ', 709, 14) ### +### ('definition', 6, 15) ('crashing', 407, 16) ('##ο', 662, 17) ('sub', 4861, 18) ('fuel', 3846, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('equipment', 0, 7) ('original', 1, 10) ('manufacturer', 2, 8) ('means', 3, 5) ('meaning', 7, 9) ### +### ('definition', 6, 15) ('originally', 4, 162) ('company', 9, 32) ('mean', 17, 13) ### +### ('manufacturers', 14, 34) ('gear', 8, 99) ('machinery', 12, 101) ('term', 29, 44) ('meant', 31, 45) ### +### ('manufactured', 39, 38) ('encompasses', 43, 68) ('originated', 33, 131) ('manufacture', 61, 67) ### +### ('factory', 21, 335) ('manufacturing', 27, 253) ### +############################################################################################################ +[2023-10-07 21:50:53,461][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:50:53,461][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:50:53,883][root][INFO] - Epoch: 8: Step: 101/1557, loss[v]=0.068991, lr=0.000013, acc@1[1]=242.0/256=0.9453125, acc@1[2]=251.5/256=0.982421875 +[2023-10-07 21:52:10,921][root][INFO] - Train batch 200 +[2023-10-07 21:52:10,922][root][INFO] - Avg. loss per last 100 batches: 0.075891 +[2023-10-07 21:52:11,635][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29190.0/29522=98.88% | mean: 0.01 | max: 5.14 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.18 | max: 5.76 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] define tire load index [SEP] ### +### [P_TEXT]: [CLS] the load index is a number that corresponds to the physical load - handling weight ### +### of a tire in kilograms or pounds. the load index can be used as a cross reference to find the ### +### weight in pounds or kilos. for example, a load index of 105 corresponds to 2, 039 pounds or 925 ### +### kilograms. this chart is found inside the vehicle's door jams and is available in tire replacement ### +### shops. the tire rack : load range identification. [SEP] ### +### ======================================= h_v_q | Gates: 26273 ======================================= ### +### ('tire', 0, 1) ('load', 1, 0) ('index', 2, 3) ('definition', 3, 20) ('defined', 4, 368) ### +### ('noun', 5, 24667) ('familiarity', 6, 24287) ('loaded', 7, 59) ('loads', 8, 41) ('.', 9, 13798) ### +### ('something', 10, 5717) ('latin', 11, 6461) ('relating', 12, 23298) ('indices', 13, 6) ### +### ('flow', 14, 4067) ('term', 15, 15639) ('encyclopedia', 16, 2313) ('refers', 17, 1964) ### +### ('loading', 18, 76) ('##load', 19, 155) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('load', 1, 0) ('tire', 0, 1) ('kilograms', 16581, 2) ('index', 2, 3) ('weight', 53, 4) ### +### ('tires', 33, 5) ('indices', 13, 6) ('crashing', 474, 7) ('encompasses', 55, 8) ('##los', 21449, 9) ### +### ('##ο', 145, 10) ('ˈ', 293, 11) ('kg', 6549, 12) ('vehicle', 487, 13) ('pounds', 5289, 14) ### +### ('handling', 9851, 15) ('vehicles', 3979, 16) ('jam', 2053, 17) ('number', 344, 18) ### +### ('##α', 108, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('load', 1, 0) ('tire', 0, 1) ('index', 2, 3) ('definition', 3, 20) ('loads', 8, 41) ### +### ('loaded', 7, 59) ('indices', 13, 6) ('defined', 4, 368) ('tires', 33, 5) ('loading', 18, 76) ### +### ('weight', 53, 4) ('define', 38, 23) ('encompasses', 55, 8) ('##load', 19, 155) ### +### ('definitions', 46, 29) ('##₂', 48, 31) ('##ο', 145, 10) ('##α', 108, 19) ('burden', 32, 149) ### +### ('julian', 59, 64) ### +############################################################################################################ +[2023-10-07 21:52:11,635][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:52:11,635][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:52:12,040][root][INFO] - Epoch: 8: Step: 201/1557, loss[v]=0.080109, lr=0.000012, acc@1[1]=242.0/256=0.9453125, acc@1[2]=247.0/256=0.96484375 +[2023-10-07 21:53:28,688][root][INFO] - Train batch 300 +[2023-10-07 21:53:28,689][root][INFO] - Avg. loss per last 100 batches: 0.075148 +[2023-10-07 21:53:29,386][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29144.3/29522=98.72% | mean: 0.01 | max: 5.37 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.17 | max: 6.15 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] who and when was velcro invented [SEP] ### +### [P_TEXT]: [CLS] velcro was invented by george de mestral a swiss electrical engineer in 1941. this ### +### idea of inventing velcro came to him when one day he returned after a walk from the hills and found ### +### cockleburs stuck to his clothes and his dogas fur. george noticed its natural hook and loop quality ### +### and started making a fabric fastener on the same quality. his idea of inventing velcro came to him ### +### when one day he returned after a walk from the hills and found cockleburs stuck to his clothes and ### +### his dogas fur. george noticed its natural hook and loop quality and started making a fabric ### +### fastener on the same quality. [SEP] ### +### ======================================= h_v_q | Gates: 27967 ======================================= ### +### ('ve', 0, 3) ('##ro', 1, 0) ('##lc', 2, 12) ('invented', 3, 2) ('who', 4, 83) ### +### ('familiarity', 5, 26271) ('was', 6, 905) ('early', 7, 953) ('founded', 8, 308) ('whose', 9, 365) ### +### ('1945', 10, 275) ('.', 11, 19963) ('1964', 12, 10736) ('2017', 13, 19342) ('november', 14, 14076) ### +### ('1996', 15, 5174) ('1944', 16, 168) ('1989', 17, 4074) ('2016', 18, 13600) ('1995', 19, 18346) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##ro', 1, 0) ('george', 282, 1) ('invented', 3, 2) ('ve', 0, 3) ('fabric', 102, 4) ### +### ('dog', 1033, 5) ('stuck', 7373, 6) ('ˈ', 107, 7) ('clothes', 2200, 8) ('fur', 1117, 9) ### +### ('##ο', 145, 10) ('cock', 10325, 11) ('##lc', 2, 12) ('##ང', 125, 13) ('quality', 4557, 14) ### +### ('swiss', 3389, 15) ('##venting', 9137, 16) ('crashing', 116, 17) ('##ener', 25686, 18) ### +### ('invention', 113, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##ro', 1, 0) ('ve', 0, 3) ('##lc', 2, 12) ('invented', 3, 2) ('who', 4, 83) ('founded', 8, 308) ### +### ('1945', 10, 275) ('fabric', 102, 4) ('whose', 9, 365) ('ˈ', 107, 7) ('was', 6, 905) ### +### ('1944', 16, 168) ('wingspan', 77, 34) ('manufactured', 38, 101) ('crashing', 116, 17) ### +### ('##ང', 125, 13) ('##ο', 145, 10) ('invention', 113, 19) ('knew', 34, 134) ('early', 7, 953) ### +############################################################################################################ +[2023-10-07 21:53:29,386][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:53:29,386][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:53:29,806][root][INFO] - Epoch: 8: Step: 301/1557, loss[v]=0.053660, lr=0.000012, acc@1[1]=243.5/256=0.951171875, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 21:54:46,094][root][INFO] - Train batch 400 +[2023-10-07 21:54:46,095][root][INFO] - Avg. loss per last 100 batches: 0.079580 +[2023-10-07 21:54:46,801][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29117.8/29522=98.63% | mean: 0.01 | max: 5.38 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.36 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how did finn in glee really die [SEP] ### +### [P_TEXT]: [CLS] finn hudson ( cory monteith ) is mourned by his classmates ( and castmates ) fox. ### +### last nightas episode of glee, athe quarterback, a paid tribute to finn hudson, the football player ### +### and shower - singer played for four seasons by cory monteith, the young actor who died of a drug ### +### overdose this summer. [SEP] ### +### ======================================= h_v_q | Gates: 27418 ======================================= ### +### ('glee', 0, 0) ('finn', 1, 1) ('died', 2, 12) ('die', 3, 72) ('.', 4, 9396) ('dies', 5, 909) ### +### ('death', 6, 220) ('familiarity', 7, 27196) ('somehow', 8, 1820) ('really', 9, 7369) ### +### ('eager', 10, 257) ('was', 11, 2118) ('killed', 12, 306) ('kill', 13, 3321) ('real', 14, 13896) ### +### ('in', 15, 23836) ('truly', 16, 7286) ('officially', 17, 375) ('genuinely', 18, 221) ### +### ('relating', 19, 26451) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('glee', 0, 0) ('finn', 1, 1) ('hudson', 485, 2) ('shower', 10126, 3) ('cory', 6988, 4) ### +### ('quarterback', 1902, 5) ('cast', 8437, 6) ('fox', 2311, 7) ('monte', 7672, 8) ('tribute', 2027, 9) ### +### ('actor', 2806, 10) ('overdose', 9964, 11) ('died', 2, 12) ('classmates', 19106, 13) ### +### ('episode', 914, 14) ('seasons', 7496, 15) ('##ith', 25467, 16) ('showers', 28216, 17) ### +### ('night', 1571, 18) ('singer', 7486, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('glee', 0, 0) ('finn', 1, 1) ('died', 2, 12) ('die', 3, 72) ('death', 6, 220) ('dies', 5, 909) ### +### ('eager', 10, 257) ('killed', 12, 306) ('dying', 22, 149) ('genuinely', 18, 221) ### +### ('somehow', 8, 1820) ('hudson', 485, 2) ('.', 4, 9396) ('knew', 33, 143) ('aired', 25, 223) ### +### ('officially', 17, 375) ('cemetery', 43, 108) ('football', 179, 23) ('dead', 37, 151) ### +### ('was', 11, 2118) ### +############################################################################################################ +[2023-10-07 21:54:46,802][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:54:46,802][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:54:47,209][root][INFO] - Epoch: 8: Step: 401/1557, loss[v]=0.088059, lr=0.000012, acc@1[1]=245.0/256=0.95703125, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 21:56:03,876][root][INFO] - Train batch 500 +[2023-10-07 21:56:03,877][root][INFO] - Avg. loss per last 100 batches: 0.075100 +[2023-10-07 21:56:04,568][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29134.4/29522=98.69% | mean: 0.01 | max: 5.20 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.00 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] importance of water [SEP] ### +### [P_TEXT]: [CLS] importance of water. water is our lifeline that bathes us and feeds us. in ancient ### +### cultures water represented the very essence of life. the romans were the first to pipe water into ### +### their growing cities, especially with their aqueducts. [SEP] ### +### ======================================= h_v_q | Gates: 26437 ======================================= ### +### ('water', 0, 0) ('importance', 1, 2) ('significance', 2, 35) ('important', 3, 17) ('value', 4, 414) ### +### ('of', 5, 7567) ('familiarity', 6, 25734) ('.', 7, 6273) ('waters', 8, 30) ('relating', 9, 24481) ### +### ('relevance', 10, 95) ('lake', 11, 483) ('significant', 12, 221) ('stream', 13, 602) ### +### ('essential', 14, 300) ('prominence', 15, 103) ('attention', 16, 836) ('river', 17, 682) ### +### (';', 18, 3380) ('intensity', 19, 944) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('water', 0, 0) ('bath', 1227, 1) ('importance', 1, 2) ('romans', 18621, 3) ('essence', 1197, 4) ### +### ('aqueduct', 8210, 5) ('baths', 13977, 6) ('life', 671, 7) ('##ο', 182, 8) ('ˈ', 113, 9) ### +### ('crashing', 276, 10) ('encompasses', 695, 11) ('##ང', 131, 12) ('pipe', 6694, 13) ### +### ('afraid', 368, 14) ('unwilling', 347, 15) ('##₂', 67, 16) ('important', 3, 17) ('−', 50, 18) ### +### ('feed', 10813, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('water', 0, 0) ('importance', 1, 2) ('significance', 2, 35) ('important', 3, 17) ('waters', 8, 30) ### +### ('value', 4, 414) ('relevance', 10, 95) ('−', 50, 18) ('prominence', 15, 103) ('##₂', 67, 16) ### +### ('julian', 31, 62) ('ˈ', 113, 9) ('liquid', 28, 80) ('freshwater', 75, 26) ('##ང', 131, 12) ### +### ('angrily', 52, 45) ('hesitated', 90, 21) ('##water', 22, 118) ('presenter', 44, 66) ### +### ('##ο', 182, 8) ### +############################################################################################################ +[2023-10-07 21:56:04,568][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:56:04,568][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:56:04,991][root][INFO] - Epoch: 8: Step: 501/1557, loss[v]=0.058513, lr=0.000012, acc@1[1]=243.0/256=0.94921875, acc@1[2]=252.0/256=0.984375 +[2023-10-07 21:57:21,457][root][INFO] - Train batch 600 +[2023-10-07 21:57:21,459][root][INFO] - Avg. loss per last 100 batches: 0.076348 +[2023-10-07 21:57:22,159][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29163.8/29522=98.79% | mean: 0.01 | max: 5.39 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.23 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is an arrhythmia [SEP] ### +### [P_TEXT]: [CLS] about arrhythmia. an arrhythmia is an abnormal heart rhythm. learn about the ### +### heart's structure and why abnormal heart rhythms may cause problems. read more. [SEP] ### +### ======================================= h_v_q | Gates: 27961 ======================================= ### +### ('##hmi', 0, 5) ('ar', 1, 2) ('##rh', 2, 43) ('##yt', 3, 19) ('##a', 4, 39) ('encompasses', 5, 7) ### +### ('familiarity', 6, 26513) ('definition', 7, 20) ('refers', 8, 8660) ('is', 9, 252) ### +### ('noun', 10, 24046) ('an', 11, 531) ('relating', 12, 18288) ('term', 13, 17242) ### +### ('association', 14, 9606) ('stands', 15, 10083) ('.', 16, 10869) ('consisting', 17, 20178) ### +### ('simon', 18, 80) ('anton', 19, 135) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('abnormal', 12790, 0) ('heart', 8464, 1) ('ar', 1, 2) ('rhythm', 4279, 3) ('heartbeat', 8700, 4) ### +### ('##hmi', 0, 5) ('crashing', 166, 6) ('encompasses', 5, 7) ('cardiac', 13404, 8) ('afraid', 97, 9) ### +### ('learn', 9973, 10) ('ˈ', 72, 11) ('unwilling', 46, 12) ('##ο', 67, 13) ('##ང', 48, 14) ### +### ('hesitated', 28, 15) ('pulse', 15727, 16) ('rhythms', 16082, 17) ('wingspan', 47, 18) ### +### ('##yt', 3, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ar', 1, 2) ('##hmi', 0, 5) ('##yt', 3, 19) ('##rh', 2, 43) ('encompasses', 5, 7) ('##a', 4, 39) ### +### ('definition', 7, 20) ('is', 9, 252) ('hesitated', 28, 15) ('##₂', 23, 23) ('julian', 22, 34) ### +### ('##α', 26, 33) ('unwilling', 46, 12) ('hugh', 21, 53) ('##ང', 48, 14) ('wingspan', 47, 18) ### +### ('simon', 18, 80) ('ˈ', 72, 11) ('##ο', 67, 13) ('−', 51, 22) ### +############################################################################################################ +[2023-10-07 21:57:22,160][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:57:22,160][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:57:22,582][root][INFO] - Epoch: 8: Step: 601/1557, loss[v]=0.098812, lr=0.000012, acc@1[1]=240.5/256=0.939453125, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 21:58:39,256][root][INFO] - Train batch 700 +[2023-10-07 21:58:39,256][root][INFO] - Avg. loss per last 100 batches: 0.078063 +[2023-10-07 21:58:39,985][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29107.8/29522=98.60% | mean: 0.01 | max: 5.22 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.6/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.15 | max: 5.99 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] population leawood, ks [SEP] ### +### [P_TEXT]: [CLS] leawood / eliewed / is a city in johnson county, kansas, united states, and part of ### +### the kansas city metropolitan area. as of the 2010 census, the city population was 31, 867. [SEP] ### +### ======================================= h_v_q | Gates: 26803 ======================================= ### +### ('##wood', 0, 2) ('lea', 1, 0) ('population', 2, 6) ('kansas', 3, 3) ('missouri', 4, 143) ### +### ('familiarity', 5, 22277) ('.', 6, 3848) ('colorado', 7, 2648) ('arkansas', 8, 310) ### +### ('volume', 9, 3710) ('wood', 10, 380) ('mississippi', 11, 715) ('minnesota', 12, 1981) ### +### ('texas', 13, 2622) ('massachusetts', 14, 1648) ('michigan', 15, 5353) ('oklahoma', 16, 172) ### +### ('music', 17, 1165) ('illinois', 18, 3997) ('georgia', 19, 6317) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('lea', 1, 0) ('eli', 2404, 1) ('##wood', 0, 2) ('kansas', 3, 3) ('johnson', 180, 4) ### +### ('metropolitan', 4367, 5) ('population', 2, 6) ('census', 1095, 7) ('encompasses', 1056, 8) ### +### ('crashing', 671, 9) ('30', 126, 10) ('##ང', 225, 11) ('afraid', 393, 12) ('where', 388, 13) ### +### ('county', 1734, 14) ('counties', 6366, 15) ('crashed', 1391, 16) ('ˈ', 534, 17) ('city', 907, 18) ### +### ('##ew', 13759, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##wood', 0, 2) ('lea', 1, 0) ('population', 2, 6) ('kansas', 3, 3) ('missouri', 4, 143) ### +### ('ks', 35, 27) ('##woods', 30, 38) ('states', 23, 67) ('simon', 27, 51) ('300', 42, 33) ### +### ('johnson', 180, 4) ('number', 22, 96) ('oklahoma', 16, 172) ('arkansas', 8, 310) ('30', 126, 10) ### +### ('residents', 24, 169) ('wood', 10, 380) ('unwilling', 136, 21) ('hugh', 84, 45) ('julian', 79, 55) ### +############################################################################################################ +[2023-10-07 21:58:39,986][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:58:39,986][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:58:40,394][root][INFO] - Epoch: 8: Step: 701/1557, loss[v]=0.083504, lr=0.000012, acc@1[1]=242.5/256=0.947265625, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 21:59:56,624][root][INFO] - Train batch 800 +[2023-10-07 21:59:56,625][root][INFO] - Avg. loss per last 100 batches: 0.078349 +[2023-10-07 21:59:57,348][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29022.0/29522=98.31% | mean: 0.01 | max: 5.29 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.07 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] definition not telling the truth [SEP] ### +### [P_TEXT]: [CLS] use lying in a sentence. link / cite add to word list. adjective. the definition of ### +### lying is not telling the truth. an example of someone who is lying is someone who is dishonest ### +### about where he was and what he did. [SEP] ### +### ======================================= h_v_q | Gates: 26709 ======================================= ### +### ('truth', 0, 1) ('definition', 1, 6) ('not', 2, 48) ('telling', 3, 10) ('noun', 4, 8611) ### +### ('defined', 5, 33) ('familiarity', 6, 26746) ('term', 7, 679) ('honest', 8, 155) ('.', 9, 5053) ### +### ('something', 10, 445) ('relating', 11, 13170) ('refers', 12, 9689) ('meaning', 13, 8) ### +### ('knowledge', 14, 5438) ('sense', 15, 3231) ('never', 16, 219) ('true', 17, 221) ('means', 18, 110) ### +### ('touching', 19, 457) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('lying', 521, 0) ('truth', 0, 1) ('##ones', 25379, 2) ('definitions', 59, 3) ('lie', 34, 4) ### +### ('define', 1330, 5) ('definition', 1, 6) ('lies', 72, 7) ('meaning', 13, 8) ('crashing', 460, 9) ### +### ('telling', 3, 10) ('sentence', 2851, 11) ('liar', 7105, 12) ('hesitated', 321, 13) ### +### ('cite', 19233, 14) ('meanings', 1185, 15) ('examples', 264, 16) ('ˈ', 596, 17) ### +### ('unwilling', 38, 18) ('example', 211, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('truth', 0, 1) ('definition', 1, 6) ('telling', 3, 10) ('not', 2, 48) ('defined', 5, 33) ### +### ('meaning', 13, 8) ('lie', 34, 4) ('definitions', 59, 3) ('honest', 8, 155) ('lies', 72, 7) ### +### ('unwilling', 38, 18) ('honesty', 22, 68) ('means', 18, 110) ('lying', 521, 0) ('ignoring', 46, 38) ### +### ('term', 7, 679) ('never', 16, 219) ('something', 10, 445) ('true', 17, 221) ('mean', 86, 42) ### +############################################################################################################ +[2023-10-07 21:59:57,348][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 21:59:57,348][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 21:59:57,769][root][INFO] - Epoch: 8: Step: 801/1557, loss[v]=0.094842, lr=0.000012, acc@1[1]=235.5/256=0.919921875, acc@1[2]=248.5/256=0.970703125 +[2023-10-07 22:01:14,417][root][INFO] - Train batch 900 +[2023-10-07 22:01:14,418][root][INFO] - Avg. loss per last 100 batches: 0.076759 +[2023-10-07 22:01:15,120][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29097.2/29522=98.56% | mean: 0.01 | max: 5.09 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.5/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.14 | max: 5.94 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] who is morpheus in greek mythology [SEP] ### +### [P_TEXT]: [CLS] in greek gods and heroes. morpheus was the god of dreams, the one with the amazing ### +### ability of appearing in dreams of mortals in any form. as his name implies a the greek word ### +### amorphea means aforma a he was the one who shaped and formed the dreams. [SEP] ### +### ======================================= h_v_q | Gates: 27047 ======================================= ### +### ('##pheus', 0, 0) ('mor', 1, 3) ('greek', 2, 7) ('mythology', 3, 236) ('whose', 4, 188) ### +### ('who', 5, 35) ('is', 6, 402) ('greece', 7, 10) ('familiarity', 8, 26535) ('.', 9, 5658) ### +### ('born', 10, 15335) ('encompasses', 11, 15) ('relating', 12, 20544) ('association', 13, 7685) ### +### ('greeks', 14, 29) ('history', 15, 9661) ('consisting', 16, 13741) ('league', 17, 13731) ### +### ('julian', 18, 77) ('person', 19, 7447) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##pheus', 0, 0) ('dreams', 1888, 1) ('amor', 8194, 2) ('mor', 1, 3) ('god', 540, 4) ### +### ('gods', 424, 5) ('af', 6133, 6) ('greek', 2, 7) ('dream', 2206, 8) ('heroes', 1380, 9) ### +### ('greece', 7, 10) ('crashing', 88, 11) ('dreaming', 19167, 12) ('##ο', 220, 13) ('ˈ', 333, 14) ### +### ('encompasses', 11, 15) ('amazing', 8458, 16) ('mortals', 23811, 17) ('##ང', 193, 18) ### +### ('hesitated', 130, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##pheus', 0, 0) ('mor', 1, 3) ('greek', 2, 7) ('greece', 7, 10) ('who', 5, 35) ### +### ('mythology', 3, 236) ('whose', 4, 188) ('encompasses', 11, 15) ('is', 6, 402) ('greeks', 14, 29) ### +### ('julian', 18, 77) ('crashing', 88, 11) ('simon', 21, 68) ('altogether', 43, 67) ### +### ('hesitated', 130, 19) ('adam', 27, 99) ('weird', 102, 31) ('gods', 424, 5) ('definition', 101, 33) ### +### ('hugh', 53, 62) ### +############################################################################################################ +[2023-10-07 22:01:15,120][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:01:15,120][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:01:15,543][root][INFO] - Epoch: 8: Step: 901/1557, loss[v]=0.080222, lr=0.000012, acc@1[1]=242.0/256=0.9453125, acc@1[2]=248.0/256=0.96875 +[2023-10-07 22:02:32,224][root][INFO] - Train batch 1000 +[2023-10-07 22:02:32,225][root][INFO] - Avg. loss per last 100 batches: 0.075773 +[2023-10-07 22:02:32,929][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29193.0/29522=98.89% | mean: 0.01 | max: 5.21 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.17 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] when ordering multifocal what is the add [SEP] ### +### [P_TEXT]: [CLS] the first sign that you may need multifocal contact lenses is if your prescription ### +### for glasses has numbers in the add column. numbers in this column indicate that you are presbyopic ### +### to some extent. although there is no set limit to what the add number can be, it usually ranges ### +### between + 0. 75 to + 2. 50. the higher your add number is, the more you stand to benefit from ### +### multifocal contact lenses. but more importantly than the add number, is how bothered you are by ### +### your vision up close. [SEP] ### +### ======================================= h_v_q | Gates: 28118 ======================================= ### +### ('multi', 0, 12) ('add', 1, 7) ('##fo', 2, 25) ('##cal', 3, 38) ('order', 4, 10648) ### +### ('ordered', 5, 17474) ('ordering', 6, 20265) ('multiple', 7, 153) ('orders', 8, 5449) ### +### ('.', 9, 14202) ('familiarity', 10, 26936) ('adding', 11, 191) ('is', 12, 13508) ('added', 13, 600) ### +### ('encompasses', 14, 546) ('definition', 15, 5867) ('when', 16, 3538) ('relating', 17, 23925) ### +### ('what', 18, 763) ('addition', 19, 114) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('lenses', 20042, 0) ('glasses', 10988, 1) ('contact', 12094, 2) ('lens', 15037, 3) ### +### ('##sby', 18884, 4) ('number', 1779, 5) ('signs', 2667, 6) ('add', 1, 7) ('numbers', 7039, 8) ### +### ('sign', 3738, 9) ('prescription', 3566, 10) ('bother', 260, 11) ('multi', 0, 12) ### +### ('importantly', 18899, 13) ('crashing', 72, 14) ('vision', 2429, 15) ('ˈ', 283, 16) ### +### ('unwilling', 45, 17) ('##ο', 106, 18) ('bothering', 3018, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('multi', 0, 12) ('add', 1, 7) ('##fo', 2, 25) ('##cal', 3, 38) ('multiple', 7, 153) ### +### ('adding', 11, 191) ('altogether', 24, 58) ('addition', 19, 114) ('simon', 23, 93) ### +### ('julian', 28, 63) ('unwilling', 45, 17) ('order', 4, 10648) ('crashing', 72, 14) ### +### ('added', 13, 600) ('##ο', 106, 18) ('sharply', 89, 26) ('encompasses', 14, 546) ### +### ('hesitated', 101, 24) ('angrily', 64, 39) ('bother', 260, 11) ### +############################################################################################################ +[2023-10-07 22:02:32,929][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:02:32,929][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:02:33,335][root][INFO] - Epoch: 8: Step: 1001/1557, loss[v]=0.084226, lr=0.000012, acc@1[1]=243.5/256=0.951171875, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 22:03:50,424][root][INFO] - Train batch 1100 +[2023-10-07 22:03:50,425][root][INFO] - Avg. loss per last 100 batches: 0.075667 +[2023-10-07 22:04:01,334][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29225.0/29522=98.99% | mean: 0.01 | max: 5.36 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.39 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how long do you have to be married to get japan visa [SEP] ### +### [P_TEXT]: [CLS] if you get divorced from your japanese spouse, it is required to report the divorce ### +### to the immigration office under the new law from july 9, 2012. the immigration may revoke your ### +### spouse visa after 6 months from the divorce. it is eventually possible to change your spouse visa ### +### to a long term resident visa if you have lived long enough in japan ( approximately 5 years ) or if ### +### you have a child of japanese nationality to raise in japan. f you get divorced from your japanese ### +### spouse, it is required to report the divorce to the immigration office under the new law from july ### +### 9, 2012. [SEP] ### +### ======================================= h_v_q | Gates: 27287 ======================================= ### +### ('japan', 0, 3) ('married', 1, 39) ('visa', 2, 8) ('days', 3, 3159) ('weeks', 4, 30) ### +### ('years', 5, 99) ('months', 6, 271) ('minutes', 7, 163) ('long', 8, 36) ('japanese', 9, 4) ### +### ('.', 10, 6150) ('you', 11, 236) ('familiarity', 12, 20818) ('wife', 13, 124) ('hours', 14, 3720) ### +### ('longer', 15, 145) ('marriage', 16, 56) ('get', 17, 94) ('jeremy', 18, 67) ('husband', 19, 29) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('divorced', 457, 0) ('divorce', 98, 1) ('spouse', 905, 2) ('japan', 0, 3) ('japanese', 9, 4) ### +### ('nationality', 14717, 5) ('immigration', 3041, 6) ('crashing', 291, 7) ('visa', 2, 8) ### +### ('##ང', 200, 9) ('july', 425, 10) ('ˈ', 212, 11) ('##ο', 194, 12) ('unwilling', 63, 13) ### +### ('eventually', 3402, 14) ('eventual', 1327, 15) ('##α', 344, 16) ('−', 150, 17) ### +### ('resident', 1270, 18) ('reports', 8201, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('japan', 0, 3) ('visa', 2, 8) ('married', 1, 39) ('weeks', 4, 30) ('japanese', 9, 4) ### +### ('years', 5, 99) ('long', 8, 36) ('months', 6, 271) ('minutes', 7, 163) ('wife', 13, 124) ### +### ('husband', 19, 29) ('marriage', 16, 56) ('you', 11, 236) ('jeremy', 18, 67) ('get', 17, 94) ### +### ('divorce', 98, 1) ('longer', 15, 145) ('days', 3, 3159) ('unwilling', 63, 13) ### +### ('duration', 25, 162) ### +############################################################################################################ +[2023-10-07 22:04:01,335][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:04:01,335][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:04:01,734][root][INFO] - Epoch: 8: Step: 1101/1557, loss[v]=0.105923, lr=0.000012, acc@1[1]=238.5/256=0.931640625, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 22:05:18,021][root][INFO] - Train batch 1200 +[2023-10-07 22:05:18,022][root][INFO] - Avg. loss per last 100 batches: 0.074693 +[2023-10-07 22:05:18,739][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29318.2/29522=99.31% | mean: 0.01 | max: 5.46 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.03 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what are types of third party liability insurance [SEP] ### +### [P_TEXT]: [CLS] third - party liability coverage can be included as part of a range of insurance ### +### types including vehicle, business and homeowner coverage. in an auto policy, the third party could ### +### be the person suing you because you rear - ended their car or the person whose fence you destroyed ### +### when you swerved off the road. hird - party liability coverage can be included as part of a range ### +### of insurance types including vehicle, business and homeowner coverage. [SEP] ### +### ======================================= h_v_q | Gates: 28482 ======================================= ### +### ('liability', 0, 1) ('types', 1, 10) ('third', 2, 9) ('insurance', 3, 5) ('party', 4, 2) ### +### ('3rd', 5, 43) ('familiarity', 6, 27080) ('categories', 7, 119) ('encompasses', 8, 62) ### +### ('.', 9, 9983) ('relating', 10, 20878) ('type', 11, 500) ('fourth', 12, 78) ('simon', 13, 70) ### +### ('unwilling', 14, 20) ('consisting', 15, 18580) ('classes', 16, 320) ('hugh', 17, 80) ### +### ('crashing', 18, 7) ('are', 19, 7151) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##erved', 25663, 0) ('liability', 0, 1) ('party', 4, 2) ('fence', 12741, 3) ('##rd', 15911, 4) ### +### ('insurance', 3, 5) ('auto', 5268, 6) ('crashing', 18, 7) ('vehicle', 3829, 8) ('third', 2, 9) ### +### ('types', 1, 10) ('coverage', 597, 11) ('included', 10593, 12) ('##ང', 40, 13) ('##ο', 49, 14) ### +### ('ˈ', 26, 15) ('car', 1693, 16) ('hi', 1982, 17) ('policy', 1351, 18) ('hesitated', 37, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('liability', 0, 1) ('types', 1, 10) ('insurance', 3, 5) ('third', 2, 9) ('party', 4, 2) ### +### ('3rd', 5, 43) ('encompasses', 8, 62) ('unwilling', 14, 20) ('crashing', 18, 7) ### +### ('categories', 7, 119) ('fourth', 12, 78) ('simon', 13, 70) ('##₂', 21, 26) ('ˈ', 26, 15) ### +### ('hugh', 17, 80) ('hesitated', 37, 19) ('##ང', 40, 13) ('##ο', 49, 14) ('include', 32, 36) ### +### ('julian', 27, 56) ### +############################################################################################################ +[2023-10-07 22:05:18,739][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:05:18,739][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:05:19,161][root][INFO] - Epoch: 8: Step: 1201/1557, loss[v]=0.066053, lr=0.000012, acc@1[1]=242.5/256=0.947265625, acc@1[2]=252.0/256=0.984375 +[2023-10-07 22:06:36,259][root][INFO] - Train batch 1300 +[2023-10-07 22:06:36,260][root][INFO] - Avg. loss per last 100 batches: 0.074789 +[2023-10-07 22:06:36,957][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29238.0/29522=99.04% | mean: 0.01 | max: 5.67 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.14 | max: 6.11 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] where is lebron james from [SEP] ### +### [P_TEXT]: [CLS] lebron james was born on december 30 1984 in akron ohio usa as lebron raymone james ### +### he is a producer and actor known for more than a game 2008 2015 nba all star all style 2015 and ### +### trainwreck 2015 he has been married to savannah brinson james since september 14 2013 [SEP] ### +### ======================================= h_v_q | Gates: 27931 ======================================= ### +### ('james', 0, 1) ('le', 1, 8) ('##bron', 2, 6) ('born', 3, 10) ('familiarity', 4, 27223) ### +### ('located', 5, 10288) ('from', 6, 12331) ('united', 7, 7223) ('where', 8, 157) ### +### ('massachusetts', 9, 7295) ('connecticut', 10, 3515) ('native', 11, 5998) ('england', 12, 7156) ### +### ('america', 13, 6447) ('brazil', 14, 5953) ('pennsylvania', 15, 646) ('is', 16, 1057) ### +### ('washington', 17, 3863) ('commonwealth', 18, 4590) ('.', 19, 7776) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('akron', 21798, 0) ('james', 0, 1) ('nba', 3347, 2) ('savannah', 1325, 3) ('married', 396, 4) ### +### ('ohio', 23, 5) ('##bron', 2, 6) ('actor', 2713, 7) ('le', 1, 8) ('producer', 1955, 9) ### +### ('born', 3, 10) ('crashing', 219, 11) ('star', 1559, 12) ('##ང', 338, 13) ('##ο', 225, 14) ### +### ('style', 1804, 15) ('usa', 31, 16) ('ray', 2288, 17) ('marry', 3886, 18) ('game', 2620, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('james', 0, 1) ('##bron', 2, 6) ('le', 1, 8) ('born', 3, 10) ('ohio', 23, 5) ('usa', 31, 16) ### +### ('where', 8, 157) ('jim', 22, 147) ('italy', 32, 106) ('simon', 57, 62) ('adam', 54, 91) ### +### ('crashing', 219, 11) ('married', 396, 4) ('##ο', 225, 14) ('santiago', 188, 35) ### +### ('atlanta', 45, 159) ('##ང', 338, 13) ('gideon', 260, 33) ('india', 20, 406) ('hesitated', 262, 32) ### +############################################################################################################ +[2023-10-07 22:06:36,957][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:06:36,958][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:06:37,364][root][INFO] - Epoch: 8: Step: 1301/1557, loss[v]=0.063787, lr=0.000012, acc@1[1]=243.5/256=0.951171875, acc@1[2]=253.0/256=0.98828125 +[2023-10-07 22:07:53,545][root][INFO] - Train batch 1400 +[2023-10-07 22:07:53,545][root][INFO] - Avg. loss per last 100 batches: 0.075457 +[2023-10-07 22:07:54,243][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29178.1/29522=98.84% | mean: 0.01 | max: 5.34 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.07 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what do meerkats look like [SEP] ### +### [P_TEXT]: [CLS] the meerkat ( suricata suricatta ) is a small member of the mongoose family, which ### +### is part of the feliformia ( cat - like carnivores ) order. they are also refered to as suricates, s ### +### lender - tailed meerkats and grey meerkats. [SEP] ### +### ======================================= h_v_q | Gates: 28190 ======================================= ### +### ('##kat', 0, 0) ('me', 1, 5) ('##er', 2, 8) ('look', 3, 6523) ('like', 4, 141) ('##s', 5, 123) ### +### ('looks', 6, 17771) ('glance', 7, 22129) ('familiarity', 8, 28269) ('looked', 9, 19516) ### +### ('.', 10, 10546) ('plural', 11, 8317) ('resembles', 12, 1903) ('simon', 13, 53) ('##स', 14, 90) ### +### ('##ers', 15, 154) ('design', 16, 14374) ('##ς', 17, 45) ('relating', 18, 22644) ### +### ('looking', 19, 2724) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##kat', 0, 0) ('##icate', 22821, 1) ('order', 1050, 2) ('tailed', 13058, 3) ('sur', 4203, 4) ### +### ('me', 1, 5) ('cat', 120, 6) ('family', 594, 7) ('##er', 2, 8) ('##vor', 22139, 9) ### +### ('encompasses', 330, 10) ('crashing', 61, 11) ('small', 4496, 12) ('##go', 7420, 13) ### +### ('grey', 5085, 14) ('mon', 16251, 15) ('##tta', 13520, 16) ('##ica', 8536, 17) ('##ང', 110, 18) ### +### ('lend', 15565, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##kat', 0, 0) ('me', 1, 5) ('##er', 2, 8) ('##s', 5, 123) ('like', 4, 141) ('simon', 13, 53) ### +### ('##ς', 17, 45) ('look', 3, 6523) ('##स', 14, 90) ('kat', 21, 80) ('crashing', 61, 11) ### +### ('##ers', 15, 154) ('cat', 120, 6) ('altogether', 37, 72) ('##ང', 110, 18) ('hesitated', 97, 25) ### +### ('##₂', 76, 38) ('unwilling', 86, 29) ('##ο', 105, 22) ('julian', 52, 75) ### +############################################################################################################ +[2023-10-07 22:07:54,244][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:07:54,244][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:07:54,649][root][INFO] - Epoch: 8: Step: 1401/1557, loss[v]=0.059761, lr=0.000012, acc@1[1]=243.0/256=0.94921875, acc@1[2]=252.5/256=0.986328125 +[2023-10-07 22:09:11,146][root][INFO] - Train batch 1500 +[2023-10-07 22:09:11,147][root][INFO] - Avg. loss per last 100 batches: 0.075931 +[2023-10-07 22:09:11,869][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29140.0/29522=98.71% | mean: 0.01 | max: 5.52 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.15 | max: 5.99 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] sources of magnesium in food list [SEP] ### +### [P_TEXT]: [CLS] the sources of magnesium include food, water, and supplements. while clearly a ### +### agooda source of magnesium is one that is readily available and easily absorbed, most experts ### +### recommend a combination of sources, taking advantage of both dietary magnesium and absorbable ### +### magnesium supplements. raditionally, foods highest in magnesium content are green vegetables, whole ### +### grain cereals, nuts and beans, and seafood. according to usda food charts ( see a complete chart of ### +### magnesium rich foods ), the five foods with the highest magnesium per typical serving are : 1 ### +### halibut. 2 mackeral. 3 boiled spinach. [SEP] ### +### ======================================= h_v_q | Gates: 27041 ======================================= ### +### ('magnesium', 0, 0) ('sources', 1, 12) ('list', 2, 7826) ('food', 3, 13) ('.', 4, 7327) ### +### ('source', 5, 11) ('foods', 6, 4) ('of', 7, 3869) ('fish', 8, 599) ('familiarity', 9, 28494) ### +### ('in', 10, 7966) ('bread', 11, 717) ('supplies', 12, 156) ('protein', 13, 647) ('simon', 14, 112) ### +### ('meat', 15, 290) ('group', 16, 5732) ('relating', 17, 22472) ('roster', 18, 9718) ### +### ('consisting', 19, 10239) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('magnesium', 0, 0) ('##ditional', 20413, 1) ('ago', 14010, 2) ('supplements', 7940, 3) ### +### ('foods', 6, 4) ('##oda', 25034, 5) ('vegetables', 149, 6) ('nuts', 1121, 7) ('include', 390, 8) ### +### ('highest', 2237, 9) ('crashing', 1059, 10) ('source', 5, 11) ('sources', 1, 12) ('food', 3, 13) ### +### ('worried', 217, 14) ('∈', 26, 15) ('dietary', 1674, 16) ('gideon', 617, 17) ('ˈ', 866, 18) ### +### ('combination', 5377, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('magnesium', 0, 0) ('sources', 1, 12) ('food', 3, 13) ('foods', 6, 4) ('source', 5, 11) ### +### ('list', 2, 7826) ('∈', 26, 15) ('.', 4, 7327) ('simon', 14, 112) ('vegetables', 149, 6) ### +### ('mushrooms', 25, 84) ('supplies', 12, 156) ('julian', 35, 75) ('calcium', 69, 42) ### +### ('altogether', 46, 76) ('fish', 8, 599) ('afraid', 107, 34) ('worried', 217, 14) ### +### ('include', 390, 8) ('unwilling', 243, 22) ### +############################################################################################################ +[2023-10-07 22:09:11,869][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:09:11,869][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:09:12,295][root][INFO] - Epoch: 8: Step: 1501/1557, loss[v]=0.063001, lr=0.000012, acc@1[1]=247.0/256=0.96484375, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 22:09:55,745][root][INFO] - rank=3; last iteration 1557 +[2023-10-07 22:09:55,746][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 22:09:55,746][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-07 22:09:55,746][root][INFO] - rank=1; last iteration 1557 +[2023-10-07 22:09:55,747][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 22:09:55,747][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-07 22:09:55,748][root][INFO] - rank=0; last iteration 1557 +[2023-10-07 22:09:55,748][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 22:09:55,748][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-07 22:09:55,748][root][INFO] - rank=2; last iteration 1557 +[2023-10-07 22:09:55,748][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 22:09:55,748][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-07 22:09:55,754][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 22:09:55,754][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 22:09:55,754][root][INFO] - Epoch finished on 3 +[2023-10-07 22:09:55,755][root][INFO] - Epoch finished on 1 +[2023-10-07 22:09:55,755][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 22:09:55,756][root][INFO] - Epoch finished on 0 +[2023-10-07 22:09:55,756][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 22:09:55,756][root][INFO] - Epoch finished on 2 +[2023-10-07 22:10:10,104][root][INFO] - Saved checkpoint at ./vdr_8 +[2023-10-07 22:10:10,105][root][INFO] - Av Loss per epoch=0.076132 +[2023-10-07 22:10:10,105][root][INFO] - epoch total (1) correct predictions=377138 +[2023-10-07 22:10:10,106][root][INFO] - epoch total (2) correct predictions=389680 +[2023-10-07 22:10:10,105][root][INFO] - Saved checkpoint at ./vdr_8 +[2023-10-07 22:10:10,106][root][INFO] - Av Loss per epoch=0.076132 +[2023-10-07 22:10:10,106][root][INFO] - epoch total (1) correct predictions=377138 +[2023-10-07 22:10:10,106][root][INFO] - epoch total (2) correct predictions=389680 +[2023-10-07 22:10:10,106][root][INFO] - Saved checkpoint at ./vdr_8 +[2023-10-07 22:10:10,107][root][INFO] - Av Loss per epoch=0.076132 +[2023-10-07 22:10:10,107][root][INFO] - epoch total (1) correct predictions=377138 +[2023-10-07 22:10:10,107][root][INFO] - epoch total (2) correct predictions=389680 +[2023-10-07 22:10:10,108][root][INFO] - Saved checkpoint at ./vdr_8 +[2023-10-07 22:10:10,109][root][INFO] - Av Loss per epoch=0.076132 +[2023-10-07 22:10:10,109][root][INFO] - epoch total (1) correct predictions=377138 +[2023-10-07 22:10:10,110][root][INFO] - epoch total (2) correct predictions=389680 +[2023-10-07 22:10:10,109][root][INFO] - ***** Epoch 9 ***** +[2023-10-07 22:10:10,109][root][INFO] - ***** Epoch 9 ***** +[2023-10-07 22:10:10,111][root][INFO] - ***** Epoch 9 ***** +[2023-10-07 22:10:10,113][root][INFO] - rank=1; Iteration start +[2023-10-07 22:10:10,113][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 22:10:10,113][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 22:10:10,114][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-07 22:10:10,116][root][INFO] - rank=0; Iteration start +[2023-10-07 22:10:10,116][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 22:10:10,116][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 22:10:10,118][root][INFO] - rank=3; Iteration start +[2023-10-07 22:10:10,118][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 22:10:10,118][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 22:10:10,118][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-07 22:10:10,120][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-07 22:10:10,118][root][INFO] - ***** Epoch 9 ***** +[2023-10-07 22:10:10,124][root][INFO] - rank=2; Iteration start +[2023-10-07 22:10:10,125][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 22:10:10,125][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 22:10:10,127][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-07 22:10:11,093][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29180.4/29522=98.84% | mean: 0.01 | max: 5.30 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.01 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] where did the name burton come from [SEP] ### +### [P_TEXT]: [CLS] making the world better, one answer at a time. the country where the last name ### +### burton originated is england. this name is used as a last name as well as a first name and means ### +### from the fortified town. [SEP] ### +### ======================================= h_v_q | Gates: 27908 ======================================= ### +### ('burton', 0, 0) ('name', 1, 14) ('comes', 2, 14622) ('surname', 3, 55) ('united', 4, 10000) ### +### ('born', 5, 518) ('from', 6, 968) ('##ュ', 7, 420) ('.', 8, 6891) ('america', 9, 12996) ### +### ('england', 10, 2) ('was', 11, 10298) ('where', 12, 48) ('washington', 13, 15261) ('come', 14, 503) ### +### ('pennsylvania', 15, 11782) ('brazil', 16, 2415) ('originated', 17, 3) ('came', 18, 853) ### +### ('relating', 19, 27477) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('burton', 0, 0) ('fortified', 2819, 1) ('england', 10, 2) ('originated', 17, 3) ### +### ('crashing', 772, 4) ('country', 34, 5) ('originate', 4511, 6) ('last', 4027, 7) ('town', 324, 8) ### +### ('ˈ', 1836, 9) ('answer', 213, 10) ('better', 4696, 11) ('names', 56, 12) ('unwilling', 1330, 13) ### +### ('name', 1, 14) ('##ο', 1087, 15) ('##ང', 782, 16) ('gideon', 1694, 17) ('ছ', 1335, 18) ### +### ('final', 2427, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('burton', 0, 0) ('name', 1, 14) ('england', 10, 2) ('surname', 3, 55) ('originated', 17, 3) ### +### ('country', 34, 5) ('where', 12, 48) ('nickname', 28, 42) ('names', 56, 12) ('somewhere', 33, 51) ### +### ('named', 35, 61) ('richard', 40, 54) ('born', 5, 518) ('##ュ', 7, 420) ('italy', 23, 148) ### +### ('uk', 72, 46) ('from', 6, 968) ('answer', 213, 10) ('come', 14, 503) ('world', 169, 25) ### +############################################################################################################ +[2023-10-07 22:10:11,094][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:10:11,094][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:10:11,490][root][INFO] - Epoch: 9: Step: 1/1557, loss[v]=0.065419, lr=0.000012, acc@1[1]=241.0/256=0.94140625, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 22:11:27,568][root][INFO] - Train batch 100 +[2023-10-07 22:11:27,568][root][INFO] - Avg. loss per last 100 batches: 0.072918 +[2023-10-07 22:11:28,280][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29052.6/29522=98.41% | mean: 0.01 | max: 5.18 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.17 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how do littoral zones differ from riparian zones? [SEP] ### +### [P_TEXT]: [CLS] riparian zones occur where the land meets the water. littoral zones occur in the ### +### transition zone between water and dry land. littoral zones extend until the a¦ water depth is ### +### approximately 15 feet. both zones are important for the health of the aquatic environment. - e2020 ### +### /. [SEP] ### +### ======================================= h_v_q | Gates: 27417 ======================================= ### +### ('rip', 0, 3) ('lit', 1, 11) ('##arian', 2, 1) ('##tor', 3, 13) ('##al', 4, 25) ('zones', 5, 4) ### +### ('zone', 6, 2) ('.', 7, 8331) ('differ', 8, 18569) ('differences', 9, 7787) ### +### ('familiarity', 10, 26079) ('somehow', 11, 394) ('light', 12, 2770) ('eager', 13, 1516) ### +### ('manual', 14, 704) ('from', 15, 7864) ('do', 16, 312) ('relating', 17, 24449) ('whereas', 18, 413) ### +### ('plural', 19, 12784) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##¦', 29510, 0) ('##arian', 2, 1) ('zone', 6, 2) ('rip', 0, 3) ('zones', 5, 4) ### +### ('transition', 2312, 5) ('aquatic', 10111, 6) ('depth', 3062, 7) ('##arians', 42, 8) ### +### ('dry', 1722, 9) ('occur', 3634, 10) ('lit', 1, 11) ('water', 3011, 12) ('##tor', 3, 13) ### +### ('land', 4111, 14) ('environment', 3175, 15) ('gideon', 137, 16) ('crashing', 120, 17) ### +### ('15', 9812, 18) ('unwilling', 41, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('rip', 0, 3) ('##arian', 2, 1) ('lit', 1, 11) ('##tor', 3, 13) ('zones', 5, 4) ('zone', 6, 2) ### +### ('##al', 4, 25) ('##arians', 42, 8) ('altogether', 21, 67) ('simon', 22, 57) ('julian', 27, 45) ### +### ('unwilling', 41, 19) ('##tors', 28, 52) ('somehow', 11, 394) ('ripping', 24, 93) ('tina', 62, 22) ### +### ('include', 46, 49) ('do', 16, 312) ('##₂', 66, 34) ('crashing', 120, 17) ### +############################################################################################################ +[2023-10-07 22:11:28,281][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:11:28,281][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:11:28,708][root][INFO] - Epoch: 9: Step: 101/1557, loss[v]=0.066598, lr=0.000012, acc@1[1]=242.5/256=0.947265625, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 22:12:44,477][root][INFO] - Train batch 200 +[2023-10-07 22:12:44,478][root][INFO] - Avg. loss per last 100 batches: 0.073889 +[2023-10-07 22:12:45,204][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29143.2/29522=98.72% | mean: 0.01 | max: 5.39 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.17 | max: 6.13 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] most popular iso standards [SEP] ### +### [P_TEXT]: [CLS] iso 9001 is the worldas most popular quality management. system standard and is all ### +### about keeping customers satisi¬ed. whatever sector you operate in, from manufacturing to. services, ### +### companies that work with bsi to adopt the principles. of quality management have benei¬ted from ### +### more efi¬cient. [SEP] ### +### ======================================= h_v_q | Gates: 26963 ======================================= ### +### ('iso', 0, 0) ('popular', 1, 7) ('standards', 2, 71) ('.', 3, 10207) ('most', 4, 87) ### +### ('popularity', 5, 18) ('standard', 6, 11) ('top', 7, 535) ('famous', 8, 254) ### +### ('familiarity', 9, 25218) ('principles', 10, 32) ('greatest', 11, 243) ('relating', 12, 25258) ### +### ('best', 13, 1803) ('highest', 14, 268) ('consisting', 15, 21443) ('largest', 16, 165) ### +### ('quality', 17, 1) ('numerous', 18, 861) ('software', 19, 204) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('iso', 0, 0) ('quality', 17, 1) ('bs', 6712, 2) ('management', 214, 3) ('##1', 3115, 4) ### +### ('##i', 7586, 5) ('900', 4198, 6) ('popular', 1, 7) ('##isi', 27343, 8) ('sat', 2047, 9) ### +### ('ˈ', 324, 10) ('standard', 6, 11) ('customers', 6205, 12) ('companies', 1326, 13) ### +### ('encompasses', 1274, 14) ('crashing', 452, 15) ('manufacturing', 1942, 16) ('icao', 4493, 17) ### +### ('popularity', 5, 18) ('keeping', 2408, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('iso', 0, 0) ('popular', 1, 7) ('standards', 2, 71) ('popularity', 5, 18) ('most', 4, 87) ### +### ('standard', 6, 11) ('quality', 17, 1) ('principles', 10, 32) ('famous', 8, 254) ('top', 7, 535) ### +### ('greatest', 11, 243) ('management', 214, 3) ('highest', 14, 268) ('.', 3, 10207) ### +### ('julian', 24, 95) ('hated', 42, 55) ('definition', 59, 41) ('simon', 32, 88) ('largest', 16, 165) ### +### ('##₂', 103, 22) ### +############################################################################################################ +[2023-10-07 22:12:45,205][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:12:45,205][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:12:45,609][root][INFO] - Epoch: 9: Step: 201/1557, loss[v]=0.051989, lr=0.000011, acc@1[1]=241.5/256=0.943359375, acc@1[2]=252.0/256=0.984375 +[2023-10-07 22:14:02,240][root][INFO] - Train batch 300 +[2023-10-07 22:14:02,241][root][INFO] - Avg. loss per last 100 batches: 0.075321 +[2023-10-07 22:14:02,915][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29155.6/29522=98.76% | mean: 0.01 | max: 5.23 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.17 | max: 6.14 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what county is hopkins sc in [SEP] ### +### [P_TEXT]: [CLS] sponsored topics. hopkins is an unincorporated community in richland county, south ### +### carolina, united states that was founded in circa 1836 named after john hopkins. it is located ### +### eleven miles from downtown columbia and is part of the columbia metropolitan statistical area. ### +### [SEP] ### +### ======================================= h_v_q | Gates: 27338 ======================================= ### +### ('hopkins', 0, 0) ('county', 1, 15) ('carolina', 2, 6) ('sc', 3, 24) ('mississippi', 4, 1453) ### +### ('florida', 5, 841) ('.', 6, 5933) ('louisiana', 7, 2729) ('georgia', 8, 878) ('ohio', 9, 5309) ### +### ('is', 10, 125) ('ward', 11, 1042) ('tennessee', 12, 614) ('counties', 13, 44) ### +### ('familiarity', 14, 27223) ('plural', 15, 9410) ('alabama', 16, 1219) ('hampshire', 17, 5603) ### +### ('virginia', 18, 566) ('city', 19, 5221) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('hopkins', 0, 0) ('unincorporated', 3871, 1) ('columbia', 55, 2) ('sponsored', 11274, 3) ### +### ('topics', 2353, 4) ('encompasses', 308, 5) ('carolina', 2, 6) ('rich', 1736, 7) ('topic', 5443, 8) ### +### ('metropolitan', 3849, 9) ('founded', 195, 10) ('john', 923, 11) ('ˈ', 1243, 12) ### +### ('downtown', 360, 13) ('crashing', 1209, 14) ('county', 1, 15) ('##ང', 697, 16) ('named', 3384, 17) ### +### ('circa', 11483, 18) ('gideon', 296, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('hopkins', 0, 0) ('county', 1, 15) ('carolina', 2, 6) ('sc', 3, 24) ('counties', 13, 44) ### +### ('columbia', 55, 2) ('is', 10, 125) ('encompasses', 308, 5) ('founded', 195, 10) ### +### ('santiago', 96, 34) ('mississippi', 4, 1453) ('south', 182, 22) ('florida', 5, 841) ### +### ('simon', 87, 66) ('hawkins', 32, 161) ('stark', 146, 42) ('edwards', 26, 219) ### +### ('unwilling', 251, 20) ('downtown', 360, 13) ('gideon', 296, 19) ### +############################################################################################################ +[2023-10-07 22:14:02,915][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:14:02,915][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:14:03,316][root][INFO] - Epoch: 9: Step: 301/1557, loss[v]=0.089122, lr=0.000011, acc@1[1]=242.0/256=0.9453125, acc@1[2]=248.5/256=0.970703125 +[2023-10-07 22:15:20,073][root][INFO] - Train batch 400 +[2023-10-07 22:15:20,073][root][INFO] - Avg. loss per last 100 batches: 0.073548 +[2023-10-07 22:15:20,811][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29162.6/29522=98.78% | mean: 0.01 | max: 5.25 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 5.93 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how difficult is it to become certified ethical hacker [SEP] ### +### [P_TEXT]: [CLS] certified ethical hacker. certified ethical hacker ( ceh ) is a qualification ### +### obtained by assessing the security of computer systems, using penetration testing techniques. the ### +### code for the ceh exam is 312 - 50, and the certification is in version 9 as of 2016. [SEP] ### +### ======================================= h_v_q | Gates: 27250 ======================================= ### +### ('hacker', 0, 0) ('ethical', 1, 1) ('difficult', 2, 3009) ('certified', 3, 4) ('ethics', 4, 9) ### +### ('become', 5, 347) ('.', 6, 7613) ('moral', 7, 37) ('impossible', 8, 635) ('becoming', 9, 6416) ### +### ('degree', 10, 622) ('hard', 11, 8948) ('certification', 12, 8) ('familiarity', 13, 26934) ### +### ('easier', 14, 3453) ('became', 15, 6009) ('difficulties', 16, 9373) ('dangerous', 17, 2672) ### +### ('graduate', 18, 851) ('difficulty', 19, 13448) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('hacker', 0, 0) ('ethical', 1, 1) ('ce', 9074, 2) ('##h', 9460, 3) ('certified', 3, 4) ### +### ('penetration', 11051, 5) ('encompasses', 88, 6) ('qualification', 732, 7) ('certification', 12, 8) ### +### ('ethics', 4, 9) ('ˈ', 295, 10) ('crashing', 269, 11) ('testing', 2252, 12) ('define', 19688, 13) ### +### ('##₂', 120, 14) ('definition', 488, 15) ('techniques', 853, 16) ('##α', 491, 17) ### +### ('computer', 24, 18) ('##ང', 715, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('hacker', 0, 0) ('ethical', 1, 1) ('certified', 3, 4) ('ethics', 4, 9) ('moral', 7, 37) ### +### ('become', 5, 347) ('certification', 12, 8) ('difficult', 2, 3009) ('computer', 24, 18) ### +### ('30', 36, 21) ('impossible', 8, 635) ('degree', 10, 622) ('encompasses', 88, 6) ### +### ('training', 29, 64) ('software', 20, 163) ('harsh', 22, 218) ('##₂', 120, 14) ('.', 6, 7613) ### +### ('hacking', 55, 58) ('is', 25, 276) ### +############################################################################################################ +[2023-10-07 22:15:20,812][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:15:20,812][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:15:21,234][root][INFO] - Epoch: 9: Step: 401/1557, loss[v]=0.083761, lr=0.000011, acc@1[1]=238.5/256=0.931640625, acc@1[2]=247.0/256=0.96484375 +[2023-10-07 22:16:37,481][root][INFO] - Train batch 500 +[2023-10-07 22:16:37,481][root][INFO] - Avg. loss per last 100 batches: 0.074997 +[2023-10-07 22:16:38,190][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29224.2/29522=98.99% | mean: 0.01 | max: 5.57 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.17 | max: 6.11 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] who discovered plastic [SEP] ### +### [P_TEXT]: [CLS] invention of plastic. the first person who invented plastic was alexander parkes, a ### +### native of birmingham, england. he called his work parkesine after himself. the plastic he invented ### +### was organic. it was made from cellulose treated with a solvent and nitric acid. it could be molded ### +### into any shape when heated. [SEP] ### +### ======================================= h_v_q | Gates: 27508 ======================================= ### +### ('plastic', 0, 0) ('discovered', 1, 62) ('whose', 2, 59) ('.', 3, 10844) ('who', 4, 51) ### +### ('found', 5, 742) ('plastics', 6, 19) ('founded', 7, 87) ('familiarity', 8, 27486) ### +### ('discovery', 9, 247) ('established', 10, 255) ('became', 11, 173) ('knew', 12, 74) ### +### ('invented', 13, 3) ('opened', 14, 6755) ('metal', 15, 361) ('developed', 16, 439) ### +### ('plural', 17, 5043) ('discover', 18, 455) ('introduced', 19, 619) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('plastic', 0, 0) ('park', 1724, 1) ('organic', 5089, 2) ('invented', 13, 3) ### +### ('birmingham', 10003, 4) ('parks', 3949, 5) ('molded', 2598, 6) ('alexander', 385, 7) ('ˈ', 703, 8) ### +### ('invention', 1417, 9) ('made', 376, 10) ('##ine', 9930, 11) ('afraid', 304, 12) ('##ο', 275, 13) ### +### ('unwilling', 265, 14) ('##ང', 364, 15) ('person', 137, 16) ('first', 1625, 17) ### +### ('sharply', 800, 18) ('plastics', 6, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('plastic', 0, 0) ('discovered', 1, 62) ('whose', 2, 59) ('who', 4, 51) ('plastics', 6, 19) ### +### ('invented', 13, 3) ('founded', 7, 87) ('knew', 12, 74) ('became', 11, 173) ('discovery', 9, 247) ### +### ('established', 10, 255) ('found', 5, 742) ('created', 30, 65) ('early', 66, 37) ### +### ('person', 137, 16) ('born', 21, 269) ('−', 118, 35) ('julian', 58, 93) ('metal', 15, 361) ### +### ('alexander', 385, 7) ### +############################################################################################################ +[2023-10-07 22:16:38,191][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:16:38,191][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:16:38,615][root][INFO] - Epoch: 9: Step: 501/1557, loss[v]=0.143841, lr=0.000011, acc@1[1]=237.0/256=0.92578125, acc@1[2]=248.5/256=0.970703125 +[2023-10-07 22:17:54,749][root][INFO] - Train batch 600 +[2023-10-07 22:17:54,749][root][INFO] - Avg. loss per last 100 batches: 0.077593 +[2023-10-07 22:17:55,481][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29102.0/29522=98.58% | mean: 0.01 | max: 5.49 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.13 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the history of for safety glasses [SEP] ### +### [P_TEXT]: [CLS] the history of safety glasses. safety glasses should be worn by personnel in ### +### manufacturing, construction, medical, some service industries and sometimes sports. although eye ### +### injury is a higher risk for these individuals, many people do not wear eye protection for one ### +### reason or another. [SEP] ### +### ======================================= h_v_q | Gates: 27387 ======================================= ### +### ('safety', 0, 1) ('glasses', 1, 0) ('history', 2, 9) ('glass', 3, 3) ('historical', 4, 25) ### +### ('.', 5, 12764) ('heritage', 6, 170) ('familiarity', 7, 27896) ('something', 8, 3234) ### +### ('for', 9, 4530) ('century', 10, 875) ('histories', 11, 155) ('1998', 12, 9121) ### +### ('sunglasses', 13, 12) ('is', 14, 12471) ('2017', 15, 20318) ('1997', 16, 20783) ### +### ('evolution', 17, 896) ('life', 18, 4441) ('relating', 19, 22877) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('glasses', 1, 0) ('safety', 0, 1) ('worn', 4300, 2) ('glass', 3, 3) ('eye', 4842, 4) ('ˈ', 497, 5) ### +### ('wear', 2064, 6) ('protection', 40, 7) ('personnel', 625, 8) ('history', 2, 9) ('##₂', 151, 10) ### +### ('unwilling', 199, 11) ('sunglasses', 13, 12) ('−', 192, 13) ('medical', 943, 14) ('##α', 841, 15) ### +### ('sharply', 884, 16) ('industries', 1498, 17) ('hating', 269, 18) ('wearing', 678, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('glasses', 1, 0) ('safety', 0, 1) ('history', 2, 9) ('glass', 3, 3) ('historical', 4, 25) ### +### ('sunglasses', 13, 12) ('heritage', 6, 170) ('protection', 40, 7) ('histories', 11, 155) ### +### ('safe', 33, 55) ('prehistoric', 50, 43) ('altogether', 72, 31) ('eyes', 90, 27) ('simon', 36, 84) ### +### ('##₂', 151, 10) ('unwilling', 199, 11) ('−', 192, 13) ('mirror', 35, 144) ('helmet', 58, 112) ### +### ('postwar', 89, 68) ### +############################################################################################################ +[2023-10-07 22:17:55,482][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:17:55,482][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:17:55,902][root][INFO] - Epoch: 9: Step: 601/1557, loss[v]=0.081361, lr=0.000011, acc@1[1]=242.0/256=0.9453125, acc@1[2]=250.5/256=0.978515625 +[2023-10-07 22:19:12,633][root][INFO] - Train batch 700 +[2023-10-07 22:19:12,634][root][INFO] - Avg. loss per last 100 batches: 0.070884 +[2023-10-07 22:19:13,313][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29124.0/29522=98.65% | mean: 0.01 | max: 5.40 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.13 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] can moldovans work in the uk [SEP] ### +### [P_TEXT]: [CLS] 18 - 45 year olds from moldova are able to apply for a work visa to the uk. ### +### successful applicants can migrate and work in the uk permanently or temporarily. hy us. established ### +### in 2002, migration expert is a private immigration company with a team of highly experienced oisc ### +### registered migration consultants who represent clients from across the globe wishing to apply for a ### +### uk visa. [SEP] ### +### ======================================= h_v_q | Gates: 27420 ======================================= ### +### ('moldova', 0, 1) ('uk', 1, 4) ('##ns', 2, 7742) ('work', 3, 14) ('england', 4, 149) ### +### ('familiarity', 5, 22530) ('can', 6, 740) ('british', 7, 212) ('.', 8, 8139) ('works', 9, 402) ### +### ('london', 10, 110) ('britain', 11, 97) ('working', 12, 75) ('couldn', 13, 556) ### +### ('hampshire', 14, 9802) ('job', 15, 176) ('kingdom', 16, 1704) ('bbc', 17, 101) ### +### ('europe', 18, 3282) ('relating', 19, 26612) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('migration', 5272, 0) ('moldova', 0, 1) ('immigration', 2272, 2) ('visa', 13529, 3) ('uk', 1, 4) ### +### ('expert', 11269, 5) ('permanently', 8090, 6) ('visas', 14995, 7) ('45', 3244, 8) ### +### ('migrant', 13923, 9) ('ˈ', 217, 10) ('applicants', 7053, 11) ('temporarily', 7666, 12) ### +### ('##is', 4714, 13) ('work', 3, 14) ('##y', 8403, 15) ('private', 833, 16) ('expertise', 761, 17) ### +### ('##α', 375, 18) ('##₂', 263, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('moldova', 0, 1) ('uk', 1, 4) ('work', 3, 14) ('england', 4, 149) ('british', 7, 212) ### +### ('london', 10, 110) ('working', 12, 75) ('britain', 11, 97) ('can', 6, 740) ('bbc', 17, 101) ### +### ('##ns', 2, 7742) ('works', 9, 402) ('job', 15, 176) ('able', 26, 69) ('us', 55, 30) ('−', 81, 29) ### +### ('couldn', 13, 556) ('australia', 21, 310) ('operate', 25, 245) ('english', 38, 156) ### +############################################################################################################ +[2023-10-07 22:19:13,313][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:19:13,313][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:19:13,731][root][INFO] - Epoch: 9: Step: 701/1557, loss[v]=0.071141, lr=0.000011, acc@1[1]=241.0/256=0.94140625, acc@1[2]=249.5/256=0.974609375 +[2023-10-07 22:20:29,948][root][INFO] - Train batch 800 +[2023-10-07 22:20:29,948][root][INFO] - Avg. loss per last 100 batches: 0.071978 +[2023-10-07 22:20:30,675][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29185.7/29522=98.86% | mean: 0.01 | max: 4.91 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.5/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.08 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how early should a fetus show on an ultrasound [SEP] ### +### [P_TEXT]: [CLS] detecting a fetus on ultrasound. a fetus can be detected as early as the sixth week ### +### of pregnancy. many ultrasound technicians prefer to wait until at least that point in the pregnancy ### +### to do an ultrasound and many prefer waiting until the eighth week of pregnancy when more of the ### +### development of the fetus can be seen. finding a fetal heartbeat. [SEP] ### +### ======================================= h_v_q | Gates: 27753 ======================================= ### +### ('ultrasound', 0, 0) ('fe', 1, 10) ('early', 2, 7) ('show', 3, 4747) ('##tus', 4, 6) ### +### ('familiarity', 5, 26027) ('weeks', 6, 5) ('days', 7, 484) ('should', 8, 16293) ('.', 9, 12252) ### +### ('months', 10, 674) ('age', 11, 84) ('shows', 12, 864) ('years', 13, 2718) ('hours', 14, 14175) ### +### ('minutes', 15, 342) ('relating', 16, 25919) ('recommended', 17, 1412) ('must', 18, 8188) ### +### ('late', 19, 50) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('ultrasound', 0, 0) ('heartbeat', 9735, 1) ('pregnancy', 3447, 2) ('detected', 19759, 3) ### +### ('detect', 5750, 4) ('weeks', 6, 5) ('##tus', 4, 6) ('early', 2, 7) ('fetal', 1397, 8) ### +### ('ˈ', 104, 9) ('fe', 1, 10) ('wingspan', 372, 11) ('detecting', 25435, 12) ('detection', 11360, 13) ### +### ('santiago', 190, 14) ('cyrillic', 328, 15) ('##ང', 169, 16) ('soon', 267, 17) ('##ο', 197, 18) ### +### ('hating', 71, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ultrasound', 0, 0) ('fe', 1, 10) ('early', 2, 7) ('##tus', 4, 6) ('weeks', 6, 5) ('age', 11, 84) ### +### ('days', 7, 484) ('late', 19, 50) ('months', 10, 674) ('onto', 22, 80) ('week', 32, 25) ### +### ('minutes', 15, 342) ('show', 3, 4747) ('shows', 12, 864) ('simon', 28, 69) ('unwilling', 52, 23) ### +### ('until', 57, 29) ('30', 21, 178) ('hating', 71, 19) ('ˈ', 104, 9) ### +############################################################################################################ +[2023-10-07 22:20:30,675][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:20:30,675][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:20:31,081][root][INFO] - Epoch: 9: Step: 801/1557, loss[v]=0.076358, lr=0.000011, acc@1[1]=240.5/256=0.939453125, acc@1[2]=246.0/256=0.9609375 +[2023-10-07 22:21:47,067][root][INFO] - Train batch 900 +[2023-10-07 22:21:47,067][root][INFO] - Avg. loss per last 100 batches: 0.071115 +[2023-10-07 22:21:47,766][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29111.0/29522=98.61% | mean: 0.01 | max: 5.54 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.18 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] enclave meaning [SEP] ### +### [P_TEXT]: [CLS] suggest new translation / definition enclave ( enclaves plural ) an enclave is an ### +### area within a country or a city where people live who have a different nationality or culture from ### +### the people living in the surrounding country or city. n - count usu with supp nagorno - karabakh is ### +### an armenian enclave inside azerbaijan. [SEP] ### +### ======================================= h_v_q | Gates: 26050 ======================================= ### +### ('enclave', 0, 0) ('.', 1, 4593) ('definition', 2, 9) ('or', 3, 17385) ('meaning', 4, 24) ### +### ('group', 5, 670) ('noun', 6, 22913) ('familiarity', 7, 26956) (';', 8, 1489) ### +### ('something', 9, 4554) ('relating', 10, 24588) ('means', 11, 245) ('circle', 12, 236) ### +### ('island', 13, 155) ('plural', 14, 13) ('division', 15, 1476) ('complex', 16, 605) ### +### ('sense', 17, 7671) ('entity', 18, 112) ('refers', 19, 13377) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('enclave', 0, 0) ('armenian', 4557, 1) ('suggest', 7952, 2) ('kara', 9719, 3) ### +### ('nationality', 10992, 4) ('azerbaijan', 2923, 5) ('definitions', 1122, 6) ('armenia', 9046, 7) ### +### ('define', 6792, 8) ('definition', 2, 9) ('##gor', 22061, 10) ('na', 7951, 11) ### +### ('suggests', 10163, 12) ('plural', 14, 13) ('living', 426, 14) ('ˈ', 1900, 15) ### +### ('encompasses', 213, 16) ('crashing', 886, 17) ('suggestion', 10292, 18) ('azerbaijani', 19774, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('enclave', 0, 0) ('definition', 2, 9) ('meaning', 4, 24) ('plural', 14, 13) ('group', 5, 670) ### +### ('defined', 20, 77) ('area', 23, 47) ('island', 13, 155) ('entity', 18, 112) ('means', 11, 245) ### +### ('circle', 12, 236) ('array', 43, 46) ('country', 91, 20) ('city', 27, 110) ('culture', 73, 34) ### +### ('.', 1, 4593) (';', 8, 1489) ('encompasses', 213, 16) ('complex', 16, 605) ('colony', 22, 321) ### +############################################################################################################ +[2023-10-07 22:21:47,767][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:21:47,767][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:21:48,185][root][INFO] - Epoch: 9: Step: 901/1557, loss[v]=0.052751, lr=0.000011, acc@1[1]=245.0/256=0.95703125, acc@1[2]=252.5/256=0.986328125 +[2023-10-07 22:23:04,494][root][INFO] - Train batch 1000 +[2023-10-07 22:23:04,495][root][INFO] - Avg. loss per last 100 batches: 0.074146 +[2023-10-07 22:23:05,172][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29247.1/29522=99.07% | mean: 0.01 | max: 5.39 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.22 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] when do you use a fire extinguisher [SEP] ### +### [P_TEXT]: [CLS] how to use a fire extinguisher. you should know exactly how to use a fire ### +### extinguisher in the event a fire develops and you feel you are safely able to fight it. it is ### +### recommended that only those trained in the proper use of fire extinguishers consider using them ### +### when appropriate. call for help before attempting to extinguish a fire. [SEP] ### +### ======================================= h_v_q | Gates: 28360 ======================================= ### +### ('fire', 0, 0) ('##ting', 1, 54) ('ex', 2, 39) ('##her', 3, 47) ('use', 4, 33) ### +### ('familiarity', 5, 21755) ('##uis', 6, 156) ('when', 7, 874) ('.', 8, 15939) ('you', 9, 1940) ### +### ('2017', 10, 23609) ('uses', 11, 42) ('relating', 12, 23337) ('used', 13, 149) ('jeremy', 14, 66) ### +### ('viewers', 15, 244) ('weeks', 16, 2333) ('consisting', 17, 18981) ('answer', 18, 11272) ### +### ('unwilling', 19, 4) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('fire', 0, 0) ('##hers', 1876, 1) ('fires', 312, 2) ('safely', 10040, 3) ('unwilling', 19, 4) ### +### ('ˈ', 95, 5) ('##ང', 68, 6) ('consider', 8079, 7) ('hating', 86, 8) ('##₂', 49, 9) ### +### ('santiago', 158, 10) ('altogether', 85, 11) ('hesitated', 47, 12) ('firefighters', 5782, 13) ### +### ('##ο', 99, 14) ('crashing', 71, 15) ('##α', 59, 16) ('−', 37, 17) ('##fires', 9097, 18) ### +### ('flame', 285, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('fire', 0, 0) ('ex', 2, 39) ('##ting', 1, 54) ('##her', 3, 47) ('use', 4, 33) ('##uis', 6, 156) ### +### ('uses', 11, 42) ('unwilling', 19, 4) ('jeremy', 14, 66) ('used', 13, 149) ('−', 37, 17) ### +### ('hesitated', 47, 12) ('##₂', 49, 9) ('simon', 29, 51) ('viewers', 15, 244) ('##ང', 68, 6) ### +### ('##α', 59, 16) ('when', 7, 874) ('ˈ', 95, 5) ('crashing', 71, 15) ### +############################################################################################################ +[2023-10-07 22:23:05,172][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:23:05,172][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:23:05,594][root][INFO] - Epoch: 9: Step: 1001/1557, loss[v]=0.044158, lr=0.000011, acc@1[1]=243.0/256=0.94921875, acc@1[2]=253.5/256=0.990234375 +[2023-10-07 22:24:21,936][root][INFO] - Train batch 1100 +[2023-10-07 22:24:21,936][root][INFO] - Avg. loss per last 100 batches: 0.070262 +[2023-10-07 22:24:22,621][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29152.7/29522=98.75% | mean: 0.01 | max: 5.20 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.01 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] budget amendment definition [SEP] ### +### [P_TEXT]: [CLS] budget amendment definition, budget amendment meaning | english dictionary. search ### +### also in : web news encyclopedia images. budget. n. 1 an itemized summary of expected income and ### +### expenditure of a country, company, etc., over a specified period, usually a financial year. [SEP] ### +### ======================================= h_v_q | Gates: 25853 ======================================= ### +### ('budget', 0, 0) ('amendment', 1, 1) ('definition', 2, 4) ('noun', 3, 19846) ('.', 4, 6731) ### +### ('something', 5, 2461) ('defined', 6, 36) ('means', 7, 10) ('familiarity', 8, 27101) ### +### ('relating', 9, 25535) ('meaning', 10, 5) ('bill', 11, 76) ('plural', 12, 9677) ('act', 13, 4952) ### +### ('refers', 14, 6542) ('or', 15, 16562) ('budgets', 16, 7) ('latin', 17, 2698) ### +### ('government', 18, 232) ('term', 19, 3952) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('budget', 0, 0) ('amendment', 1, 1) ('expenditure', 2184, 2) ('definitions', 48, 3) ### +### ('definition', 2, 4) ('meaning', 10, 5) ('define', 689, 6) ('budgets', 16, 7) ('amendments', 33, 8) ### +### ('ˈ', 273, 9) ('means', 7, 10) ('expected', 8366, 11) ('crashing', 587, 12) ### +### ('encyclopedia', 20, 13) ('financial', 240, 14) ('summary', 10043, 15) ('country', 1262, 16) ### +### ('mean', 129, 17) ('finances', 3628, 18) ('gideon', 258, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('budget', 0, 0) ('amendment', 1, 1) ('definition', 2, 4) ('means', 7, 10) ('defined', 6, 36) ### +### ('meaning', 10, 5) ('budgets', 16, 7) ('encyclopedia', 20, 13) ('bill', 11, 76) ### +### ('definitions', 48, 3) ('amendments', 33, 8) ('dictionary', 24, 32) ('specified', 36, 39) ### +### ('english', 52, 20) ('simon', 42, 37) ('something', 5, 2461) ('unwilling', 95, 21) ### +### ('mean', 129, 17) ('income', 105, 25) ('government', 18, 232) ### +############################################################################################################ +[2023-10-07 22:24:22,621][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:24:22,622][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:24:23,043][root][INFO] - Epoch: 9: Step: 1101/1557, loss[v]=0.082786, lr=0.000011, acc@1[1]=241.0/256=0.94140625, acc@1[2]=248.5/256=0.970703125 +[2023-10-07 22:25:39,648][root][INFO] - Train batch 1200 +[2023-10-07 22:25:39,648][root][INFO] - Avg. loss per last 100 batches: 0.075220 +[2023-10-07 22:25:40,379][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29155.9/29522=98.76% | mean: 0.01 | max: 5.17 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.17 | max: 5.97 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] causes for dental abfraction [SEP] ### +### [P_TEXT]: [CLS] 2013 2 apr. a dental abfraction is a notched - out area on the root of a tooth at ### +### the gumline. there are several causes for this. it can be caused by toothbrush wear over a period ### +### of timea¦usually by vigorous brushing in certain areas or by the use of a hard - bristled ### +### toothbrush. [SEP] ### +### ======================================= h_v_q | Gates: 27408 ======================================= ### +### ('dental', 0, 2) ('ab', 1, 8) ('##fra', 2, 3) ('##ction', 3, 26) ('causes', 4, 18) ### +### ('cause', 5, 106) ('caused', 6, 352) ('.', 7, 11449) ('familiarity', 8, 27601) ('for', 9, 1440) ### +### ('relating', 10, 24821) ('medical', 11, 4308) ('health', 12, 8204) ('action', 13, 2424) ### +### ('sources', 14, 3582) ('...', 15, 8292) ('consisting', 16, 26715) ('factors', 17, 1273) ### +### ('reasons', 18, 424) ('division', 19, 1757) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('notch', 17811, 0) ('tooth', 350, 1) ('dental', 0, 2) ('##fra', 2, 3) ('teeth', 843, 4) ### +### ('##brush', 15464, 5) ('##¦', 29511, 6) ('wear', 6804, 7) ('ab', 1, 8) ('##line', 3874, 9) ### +### ('encompasses', 1237, 10) ('gum', 2196, 11) ('root', 1663, 12) ('sharply', 165, 13) ### +### ('areas', 2399, 14) ('ˈ', 148, 15) ('brushing', 4202, 16) ('gideon', 107, 17) ('causes', 4, 18) ### +### ('##ང', 419, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('dental', 0, 2) ('##fra', 2, 3) ('ab', 1, 8) ('causes', 4, 18) ('##ction', 3, 26) ### +### ('cause', 5, 106) ('caused', 6, 352) ('dentist', 22, 21) ('simon', 24, 48) ('−', 31, 33) ### +### ('tooth', 350, 1) ('##₂', 44, 41) ('hesitated', 61, 29) ('gideon', 107, 17) ('tina', 38, 62) ### +### ('for', 9, 1440) ('unwilling', 78, 34) ('ˈ', 148, 15) ('sharply', 165, 13) ('bare', 40, 84) ### +############################################################################################################ +[2023-10-07 22:25:40,380][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:25:40,380][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:25:40,801][root][INFO] - Epoch: 9: Step: 1201/1557, loss[v]=0.127610, lr=0.000011, acc@1[1]=241.0/256=0.94140625, acc@1[2]=248.0/256=0.96875 +[2023-10-07 22:26:58,364][root][INFO] - Train batch 1300 +[2023-10-07 22:26:58,365][root][INFO] - Avg. loss per last 100 batches: 0.078613 +[2023-10-07 22:26:59,098][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29091.9/29522=98.54% | mean: 0.01 | max: 5.31 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.00 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] is hpv a std [SEP] ### +### [P_TEXT]: [CLS] genital human papillomavirus ( or hpv ) is the most common sexually transmitted ### +### disease ( std, commonly known as sexually transmitted infection, sti ). some types of genital hpv ### +### primarily infect skin near the genitals and anus. ome types of sexually transmitted hpv cause ### +### genital or anal warts. warts appear as growths or bumps and may be raised or flat, single or ### +### multiple, small or large. they tend to be flesh - colored or whitish in appearance. [SEP] ### +### ======================================= h_v_q | Gates: 27472 ======================================= ### +### ('hp', 0, 0) ('##v', 1, 7) ('st', 2, 5) ('##d', 3, 17) ('is', 4, 721) ('relating', 5, 25232) ### +### ('virginia', 6, 4006) ('.', 7, 15383) ('saint', 8, 127) ('familiarity', 9, 28328) ('a', 10, 23652) ### +### ('consisting', 11, 24213) ('##va', 12, 252) ('plural', 13, 11243) ('design', 14, 6120) ### +### ('encompasses', 15, 6) ('ward', 16, 5493) ('provides', 17, 7765) ('government', 18, 4897) ### +### ('stands', 19, 5380) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('hp', 0, 0) ('##ital', 24087, 1) ('anal', 15674, 2) ('##virus', 5243, 3) ('sexually', 23712, 4) ### +### ('st', 2, 5) ('encompasses', 15, 6) ('##v', 1, 7) ('transmitted', 14270, 8) ('gen', 6600, 9) ### +### ('types', 4377, 10) ('hating', 83, 11) ('##i', 5267, 12) ('infection', 5651, 13) ('om', 7964, 14) ### +### ('gideon', 57, 15) ('disease', 849, 16) ('##d', 3, 17) ('war', 510, 18) ('ˈ', 144, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('hp', 0, 0) ('##v', 1, 7) ('st', 2, 5) ('##d', 3, 17) ('encompasses', 15, 6) ('saint', 8, 127) ### +### ('is', 4, 721) ('unwilling', 40, 23) ('gideon', 57, 15) ('hesitated', 39, 29) ('beside', 29, 47) ### +### ('afraid', 31, 51) ('##va', 12, 252) ('hating', 83, 11) ('crashing', 63, 27) ('simon', 33, 73) ### +### ('santiago', 98, 25) ('ˈ', 144, 19) ('##ང', 112, 35) ('definition', 43, 98) ### +############################################################################################################ +[2023-10-07 22:26:59,098][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:26:59,098][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:26:59,522][root][INFO] - Epoch: 9: Step: 1301/1557, loss[v]=0.040618, lr=0.000011, acc@1[1]=245.0/256=0.95703125, acc@1[2]=254.0/256=0.9921875 +[2023-10-07 22:28:16,743][root][INFO] - Train batch 1400 +[2023-10-07 22:28:16,744][root][INFO] - Avg. loss per last 100 batches: 0.073220 +[2023-10-07 22:28:17,446][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29237.0/29522=99.03% | mean: 0.01 | max: 5.52 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.18 | max: 6.28 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the definition of a nerd [SEP] ### +### [P_TEXT]: [CLS] a nerd is described as'a foolish or contemptible person who lacks social skills or ### +### is boringly studious. '.'a single - minded expert in a particular technical field'and'a computer ### +### nerd. '. the first known use of the word nerd is quoted as the name of a creature in dr. seuss's ### +### book if i ran the zoo from 1950. [SEP] ### +### ======================================= h_v_q | Gates: 28019 ======================================= ### +### ('##rd', 0, 1) ('ne', 1, 5) ('definition', 2, 10) ('##rds', 3, 6) ('noun', 4, 18074) ### +### ('relating', 5, 23778) ('plural', 6, 6389) ('defined', 7, 147) ('familiarity', 8, 24815) ### +### ('a', 9, 542) ('consisting', 10, 24435) ('encompasses', 11, 13) ('refers', 12, 7169) ### +### ('something', 13, 3391) ('or', 14, 20726) ('term', 15, 464) ('##rt', 16, 106) ### +### ('stylized', 17, 19544) ('is', 18, 682) ('definitions', 19, 29) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('zoo', 3720, 0) ('##rd', 0, 1) ('boring', 2863, 2) ('studio', 133, 3) ('contempt', 7023, 4) ### +### ('ne', 1, 5) ('##rds', 3, 6) ('foolish', 9460, 7) ('define', 581, 8) ('creature', 7916, 9) ### +### ('definition', 2, 10) ('minded', 23141, 11) ('ˈ', 32, 12) ('encompasses', 11, 13) ### +### ('quoted', 5715, 14) ('crashing', 64, 15) ('bored', 18967, 16) ('wingspan', 174, 17) ### +### ('single', 846, 18) ('gideon', 70, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##rd', 0, 1) ('ne', 1, 5) ('definition', 2, 10) ('##rds', 3, 6) ('encompasses', 11, 13) ### +### ('defined', 7, 147) ('definitions', 19, 29) ('meaning', 20, 28) ('ˈ', 32, 12) ('##rt', 16, 106) ### +### ('##₂', 38, 25) ('a', 9, 542) ('unwilling', 30, 39) ('studio', 133, 3) ('crashing', 64, 15) ### +### ('##ང', 53, 30) ('gideon', 70, 19) ('means', 21, 169) ('term', 15, 464) ('angrily', 73, 40) ### +############################################################################################################ +[2023-10-07 22:28:17,446][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:28:17,446][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:28:17,851][root][INFO] - Epoch: 9: Step: 1401/1557, loss[v]=0.075800, lr=0.000011, acc@1[1]=246.5/256=0.962890625, acc@1[2]=251.5/256=0.982421875 +[2023-10-07 22:29:34,420][root][INFO] - Train batch 1500 +[2023-10-07 22:29:34,421][root][INFO] - Avg. loss per last 100 batches: 0.070701 +[2023-10-07 22:29:35,154][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29141.9/29522=98.71% | mean: 0.01 | max: 5.41 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.18 | max: 6.15 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what heals skin sores fast [SEP] ### +### [P_TEXT]: [CLS] more views. intense fast healinga® cream. quick overview. our best selling cream ### +### for the past 20 years. this unique, multi - purpose cream contains more than 200 healing properties ### +### to help provide faster healing for minor to severely damaged skin from cuts, scrapes, rashes, ### +### blisters, burns, sunburns, sores, and other hard - to - heal skin irritations. [SEP] ### +### ======================================= h_v_q | Gates: 27639 ======================================= ### +### ('sore', 0, 9) ('heal', 1, 2) ('skin', 2, 8) ('fast', 3, 3) ('healing', 4, 0) ('.', 5, 21705) ### +### ('##s', 6, 748) ('speed', 7, 107) ('familiarity', 8, 27819) ('quick', 9, 97) ('healed', 10, 47) ### +### ('relating', 11, 24433) ('slow', 12, 463) ('tight', 13, 319) ('repair', 14, 235) ### +### ('consisting', 15, 26521) ('simon', 16, 79) ('faster', 17, 15) ('refers', 18, 22805) ### +### ('answer', 19, 17514) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('healing', 4, 0) ('cream', 8435, 1) ('heal', 1, 2) ('fast', 3, 3) ('intense', 255, 4) ### +### ('selling', 11785, 5) ('ˈ', 66, 6) ('burns', 496, 7) ('skin', 2, 8) ('sore', 0, 9) ### +### ('##®', 11677, 10) ('overview', 13680, 11) ('irritation', 4600, 12) ('damaged', 1262, 13) ### +### ('unique', 3591, 14) ('faster', 17, 15) ('hating', 103, 16) ('##α', 85, 17) ('helps', 851, 18) ### +### ('unwilling', 38, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('heal', 1, 2) ('sore', 0, 9) ('fast', 3, 3) ('skin', 2, 8) ('healing', 4, 0) ('healed', 10, 47) ### +### ('speed', 7, 107) ('quick', 9, 97) ('##s', 6, 748) ('faster', 17, 15) ('simon', 16, 79) ### +### ('annoyance', 24, 32) ('ˈ', 66, 6) ('unwilling', 38, 19) ('tight', 13, 319) ('slow', 12, 463) ### +### ('repair', 14, 235) ('##α', 85, 17) ('##₂', 51, 36) ('dante', 33, 60) ### +############################################################################################################ +[2023-10-07 22:29:35,155][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:29:35,155][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:29:35,583][root][INFO] - Epoch: 9: Step: 1501/1557, loss[v]=0.054554, lr=0.000011, acc@1[1]=246.0/256=0.9609375, acc@1[2]=252.0/256=0.984375 +[2023-10-07 22:30:18,522][root][INFO] - rank=2; last iteration 1557 +[2023-10-07 22:30:18,523][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 22:30:18,523][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-07 22:30:18,523][root][INFO] - rank=1; last iteration 1557 +[2023-10-07 22:30:18,523][root][INFO] - rank=3; last iteration 1557 +[2023-10-07 22:30:18,523][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 22:30:18,523][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 22:30:18,523][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-07 22:30:18,524][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-07 22:30:18,525][root][INFO] - rank=0; last iteration 1557 +[2023-10-07 22:30:18,525][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 22:30:18,525][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-07 22:30:18,531][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 22:30:18,531][root][INFO] - Epoch finished on 2 +[2023-10-07 22:30:18,531][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 22:30:18,532][root][INFO] - Epoch finished on 1 +[2023-10-07 22:30:18,533][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 22:30:18,533][root][INFO] - Epoch finished on 3 +[2023-10-07 22:30:18,533][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 22:30:18,533][root][INFO] - Epoch finished on 0 +[2023-10-07 22:30:35,781][root][INFO] - Saved checkpoint at ./vdr_9 +[2023-10-07 22:30:35,781][root][INFO] - Saved checkpoint at ./vdr_9 +[2023-10-07 22:30:35,782][root][INFO] - Av Loss per epoch=0.073400 +[2023-10-07 22:30:35,782][root][INFO] - epoch total (1) correct predictions=377705 +[2023-10-07 22:30:35,782][root][INFO] - Av Loss per epoch=0.073400 +[2023-10-07 22:30:35,782][root][INFO] - epoch total (2) correct predictions=390020 +[2023-10-07 22:30:35,782][root][INFO] - epoch total (1) correct predictions=377705 +[2023-10-07 22:30:35,782][root][INFO] - Saved checkpoint at ./vdr_9 +[2023-10-07 22:30:35,782][root][INFO] - epoch total (2) correct predictions=390020 +[2023-10-07 22:30:35,782][root][INFO] - Av Loss per epoch=0.073400 +[2023-10-07 22:30:35,783][root][INFO] - epoch total (1) correct predictions=377705 +[2023-10-07 22:30:35,783][root][INFO] - epoch total (2) correct predictions=390020 +[2023-10-07 22:30:35,785][root][INFO] - Saved checkpoint at ./vdr_9 +[2023-10-07 22:30:35,786][root][INFO] - Av Loss per epoch=0.073400 +[2023-10-07 22:30:35,786][root][INFO] - epoch total (1) correct predictions=377705 +[2023-10-07 22:30:35,786][root][INFO] - epoch total (2) correct predictions=390020 +[2023-10-07 22:30:35,786][root][INFO] - ***** Epoch 10 ***** +[2023-10-07 22:30:35,786][root][INFO] - ***** Epoch 10 ***** +[2023-10-07 22:30:35,790][root][INFO] - rank=2; Iteration start +[2023-10-07 22:30:35,790][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 22:30:35,790][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 22:30:35,791][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-07 22:30:35,790][root][INFO] - ***** Epoch 10 ***** +[2023-10-07 22:30:35,793][root][INFO] - rank=0; Iteration start +[2023-10-07 22:30:35,793][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 22:30:35,793][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 22:30:35,793][root][INFO] - ***** Epoch 10 ***** +[2023-10-07 22:30:35,795][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-07 22:30:35,798][root][INFO] - rank=1; Iteration start +[2023-10-07 22:30:35,798][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 22:30:35,798][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 22:30:35,799][root][INFO] - rank=3; Iteration start +[2023-10-07 22:30:35,799][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 22:30:35,800][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 22:30:35,800][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-07 22:30:35,802][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-07 22:30:36,786][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29144.6/29522=98.72% | mean: 0.01 | max: 5.52 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.18 | max: 6.37 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] is there a swan creek exit [SEP] ### +### [P_TEXT]: [CLS] from laurel gray, exit right out of the parking lot onto old highway 421 and ### +### continue 1. 4 miles to swan creek road turn right and follow swan creek road for 2. 7 miles. turn ### +### left onto cedar forest road. go 1. 1 mile to groce road on your left. take groce road and ### +### raffaldini vineyards is just 0. 2 mile ahead on the right. [SEP] ### +### ======================================= h_v_q | Gates: 26814 ======================================= ### +### ('swan', 0, 1) ('exit', 1, 2) ('creek', 2, 6) ('there', 3, 3440) ('exits', 4, 20) ('is', 5, 11228) ### +### ('relating', 6, 27587) ('river', 7, 80) ('familiarity', 8, 23010) ('.', 9, 10585) ### +### ('entrance', 10, 69) ('downtown', 11, 2078) ('stream', 12, 730) ('lake', 13, 539) ### +### ('interchange', 14, 722) ('conclusion', 15, 838) ('nearby', 16, 868) ('end', 17, 7301) ### +### ('door', 18, 1461) ('escape', 19, 1300) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('laurel', 13036, 0) ('swan', 0, 1) ('exit', 1, 2) ('gray', 2669, 3) ('cedar', 3415, 4) ### +### ('vineyards', 16665, 5) ('creek', 2, 6) ('forest', 3762, 7) ('highway', 1026, 8) ('onto', 797, 9) ### +### ('raf', 2206, 10) ('ahead', 4915, 11) ('winery', 6562, 12) ('mile', 1792, 13) ('parking', 137, 14) ### +### ('miles', 328, 15) ('vines', 2296, 16) ('unwilling', 201, 17) ('follow', 6608, 18) ### +### ('##fa', 13787, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('swan', 0, 1) ('exit', 1, 2) ('creek', 2, 6) ('exits', 4, 20) ('river', 7, 80) ### +### ('entrance', 10, 69) ('there', 3, 3440) ('exited', 23, 135) ('lake', 13, 539) ('jeremy', 31, 118) ### +### ('stream', 12, 730) ('parking', 137, 14) ('swans', 40, 98) ('interchange', 14, 722) ### +### ('unwilling', 201, 17) ('turn', 186, 23) ('conclusion', 15, 838) ('crashing', 124, 38) ### +### ('ছ', 125, 47) ('miles', 328, 15) ### +############################################################################################################ +[2023-10-07 22:30:36,786][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:30:36,786][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:30:37,200][root][INFO] - Epoch: 10: Step: 1/1557, loss[v]=0.058671, lr=0.000011, acc@1[1]=245.0/256=0.95703125, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 22:31:53,423][root][INFO] - Train batch 100 +[2023-10-07 22:31:53,424][root][INFO] - Avg. loss per last 100 batches: 0.070331 +[2023-10-07 22:31:54,124][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29184.5/29522=98.86% | mean: 0.01 | max: 5.46 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.17 | max: 6.16 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] resort in ko olina [SEP] ### +### [P_TEXT]: [CLS] ko olina resort. ko olina resort is a 642 - acre ( 2. 60 km2 ) master - planned ### +### vacation and residential community on the leeward coast of oahu, 17 miles ( 27 km ) northwest of ### +### honolulu. ko olina has 2 miles ( 3. 2 km ) of coastal frontage and includes three natural and four ### +### man - made lagoons with white - sand beaches. [SEP] ### +### ======================================= h_v_q | Gates: 27918 ======================================= ### +### ('ol', 0, 3) ('resort', 1, 2) ('ko', 2, 1) ('##ina', 3, 15) ('resorts', 4, 19) ### +### ('familiarity', 5, 27403) ('relating', 6, 24073) ('plural', 7, 12393) ('.', 8, 9299) ### +### ('hotel', 9, 112) ('lodge', 10, 130) ('ku', 11, 144) ('consisting', 12, 26050) ### +### ('stylized', 13, 27936) ('in', 14, 27524) ('##ine', 15, 1318) ('castle', 16, 3209) ### +### ('leisure', 17, 5702) ('casino', 18, 689) ('∈', 19, 943) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('honolulu', 8500, 0) ('ko', 2, 1) ('resort', 1, 2) ('ol', 0, 3) ('##ward', 26267, 4) ### +### ('beaches', 20268, 5) ('acres', 18908, 6) ('residential', 1987, 7) ('vacation', 21, 8) ### +### ('##ahu', 19503, 9) ('planned', 630, 10) ('hawaii', 3363, 11) ('coast', 4184, 12) ### +### ('lagoon', 2158, 13) ('acre', 21493, 14) ('##ina', 3, 15) ('lee', 689, 16) ('ˈ', 25, 17) ### +### ('encompasses', 1786, 18) ('resorts', 4, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('resort', 1, 2) ('ol', 0, 3) ('ko', 2, 1) ('##ina', 3, 15) ('resorts', 4, 19) ('vacation', 21, 8) ### +### ('hotel', 9, 112) ('ˈ', 25, 17) ('unwilling', 20, 22) ('lodge', 10, 130) ('hesitated', 43, 26) ### +### ('tina', 27, 54) ('ku', 11, 144) ('gideon', 46, 28) ('ছ', 30, 42) ('−', 34, 47) ### +### ('crashing', 85, 27) ('simon', 42, 52) ('annoyance', 41, 60) ('##₂', 76, 35) ### +############################################################################################################ +[2023-10-07 22:31:54,124][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:31:54,124][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:31:54,531][root][INFO] - Epoch: 10: Step: 101/1557, loss[v]=0.128321, lr=0.000010, acc@1[1]=242.5/256=0.947265625, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 22:33:11,150][root][INFO] - Train batch 200 +[2023-10-07 22:33:11,151][root][INFO] - Avg. loss per last 100 batches: 0.073883 +[2023-10-07 22:33:11,851][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29205.0/29522=98.93% | mean: 0.01 | max: 5.01 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.17 | max: 5.94 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] where is newport beach, ca? [SEP] ### +### [P_TEXT]: [CLS] newport beach, ca. sponsored topics. newport beach, incorporated in 1906, is a city ### +### in orange county, california, 10 miles ( 16 km ) south of downtown santa ana. the population was ### +### 85, 186 at the 2010 census. the city's median family income and property values consistently place ### +### high in national rankings. [SEP] ### +### ======================================= h_v_q | Gates: 26847 ======================================= ### +### ('newport', 0, 0) ('beach', 1, 2) ('california', 2, 7) ('located', 3, 6303) ('is', 4, 255) ### +### ('familiarity', 5, 24999) ('united', 6, 17377) ('where', 7, 17) ('downtown', 8, 9) ('county', 9, 8) ### +### ('ca', 10, 22) ('district', 11, 135) ('plural', 12, 9111) ('relating', 13, 23983) ('city', 14, 91) ### +### ('america', 15, 16698) ('washington', 16, 3613) ('was', 17, 243) ('village', 18, 362) ### +### ('.', 19, 2345) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('newport', 0, 0) ('incorporated', 4998, 1) ('beach', 1, 2) ('beaches', 978, 3) ('orange', 2478, 4) ### +### ('sponsored', 10447, 5) ('ana', 2894, 6) ('california', 2, 7) ('county', 9, 8) ('downtown', 8, 9) ### +### ('1906', 9763, 10) ('topics', 5041, 11) ('population', 709, 12) ('median', 22249, 13) ### +### ('encompasses', 28, 14) ('crashing', 142, 15) ('counties', 2766, 16) ('where', 7, 17) ### +### ('ˈ', 221, 18) ('incorporate', 15086, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('newport', 0, 0) ('beach', 1, 2) ('california', 2, 7) ('downtown', 8, 9) ('where', 7, 17) ### +### ('county', 9, 8) ('ca', 10, 22) ('encompasses', 28, 14) ('is', 4, 255) ('city', 14, 91) ### +### ('district', 11, 135) ('somewhere', 23, 92) ('coast', 30, 74) ('crashing', 142, 15) ### +### ('was', 17, 243) ('village', 18, 362) ('ˈ', 221, 18) ('nearby', 66, 102) ('unwilling', 250, 21) ### +### ('hesitated', 139, 45) ### +############################################################################################################ +[2023-10-07 22:33:11,852][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:33:11,852][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:33:12,274][root][INFO] - Epoch: 10: Step: 201/1557, loss[v]=0.060496, lr=0.000010, acc@1[1]=243.0/256=0.94921875, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 22:34:29,507][root][INFO] - Train batch 300 +[2023-10-07 22:34:29,507][root][INFO] - Avg. loss per last 100 batches: 0.067049 +[2023-10-07 22:34:30,211][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29180.6/29522=98.84% | mean: 0.01 | max: 5.27 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.21 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a permanent place of abode in ma [SEP] ### +### [P_TEXT]: [CLS] the ma department of revenue defines a apermanent place of abodea as a dwelling ### +### place continually maintained by a person, whether or not owned by such person, and will include a ### +### dwelling place owned or leased by a person's spouse. [SEP] ### +### ======================================= h_v_q | Gates: 27460 ======================================= ### +### ('ma', 0, 1) ('ab', 1, 6) ('permanent', 2, 1037) ('##ode', 3, 4) ('place', 4, 3) ### +### ('massachusetts', 5, 64) ('places', 6, 11) ('is', 7, 1149) ('familiarity', 8, 26381) ### +### ('permanently', 9, 3124) ('space', 10, 192) ('hampshire', 11, 1549) ('temporary', 12, 2067) ### +### ('plural', 13, 11655) ('relating', 14, 20106) ('stylized', 15, 28000) ('stands', 16, 2505) ### +### ('colorado', 17, 6933) ('refers', 18, 6166) ('home', 19, 1903) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('dwelling', 265, 0) ('ma', 0, 1) ('revenue', 7365, 2) ('place', 4, 3) ('##ode', 3, 4) ### +### ('ape', 19633, 5) ('ab', 1, 6) ('leased', 17677, 7) ('department', 3922, 8) ('define', 12689, 9) ### +### ('definition', 25, 10) ('places', 6, 11) ('##α', 471, 12) ('spouse', 18024, 13) ### +### ('definitions', 7170, 14) ('meaning', 2896, 15) ('owned', 896, 16) ('ˈ', 68, 17) ### +### ('##ent', 6398, 18) ('crashing', 42, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ma', 0, 1) ('##ode', 3, 4) ('ab', 1, 6) ('place', 4, 3) ('places', 6, 11) ### +### ('massachusetts', 5, 64) ('permanent', 2, 1037) ('definition', 25, 10) ('encompasses', 20, 51) ### +### ('crashing', 42, 19) ('defined', 44, 24) ('space', 10, 192) ('##odes', 24, 74) ('dwelling', 265, 0) ### +### ('ˈ', 68, 17) ('angrily', 63, 28) ('unwilling', 86, 23) ('spatial', 28, 119) ('is', 7, 1149) ### +### ('person', 82, 43) ### +############################################################################################################ +[2023-10-07 22:34:30,211][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:34:30,211][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:34:30,635][root][INFO] - Epoch: 10: Step: 301/1557, loss[v]=0.048341, lr=0.000010, acc@1[1]=246.5/256=0.962890625, acc@1[2]=251.5/256=0.982421875 +[2023-10-07 22:35:47,423][root][INFO] - Train batch 400 +[2023-10-07 22:35:47,424][root][INFO] - Avg. loss per last 100 batches: 0.070741 +[2023-10-07 22:35:48,122][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29209.2/29522=98.94% | mean: 0.01 | max: 5.43 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.17 | max: 6.04 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is biliary removal [SEP] ### +### [P_TEXT]: [CLS] a biliary obstruction blocks the bile ducts, which carry bile to the small ### +### intestine for digestion and waste removal. learn about symptoms, causes and more. a biliary ### +### obstruction blocks the bile ducts, which carry bile to the small intestine for digestion and waste ### +### removal. [SEP] ### +### ======================================= h_v_q | Gates: 28001 ======================================= ### +### ('bi', 0, 7) ('removal', 1, 13) ('##lia', 2, 50) ('##ry', 3, 63) ('is', 4, 12021) ### +### ('definition', 5, 682) ('plural', 6, 14798) ('familiarity', 7, 28204) ('refers', 8, 21440) ### +### ('encompasses', 9, 88) ('relating', 10, 26170) ('remove', 11, 31) ('stylized', 12, 29076) ### +### ('consisting', 13, 24864) ('removed', 14, 78) ('noun', 15, 28350) ('.', 16, 10952) ### +### ('defined', 17, 1029) ('simon', 18, 43) ('stands', 19, 7247) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('obstruction', 7109, 0) ('bile', 20056, 1) ('blocks', 13459, 2) ('waste', 13457, 3) ('ˈ', 30, 4) ### +### ('crashing', 27, 5) ('duct', 17722, 6) ('bi', 0, 7) ('hating', 51, 8) ('blocked', 5747, 9) ### +### ('block', 4565, 10) ('digest', 25227, 11) ('unwilling', 23, 12) ('removal', 1, 13) ### +### ('hesitated', 48, 14) ('gideon', 67, 15) ('sharply', 46, 16) ('crashed', 103, 17) ### +### ('carry', 7920, 18) ('##₂', 69, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('bi', 0, 7) ('removal', 1, 13) ('##lia', 2, 50) ('##ry', 3, 63) ('remove', 11, 31) ### +### ('encompasses', 9, 88) ('ˈ', 30, 4) ('definition', 5, 682) ('crashing', 27, 5) ### +### ('unwilling', 23, 12) ('simon', 18, 43) ('removed', 14, 78) ('hating', 51, 8) ('hesitated', 48, 14) ### +### ('hugh', 25, 51) ('wingspan', 40, 22) ('sharply', 46, 16) ('angrily', 39, 23) ('gideon', 67, 15) ### +### ('−', 41, 28) ### +############################################################################################################ +[2023-10-07 22:35:48,122][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:35:48,122][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:35:48,546][root][INFO] - Epoch: 10: Step: 401/1557, loss[v]=0.129854, lr=0.000010, acc@1[1]=240.5/256=0.939453125, acc@1[2]=248.0/256=0.96875 +[2023-10-07 22:37:04,533][root][INFO] - Train batch 500 +[2023-10-07 22:37:04,534][root][INFO] - Avg. loss per last 100 batches: 0.071065 +[2023-10-07 22:37:05,263][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29180.2/29522=98.84% | mean: 0.01 | max: 5.44 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.17 | max: 6.10 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what does a horse cost [SEP] ### +### [P_TEXT]: [CLS] responses to a horse - ownership survey from the university of maine found that the ### +### average annual cost of horse ownership is $ 3, 876 per horse, while the median cost is $ 2, 419. ### +### that puts the average monthly expense anywhere from $ 200 to $ 325 a on par with a car payment. ### +### [SEP] ### +### ======================================= h_v_q | Gates: 26906 ======================================= ### +### ('horse', 0, 1) ('cost', 1, 2) ('$', 2, 5) ('costs', 3, 6) ('##£', 4, 20508) ('.', 5, 8912) ### +### ('horses', 6, 4) ('familiarity', 7, 27400) ('relating', 8, 24547) ('answer', 9, 5000) ### +### ('does', 10, 8984) ('plural', 11, 19149) ('definition', 12, 11764) ('something', 13, 11814) ### +### ('animal', 14, 251) ('refers', 15, 25134) ('price', 16, 11) ('is', 17, 2363) ('doesn', 18, 1204) ### +### ('stylized', 19, 28726) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('maine', 5497, 0) ('horse', 0, 1) ('cost', 1, 2) ('ownership', 3062, 3) ('horses', 6, 4) ### +### ('$', 2, 5) ('costs', 3, 6) ('expense', 72, 7) ('ˈ', 205, 8) ('horseback', 3942, 9) ### +### ('payment', 109, 10) ('price', 16, 11) ('saddle', 43, 12) ('unwilling', 173, 13) ### +### ('hating', 118, 14) ('median', 17597, 15) ('wingspan', 638, 16) ('crashing', 135, 17) ### +### ('stumbled', 130, 18) ('hesitated', 117, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('horse', 0, 1) ('cost', 1, 2) ('$', 2, 5) ('costs', 3, 6) ('horses', 6, 4) ('price', 16, 11) ### +### ('prices', 28, 21) ('expense', 72, 7) ('saddle', 43, 12) ('fee', 32, 31) ('annoyance', 36, 26) ### +### ('car', 26, 59) ('payment', 109, 10) ('mare', 22, 123) ('ˈ', 205, 8) ('animal', 14, 251) ### +### ('hating', 118, 14) ('simon', 45, 56) ('stallion', 69, 35) ('sharply', 74, 34) ### +############################################################################################################ +[2023-10-07 22:37:05,264][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:37:05,264][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:37:05,693][root][INFO] - Epoch: 10: Step: 501/1557, loss[v]=0.085064, lr=0.000010, acc@1[1]=239.5/256=0.935546875, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 22:38:22,508][root][INFO] - Train batch 600 +[2023-10-07 22:38:22,509][root][INFO] - Avg. loss per last 100 batches: 0.072328 +[2023-10-07 22:38:23,198][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29173.4/29522=98.82% | mean: 0.01 | max: 5.39 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.5/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 6.14 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what did a sculpture do in ancient india [SEP] ### +### [P_TEXT]: [CLS] sculpture in india sculpture in india suitable the ideal medium to represent the ### +### image of gods and demigods and at the same time embodied in a tangible way their traditions, ### +### philosophical and religious beliefs with an artistic medium allowing to be appreciated from ### +### different visual angles. [SEP] ### +### ======================================= h_v_q | Gates: 27222 ======================================= ### +### ('sculpture', 0, 0) ('india', 1, 1) ('ancient', 2, 3540) ('sculptures', 3, 4) ('was', 4, 6298) ### +### ('.', 5, 3373) ('sculptor', 6, 14) ('statue', 7, 60) ('doing', 8, 231) ('do', 9, 4986) ### +### ('did', 10, 5329) ('aired', 11, 20037) ('began', 12, 24504) ('brazil', 13, 1450) ('indian', 14, 24) ### +### ('canada', 15, 343) ('art', 16, 70) ('perform', 17, 19404) ('knew', 18, 784) ('china', 19, 227) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('sculpture', 0, 0) ('india', 1, 1) ('ideal', 3261, 2) ('tangible', 9426, 3) ('sculptures', 3, 4) ### +### ('demi', 28730, 5) ('embodied', 11916, 6) ('medium', 13660, 7) ('gods', 1187, 8) ### +### ('represent', 2365, 9) ('crashing', 388, 10) ('ˈ', 656, 11) ('unwilling', 378, 12) ### +### ('suitable', 14649, 13) ('sculptor', 6, 14) ('statues', 88, 15) ('artistic', 1189, 16) ### +### ('crashed', 1323, 17) ('image', 44, 18) ('stumbled', 443, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('sculpture', 0, 0) ('india', 1, 1) ('sculptures', 3, 4) ('sculptor', 6, 14) ('statue', 7, 60) ### +### ('indian', 14, 24) ('art', 16, 70) ('doing', 8, 231) ('carving', 25, 39) ('ancient', 2, 3540) ### +### ('image', 44, 18) ('indians', 36, 55) ('monument', 30, 95) ('statues', 88, 15) ('ト', 26, 148) ### +### ('painting', 21, 226) ('china', 19, 227) ('symbol', 57, 53) ('canada', 15, 343) ('.', 5, 3373) ### +############################################################################################################ +[2023-10-07 22:38:23,198][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:38:23,198][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:38:23,619][root][INFO] - Epoch: 10: Step: 601/1557, loss[v]=0.092756, lr=0.000010, acc@1[1]=235.5/256=0.919921875, acc@1[2]=247.0/256=0.96484375 +[2023-10-07 22:39:39,837][root][INFO] - Train batch 700 +[2023-10-07 22:39:39,838][root][INFO] - Avg. loss per last 100 batches: 0.070195 +[2023-10-07 22:39:40,525][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29161.0/29522=98.78% | mean: 0.01 | max: 5.63 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 6.16 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] tauk meaning [SEP] ### +### [P_TEXT]: [CLS] cognate with tibetan a½a½´a½ ( dug, a poison ; toxin a ), a½a½a½´a½ ( gdug ), ### +### a½a½a½´a½a¼a½ ( gdug pa, a vicious ; evil ; poisonous a ), burmese aa±a¬aaº ( tauk, a to suffer ### +### from toxicity ; to be ill ; to be poisonous a ). a departing tone variant, meaning ato poisona, is ### +### preserved in cantonese, hakka and min dialects. it is derived from the sense apoisona with the * as ### +### suffix in old chinese. pronunciation 1 [SEP] ### +### ======================================= h_v_q | Gates: 27171 ======================================= ### +### ('tau', 0, 0) ('##k', 1, 17) ('meaning', 2, 40) ('noun', 3, 19137) ('definition', 4, 46) ### +### ('means', 5, 69) ('familiarity', 6, 27504) ('relating', 7, 25464) ('refers', 8, 8975) ### +### ('symbol', 9, 593) ('k', 10, 429) ('plural', 11, 16894) ('sense', 12, 137) ('defined', 13, 154) ### +### ('##ka', 14, 2832) ('##ks', 15, 1412) ('something', 16, 3347) ('stylized', 17, 27016) ### +### ('term', 18, 968) ('.', 19, 3629) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('tau', 0, 0) ('burmese', 16962, 1) ('dug', 15500, 2) ('tone', 1521, 3) ('poison', 2686, 4) ### +### ('tibetan', 3130, 5) ('poisonous', 20363, 6) ('toxin', 10129, 7) ('toxicity', 15019, 8) ### +### ('chinese', 1516, 9) ('departing', 18367, 10) ('##gna', 23433, 11) ('burma', 14489, 12) ### +### ('##du', 19499, 13) ('##½', 14615, 14) ('cantonese', 22069, 15) ('min', 17958, 16) ('##k', 1, 17) ### +### ('vicious', 6618, 18) ('myanmar', 11856, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('tau', 0, 0) ('##k', 1, 17) ('meaning', 2, 40) ('definition', 4, 46) ('means', 5, 69) ### +### ('sense', 12, 137) ('defined', 13, 154) ('meanings', 28, 80) ('mean', 35, 56) ('symbol', 9, 593) ### +### ('k', 10, 429) ('##g', 52, 54) ('##α', 75, 59) ('encompasses', 44, 110) ('आ', 104, 57) ### +### ('gideon', 87, 74) ('meant', 55, 114) ('##kka', 376, 23) ('china', 220, 43) ('##ང', 234, 50) ### +############################################################################################################ +[2023-10-07 22:39:40,526][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:39:40,526][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:39:40,929][root][INFO] - Epoch: 10: Step: 701/1557, loss[v]=0.063968, lr=0.000010, acc@1[1]=240.5/256=0.939453125, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 22:40:57,586][root][INFO] - Train batch 800 +[2023-10-07 22:40:57,587][root][INFO] - Avg. loss per last 100 batches: 0.069969 +[2023-10-07 22:40:58,288][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29097.2/29522=98.56% | mean: 0.01 | max: 5.36 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 6.15 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how much does the average pool table weigh [SEP] ### +### [P_TEXT]: [CLS] a pool table weight will vary between manufacture, style and thickness of slate. a ### +### good rule of thumb is to expect the table to weigh between 500 - 1000 pounds. this is about ### +### equivalent to 4 - 5 men standing in the same room. better floor support will enhance the ### +### playability of the game. floors that have minimal support will flex slightly as players move around ### +### the room. [SEP] ### +### ======================================= h_v_q | Gates: 26773 ======================================= ### +### ('pool', 0, 0) ('table', 1, 2) ('weighs', 2, 122) ('average', 3, 524) ('weigh', 4, 3) ### +### ('weight', 5, 4) ('$', 6, 4760) ('.', 7, 8368) ('familiarity', 8, 23868) ('##£', 9, 28234) ### +### ('answer', 10, 17259) ('pounds', 11, 21) ('much', 12, 58) ('approximately', 13, 3046) ### +### ('weighed', 14, 73) ('lake', 15, 2714) ('relating', 16, 25795) ('tables', 17, 18) ### +### ('considerable', 18, 44) ('30', 19, 70) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('pool', 0, 0) ('slate', 10031, 1) ('table', 1, 2) ('weigh', 4, 3) ('weight', 5, 4) ### +### ('thumb', 6718, 5) ('thickness', 1103, 6) ('##ability', 6729, 7) ('expecting', 1098, 8) ### +### ('men', 1068, 9) ('ˈ', 252, 10) ('minimal', 228, 11) ('expect', 6288, 12) ('pools', 75, 13) ### +### ('floor', 608, 14) ('vary', 5623, 15) ('enhanced', 4598, 16) ('manufacture', 8125, 17) ### +### ('tables', 17, 18) ('unwilling', 156, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('pool', 0, 0) ('table', 1, 2) ('weigh', 4, 3) ('weight', 5, 4) ('weighs', 2, 122) ### +### ('average', 3, 524) ('pounds', 11, 21) ('much', 12, 58) ('tables', 17, 18) ('weighed', 14, 73) ### +### ('considerable', 18, 44) ('30', 19, 70) ('pound', 21, 76) ('$', 6, 4760) ('pools', 75, 13) ### +### ('300', 51, 36) ('size', 28, 128) ('20', 32, 99) ('weighing', 26, 141) ('hesitated', 83, 34) ### +############################################################################################################ +[2023-10-07 22:40:58,289][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:40:58,289][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:40:58,709][root][INFO] - Epoch: 10: Step: 801/1557, loss[v]=0.070575, lr=0.000010, acc@1[1]=243.0/256=0.94921875, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 22:42:14,620][root][INFO] - Train batch 900 +[2023-10-07 22:42:14,621][root][INFO] - Avg. loss per last 100 batches: 0.068469 +[2023-10-07 22:42:15,326][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29183.6/29522=98.85% | mean: 0.01 | max: 5.49 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.15 | max: 6.35 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is topiramate [SEP] ### +### [P_TEXT]: [CLS] topiramate was originally produced by ortho - mcneil neurologics and noramco, inc., ### +### both divisions of the johnson & johnson corporation. this medication was discovered in 1979 by ### +### bruce e. maryanoff and joseph f. gardocki during their research work at mcneil pharmaceutical. ### +### opiramate is used to treat epilepsy in children and adults, and it was originally used as an ### +### anticonvulsant. in children, it is indicated for the treatment of lennox - gastaut syndrome, a ### +### disorder that causes seizures and developmental delay. [SEP] ### +### ======================================= h_v_q | Gates: 27998 ======================================= ### +### ('##ira', 0, 0) ('##mate', 1, 3) ('top', 2, 7) ('encompasses', 3, 15) ('familiarity', 4, 27422) ### +### ('is', 5, 2281) ('definition', 6, 1057) ('refers', 7, 19039) ('plural', 8, 17083) ### +### ('noun', 9, 28086) ('consisting', 10, 23248) ('relating', 11, 26988) ('mate', 12, 220) ### +### ('stylized', 13, 27011) ('.', 14, 16610) ('##mates', 15, 38) ('term', 16, 15833) ### +### ('bottom', 17, 515) ('##yra', 18, 5993) ('something', 19, 4602) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##ira', 0, 0) ('op', 16470, 1) ('nora', 161, 2) ('##mate', 1, 3) ('lennox', 17109, 4) ### +### ('medication', 14775, 5) ('discovered', 1719, 6) ('top', 2, 7) ('hesitated', 26, 8) ('ˈ', 47, 9) ### +### ('johnson', 2598, 10) ('##logic', 16208, 11) ('crashing', 162, 12) ('unwilling', 53, 13) ### +### ('stumbled', 34, 14) ('encompasses', 3, 15) ('hating', 56, 16) ('seizures', 25854, 17) ### +### ('medications', 8945, 18) ('pharmaceutical', 11617, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##ira', 0, 0) ('##mate', 1, 3) ('top', 2, 7) ('encompasses', 3, 15) ('##mates', 15, 38) ### +### ('hesitated', 26, 8) ('mate', 12, 220) ('definition', 6, 1057) ('stumbled', 34, 14) ('−', 32, 21) ### +### ('ˈ', 47, 9) ('unwilling', 53, 13) ('##₂', 36, 29) ('is', 5, 2281) ('##α', 30, 56) ### +### ('hating', 56, 16) ('sharply', 61, 20) ('##ο', 50, 31) ('nora', 161, 2) ('gideon', 59, 25) ### +############################################################################################################ +[2023-10-07 22:42:15,326][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:42:15,326][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:42:15,751][root][INFO] - Epoch: 10: Step: 901/1557, loss[v]=0.055946, lr=0.000010, acc@1[1]=244.5/256=0.955078125, acc@1[2]=252.0/256=0.984375 +[2023-10-07 22:43:32,917][root][INFO] - Train batch 1000 +[2023-10-07 22:43:32,918][root][INFO] - Avg. loss per last 100 batches: 0.071261 +[2023-10-07 22:43:33,631][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29158.4/29522=98.77% | mean: 0.01 | max: 5.32 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.15 | max: 6.07 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] who is carson daly married to [SEP] ### +### [P_TEXT]: [CLS] view all photos ( 7 ). carson daly is currently engaged to siri pinter. he has been ### +### in seven celebrity relationships averaging approximately 2. 0 years each. he has never been ### +### married. given name : carson jones daly. [SEP] ### +### ======================================= h_v_q | Gates: 27463 ======================================= ### +### ('daly', 0, 0) ('carson', 1, 1) ('married', 2, 5) ('wife', 3, 130) ('marriage', 4, 60) ### +### ('familiarity', 5, 24937) ('.', 6, 12527) ('husband', 7, 93) ('declan', 8, 77) ('whose', 9, 212) ### +### ('marry', 10, 14) ('america', 11, 21033) ('wedding', 12, 227) ('is', 13, 2505) ### +### ('relating', 14, 26094) ('plural', 15, 6907) ('ryan', 16, 284) ('consisting', 17, 25045) ### +### ('stylized', 18, 27691) ('songwriter', 19, 9481) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('daly', 0, 0) ('carson', 1, 1) ('engaged', 87, 2) ('celebrity', 3452, 3) ('pin', 7044, 4) ### +### ('married', 2, 5) ('sir', 4009, 6) ('jones', 5924, 7) ('##ter', 411, 8) ('ˈ', 566, 9) ### +### ('relationships', 4918, 10) ('##i', 12215, 11) ('currently', 4005, 12) ('never', 4595, 13) ### +### ('marry', 10, 14) ('cyrillic', 868, 15) ('##ང', 545, 16) ('wingspan', 2236, 17) ('##ο', 892, 18) ### +### ('unwilling', 543, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('daly', 0, 0) ('carson', 1, 1) ('married', 2, 5) ('marriage', 4, 60) ('wife', 3, 130) ### +### ('marry', 10, 14) ('husband', 7, 93) ('declan', 8, 77) ('engaged', 87, 2) ('who', 21, 45) ### +### ('julian', 20, 79) ('whose', 9, 212) ('wedding', 12, 227) ('relationship', 88, 30) ### +### ('dodgers', 53, 69) ('seth', 89, 40) ('divorce', 29, 122) ('hesitated', 155, 24) ### +### ('stumbled', 163, 23) ('ryan', 16, 284) ### +############################################################################################################ +[2023-10-07 22:43:33,631][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:43:33,631][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:43:34,034][root][INFO] - Epoch: 10: Step: 1001/1557, loss[v]=0.080828, lr=0.000010, acc@1[1]=240.0/256=0.9375, acc@1[2]=253.0/256=0.98828125 +[2023-10-07 22:44:50,613][root][INFO] - Train batch 1100 +[2023-10-07 22:44:50,614][root][INFO] - Avg. loss per last 100 batches: 0.068368 +[2023-10-07 22:44:51,300][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29182.3/29522=98.85% | mean: 0.01 | max: 5.69 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.15 | max: 6.20 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] effect of great depression on india [SEP] ### +### [P_TEXT]: [CLS] great depression in india. the great depression of 1929 had a very severe impact on ### +### india, which was then under the rule of the british raj. the government of british india adopted a ### +### protective trade policy which, though beneficial to the united kingdom, caused great damage to the ### +### indian economy. [SEP] ### +### ======================================= h_v_q | Gates: 27277 ======================================= ### +### ('depression', 0, 1) ('india', 1, 0) ('great', 2, 4) ('effect', 3, 196) ('familiarity', 4, 27350) ### +### ('.', 5, 3551) ('effects', 6, 60) ('war', 7, 12014) ('impact', 8, 6) ('relating', 9, 23509) ### +### ('affect', 10, 25) ('1930', 11, 89) ('consisting', 12, 24222) ('china', 13, 571) ('julian', 14, 51) ### +### ('plural', 15, 9101) ('simon', 16, 54) ('considerable', 17, 34) ('israel', 18, 1747) ### +### ('stylized', 19, 27879) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('india', 1, 0) ('depression', 0, 1) ('indian', 22, 2) ('raj', 733, 3) ('great', 2, 4) ### +### ('ˈ', 129, 5) ('impact', 8, 6) ('wingspan', 320, 7) ('stumbled', 74, 8) ('unwilling', 168, 9) ### +### ('crashing', 323, 10) ('gideon', 39, 11) ('angrily', 133, 12) ('sharply', 62, 13) ### +### ('indians', 305, 14) ('hating', 325, 15) ('1929', 188, 16) ('##₂', 69, 17) ('##ང', 345, 18) ### +### ('british', 121, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('depression', 0, 1) ('india', 1, 0) ('great', 2, 4) ('effect', 3, 196) ('impact', 8, 6) ### +### ('effects', 6, 60) ('affect', 10, 25) ('indian', 22, 2) ('considerable', 17, 34) ('1930', 11, 89) ### +### ('julian', 14, 51) ('simon', 16, 54) ('−', 31, 20) ('gideon', 39, 11) ('sharply', 62, 13) ### +### ('vast', 24, 62) ('stumbled', 74, 8) ('##₂', 69, 17) ('ˈ', 129, 5) ('dante', 37, 42) ### +############################################################################################################ +[2023-10-07 22:44:51,300][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:44:51,300][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:44:51,705][root][INFO] - Epoch: 10: Step: 1101/1557, loss[v]=0.044497, lr=0.000010, acc@1[1]=242.5/256=0.947265625, acc@1[2]=253.0/256=0.98828125 +[2023-10-07 22:46:07,937][root][INFO] - Train batch 1200 +[2023-10-07 22:46:07,938][root][INFO] - Avg. loss per last 100 batches: 0.070760 +[2023-10-07 22:46:08,668][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29220.2/29522=98.98% | mean: 0.01 | max: 5.54 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.21 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] nadir sepsis definition [SEP] ### +### [P_TEXT]: [CLS] thread : nadir sepsis. can anyone explain what nadir sepsis is.. also, any resource ### +### or paper on the topic.. thanks a lot!!! nadir sepsis nadir sepsis it is usually an unidentifiable ### +### cause of sepsis. usually in neutropenic patients. and it correlates with the neutropenic nadir ( 7 ### +### - 10 days post chemotherapy ). you need to learn how to use medical resource such as pubmed and ### +### uptodatecom. [SEP] ### +### ======================================= h_v_q | Gates: 27809 ======================================= ### +### ('nad', 0, 0) ('##ir', 1, 3) ('sep', 2, 2) ('##sis', 3, 5) ('definition', 4, 1459) ### +### ('familiarity', 5, 25990) ('noun', 6, 28925) ('defined', 7, 9782) ('.', 8, 15042) ### +### ('relating', 9, 22077) ('something', 10, 2360) ('consisting', 11, 26149) ('refers', 12, 23069) ### +### ('latin', 13, 15292) ('stylized', 14, 28539) ('plural', 15, 23233) ('specified', 16, 14490) ### +### ('definitions', 17, 137) ('##irs', 18, 70) ('−', 19, 18) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('nad', 0, 0) ('thread', 11401, 1) ('sep', 2, 2) ('##ir', 1, 3) ('ˈ', 77, 4) ('##sis', 3, 5) ### +### ('chemotherapy', 12486, 6) ('unwilling', 132, 7) ('crashing', 314, 8) ('##ང', 410, 9) ### +### ('##lates', 16475, 10) ('hating', 154, 11) ('gideon', 90, 12) ('##₂', 40, 13) ('cyrillic', 374, 14) ### +### ('post', 9854, 15) ('wingspan', 356, 16) ('angrily', 42, 17) ('−', 19, 18) ('crashed', 383, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('nad', 0, 0) ('sep', 2, 2) ('##ir', 1, 3) ('##sis', 3, 5) ('definition', 4, 1459) ('−', 19, 18) ### +### ('##α', 26, 30) ('ˈ', 77, 4) ('##₂', 40, 13) ('encompasses', 28, 37) ('##irs', 18, 70) ### +### ('angrily', 42, 17) ('simon', 22, 67) ('definitions', 17, 137) ('hugh', 35, 55) ('gideon', 90, 12) ### +### ('unwilling', 132, 7) ('sharply', 100, 20) ('hesitated', 95, 24) ('##ο', 103, 23) ### +############################################################################################################ +[2023-10-07 22:46:08,669][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:46:08,669][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:46:09,098][root][INFO] - Epoch: 10: Step: 1201/1557, loss[v]=0.048612, lr=0.000010, acc@1[1]=246.0/256=0.9609375, acc@1[2]=254.0/256=0.9921875 +[2023-10-07 22:47:25,863][root][INFO] - Train batch 1300 +[2023-10-07 22:47:25,864][root][INFO] - Avg. loss per last 100 batches: 0.067320 +[2023-10-07 22:47:26,544][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29247.0/29522=99.07% | mean: 0.01 | max: 5.25 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.17 | max: 6.06 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is dt environment for software developers [SEP] ### +### [P_TEXT]: [CLS] best integrated development environment ( ide ) software. integrated development ### +### environments, or ides, are software platforms that provide programmers and developers a ### +### comprehensive set of tools for software development in a single product. ides are built to work ### +### with specific application platforms and remove barriers involved in the lifecycle of software ### +### development. [SEP] ### +### ======================================= h_v_q | Gates: 27675 ======================================= ### +### ('dt', 0, 4390) ('software', 1, 4) ('environment', 2, 8) ('developers', 3, 23) ### +### ('familiarity', 4, 27285) ('developer', 5, 31) ('.', 6, 10463) ('digital', 7, 3278) ('is', 8, 1020) ### +### ('development', 9, 2) ('encompasses', 10, 11) ('relating', 11, 24650) ('community', 12, 1433) ### +### ('consisting', 13, 23709) ('atmosphere', 14, 84) ('plural', 15, 15873) ('infrastructure', 16, 139) ### +### ('computer', 17, 378) ('for', 18, 20336) ('st', 19, 11627) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('integrated', 4007, 0) ('id', 370, 1) ('development', 9, 2) ('##es', 13293, 3) ('software', 1, 4) ### +### ('environments', 30, 5) ('##e', 11694, 6) ('platforms', 5957, 7) ('environment', 2, 8) ### +### ('integrating', 22865, 9) ('ˈ', 221, 10) ('encompasses', 10, 11) ('##₂', 172, 12) ### +### ('crashing', 135, 13) ('unwilling', 356, 14) ('tools', 725, 15) ('stumbled', 232, 16) ### +### ('programmers', 304, 17) ('best', 4685, 18) ('gideon', 619, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('software', 1, 4) ('environment', 2, 8) ('developers', 3, 23) ('development', 9, 2) ### +### ('developer', 5, 31) ('encompasses', 10, 11) ('dt', 0, 4390) ('environments', 30, 5) ### +### ('atmosphere', 14, 84) ('id', 370, 1) ('−', 49, 21) ('infrastructure', 16, 139) ### +### ('definition', 28, 51) ('angrily', 57, 32) ('crashing', 135, 13) ('##₂', 172, 12) ('is', 8, 1020) ### +### ('sharply', 89, 28) ('ˈ', 221, 10) ('julian', 27, 123) ### +############################################################################################################ +[2023-10-07 22:47:26,545][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:47:26,545][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:47:26,962][root][INFO] - Epoch: 10: Step: 1301/1557, loss[v]=0.067301, lr=0.000010, acc@1[1]=244.5/256=0.955078125, acc@1[2]=252.0/256=0.984375 +[2023-10-07 22:48:43,044][root][INFO] - Train batch 1400 +[2023-10-07 22:48:43,045][root][INFO] - Avg. loss per last 100 batches: 0.070198 +[2023-10-07 22:48:43,729][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29190.6/29522=98.88% | mean: 0.01 | max: 5.54 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 6.23 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what do placebo pills do [SEP] ### +### [P_TEXT]: [CLS] placebo pills are just pills with no active ingredient in them. they are just ### +### psychological medication which does often work. they are given to patients who think they are ill ### +### but are not, it is just in their mind, so they take the placebo pill thinking it will make them ### +### better and very often it works. [SEP] ### +### ======================================= h_v_q | Gates: 27798 ======================================= ### +### ('##bo', 0, 11) ('place', 1, 5) ('pills', 2, 0) ('.', 3, 18114) ('familiarity', 4, 26973) ### +### ('operate', 5, 223) ('ト', 6, 201) ('do', 7, 6612) ('places', 8, 56) ('relating', 9, 25635) ### +### ('doing', 10, 189) ('perform', 11, 14160) ('what', 12, 316) ('something', 13, 4610) ### +### ('##bos', 14, 75) ('include', 15, 286) ('ability', 16, 2780) ('destroy', 17, 2038) ('are', 18, 582) ### +### ('julian', 19, 92) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('pills', 2, 0) ('pill', 32, 1) ('medication', 178, 2) ('ˈ', 334, 3) ('unwilling', 98, 4) ### +### ('place', 1, 5) ('hating', 204, 6) ('ill', 12248, 7) ('##₂', 109, 8) ('crashing', 159, 9) ### +### ('psychological', 10390, 10) ('##bo', 0, 11) ('stumbled', 45, 12) ('realizing', 4464, 13) ### +### ('wingspan', 970, 14) ('−', 69, 15) ('sharply', 131, 16) ('ingredient', 15414, 17) ('##ο', 184, 18) ### +### ('ingredients', 4323, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##bo', 0, 11) ('place', 1, 5) ('pills', 2, 0) ('places', 8, 56) ('pill', 32, 1) ('ト', 6, 201) ### +### ('operate', 5, 223) ('doing', 10, 189) ('##bos', 14, 75) ('julian', 19, 92) ('simon', 22, 71) ### +### ('stumbled', 45, 12) ('improve', 21, 91) ('what', 12, 316) ('unwilling', 98, 4) ### +### ('medications', 46, 23) ('medicine', 33, 44) ('##α', 51, 21) ('−', 69, 15) ('medication', 178, 2) ### +############################################################################################################ +[2023-10-07 22:48:43,730][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:48:43,730][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:48:44,132][root][INFO] - Epoch: 10: Step: 1401/1557, loss[v]=0.061491, lr=0.000010, acc@1[1]=243.0/256=0.94921875, acc@1[2]=251.5/256=0.982421875 +[2023-10-07 22:50:00,187][root][INFO] - Train batch 1500 +[2023-10-07 22:50:00,188][root][INFO] - Avg. loss per last 100 batches: 0.066952 +[2023-10-07 22:50:00,895][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29212.0/29522=98.95% | mean: 0.01 | max: 5.35 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.15 | max: 5.99 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] does garcinia cambogia curb appetite [SEP] ### +### [P_TEXT]: [CLS] garcinia cambogia is an impressively effective appetite suppressant, but as i ### +### mentioned in a previous post, you can eat through it if you aren't conscious of your eating. if ### +### you're used to eating large meals and subconsciously overfill your plate, garcinia cambogia might ### +### not keep you from mindlessly clearing it. [SEP] ### +### ======================================= h_v_q | Gates: 28343 ======================================= ### +### ('appetite', 0, 0) ('##rc', 1, 77) ('curb', 2, 13195) ('cam', 3, 4) ('ga', 4, 8) ('##gia', 5, 3) ### +### ('.', 6, 12515) ('does', 7, 13754) ('##bo', 8, 115) ('familiarity', 9, 26524) ('doesn', 10, 734) ### +### ('doing', 11, 187) ('eager', 12, 252) ('consisting', 13, 24054) ('simon', 14, 67) ### +### ('did', 15, 12677) ('plural', 16, 22457) ('##inia', 17, 6) ('answer', 18, 14213) ### +### ('relating', 19, 25182) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('appetite', 0, 0) ('##lessly', 3406, 1) ('impressive', 4116, 2) ('##gia', 5, 3) ('cam', 3, 4) ### +### ('suppress', 2838, 5) ('##inia', 17, 6) ('conscious', 1407, 7) ('ga', 4, 8) ('effective', 1097, 9) ### +### ('subconscious', 13270, 10) ('suppressed', 165, 11) ('eat', 138, 12) ('ˈ', 99, 13) ### +### ('eating', 500, 14) ('hating', 50, 15) ('unwilling', 23, 16) ('##₂', 40, 17) ('gideon', 36, 18) ### +### ('encompasses', 728, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('appetite', 0, 0) ('cam', 3, 4) ('##gia', 5, 3) ('ga', 4, 8) ('##rc', 1, 77) ('##bo', 8, 115) ### +### ('##inia', 17, 6) ('simon', 14, 67) ('doing', 11, 187) ('unwilling', 23, 16) ('eager', 12, 252) ### +### ('stumbled', 26, 28) ('gideon', 36, 18) ('−', 38, 24) ('##α', 28, 35) ('##₂', 40, 17) ### +### ('doesn', 10, 734) ('hating', 50, 15) ('crashing', 46, 23) ('curb', 2, 13195) ### +############################################################################################################ +[2023-10-07 22:50:00,895][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:50:00,895][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:50:01,315][root][INFO] - Epoch: 10: Step: 1501/1557, loss[v]=0.075294, lr=0.000010, acc@1[1]=241.0/256=0.94140625, acc@1[2]=248.0/256=0.96875 +[2023-10-07 22:50:44,273][root][INFO] - rank=2; last iteration 1557 +[2023-10-07 22:50:44,274][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 22:50:44,274][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-07 22:50:44,279][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 22:50:44,279][root][INFO] - Epoch finished on 2 +[2023-10-07 22:50:44,298][root][INFO] - rank=1; last iteration 1557 +[2023-10-07 22:50:44,298][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 22:50:44,298][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-07 22:50:44,299][root][INFO] - rank=3; last iteration 1557 +[2023-10-07 22:50:44,300][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 22:50:44,300][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-07 22:50:44,301][root][INFO] - rank=0; last iteration 1557 +[2023-10-07 22:50:44,302][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 22:50:44,302][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-07 22:50:44,305][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 22:50:44,305][root][INFO] - Epoch finished on 1 +[2023-10-07 22:50:44,309][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 22:50:44,309][root][INFO] - Epoch finished on 3 +[2023-10-07 22:50:44,309][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 22:50:44,310][root][INFO] - Epoch finished on 0 +[2023-10-07 22:50:58,877][root][INFO] - Saved checkpoint at ./vdr_10 +[2023-10-07 22:50:58,878][root][INFO] - Saved checkpoint at ./vdr_10 +[2023-10-07 22:50:58,877][root][INFO] - Saved checkpoint at ./vdr_10 +[2023-10-07 22:50:58,879][root][INFO] - Av Loss per epoch=0.069986 +[2023-10-07 22:50:58,879][root][INFO] - Av Loss per epoch=0.069986 +[2023-10-07 22:50:58,879][root][INFO] - epoch total (1) correct predictions=378249 +[2023-10-07 22:50:58,879][root][INFO] - Av Loss per epoch=0.069986 +[2023-10-07 22:50:58,879][root][INFO] - epoch total (1) correct predictions=378249 +[2023-10-07 22:50:58,879][root][INFO] - epoch total (2) correct predictions=390556 +[2023-10-07 22:50:58,879][root][INFO] - epoch total (1) correct predictions=378249 +[2023-10-07 22:50:58,879][root][INFO] - epoch total (2) correct predictions=390556 +[2023-10-07 22:50:58,879][root][INFO] - epoch total (2) correct predictions=390556 +[2023-10-07 22:50:58,880][root][INFO] - Saved checkpoint at ./vdr_10 +[2023-10-07 22:50:58,881][root][INFO] - Av Loss per epoch=0.069986 +[2023-10-07 22:50:58,881][root][INFO] - epoch total (1) correct predictions=378249 +[2023-10-07 22:50:58,882][root][INFO] - epoch total (2) correct predictions=390556 +[2023-10-07 22:50:58,882][root][INFO] - ***** Epoch 11 ***** +[2023-10-07 22:50:58,883][root][INFO] - ***** Epoch 11 ***** +[2023-10-07 22:50:58,883][root][INFO] - ***** Epoch 11 ***** +[2023-10-07 22:50:58,885][root][INFO] - ***** Epoch 11 ***** +[2023-10-07 22:50:58,888][root][INFO] - rank=2; Iteration start +[2023-10-07 22:50:58,889][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 22:50:58,889][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 22:50:58,889][root][INFO] - rank=0; Iteration start +[2023-10-07 22:50:58,890][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 22:50:58,890][root][INFO] - rank=1; Iteration start +[2023-10-07 22:50:58,890][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 22:50:58,890][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 22:50:58,890][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 22:50:58,890][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-07 22:50:58,892][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-07 22:50:58,892][root][INFO] - rank=3; Iteration start +[2023-10-07 22:50:58,892][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 22:50:58,892][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-07 22:50:58,892][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 22:50:58,894][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-07 22:50:59,832][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29224.0/29522=98.99% | mean: 0.01 | max: 5.25 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 6.11 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] amalgamation meaning [SEP] ### +### [P_TEXT]: [CLS] definition of amalgamation in general, the definition of amalgamation can be stated ### +### as follows. amalgamation is a union of two or more companies, made with an intention to form a new ### +### company. [SEP] ### +### ======================================= h_v_q | Gates: 27625 ======================================= ### +### ('amalgamation', 0, 0) ('meaning', 1, 12) ('amalgamated', 2, 4) ('merger', 3, 18) ### +### ('relating', 4, 12198) ('definition', 5, 2) (';', 6, 863) ('noun', 7, 18766) ('union', 8, 5) ### +### ('.', 9, 4144) ('means', 10, 38) ('or', 11, 12795) ('consolidation', 12, 50) ('combining', 13, 40) ### +### ('association', 14, 2223) ('incorporation', 15, 34) ('mixture', 16, 70) ('sense', 17, 6358) ### +### ('familiarity', 18, 24975) ('collaboration', 19, 1484) ### +### ======================================= h_v_p | Gates: 29520 ======================================= ### +### ('amalgamation', 0, 0) ('definitions', 879, 1) ('definition', 5, 2) ('define', 8565, 3) ### +### ('amalgamated', 2, 4) ('union', 8, 5) ('companies', 2063, 6) ('crashing', 2579, 7) ('ˈ', 3122, 8) ### +### ('unions', 88, 9) ('##₂', 873, 10) ('unwilling', 1385, 11) ('meaning', 1, 12) ('stumbled', 737, 13) ### +### ('hating', 2159, 14) ('encompasses', 340, 15) ('##α', 1187, 16) ('wingspan', 3800, 17) ### +### ('merger', 3, 18) ('crashed', 3656, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('amalgamation', 0, 0) ('meaning', 1, 12) ('amalgamated', 2, 4) ('definition', 5, 2) ### +### ('merger', 3, 18) ('union', 8, 5) ('means', 10, 38) ('consolidation', 12, 50) ### +### ('incorporation', 15, 34) ('combining', 13, 40) ('merge', 20, 27) ('mixture', 16, 70) ### +### ('fusion', 23, 45) ('unions', 88, 9) ('entity', 38, 73) (';', 6, 863) ('linking', 26, 105) ### +### ('integrated', 45, 64) ('defined', 76, 28) ('integration', 33, 124) ### +############################################################################################################ +[2023-10-07 22:50:59,833][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:50:59,833][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:51:00,248][root][INFO] - Epoch: 11: Step: 1/1557, loss[v]=0.067510, lr=0.000009, acc@1[1]=241.5/256=0.943359375, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 22:52:17,279][root][INFO] - Train batch 100 +[2023-10-07 22:52:17,280][root][INFO] - Avg. loss per last 100 batches: 0.067014 +[2023-10-07 22:52:18,001][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29187.2/29522=98.87% | mean: 0.01 | max: 5.21 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.05 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] who is john brown civil rights [SEP] ### +### [P_TEXT]: [CLS] reynolds, david s. ( 2005 ) : john brown, abolitionist : the man who killed ### +### slavery, sparked the civil war, and seeded civil rights ( 2005 ) rodriguez, junius p., ed. ### +### encyclopedia of slave resistance and rebellion. [SEP] ### +### ======================================= h_v_q | Gates: 27124 ======================================= ### +### ('brown', 0, 7) ('civil', 1, 17) ('john', 2, 8) ('rights', 3, 18) ('familiarity', 4, 26589) ### +### ('relating', 5, 27028) ('.', 6, 14421) ('who', 7, 48) ('whose', 8, 87) ('plural', 9, 19555) ### +### ('green', 10, 907) ('association', 11, 8343) ('consisting', 12, 23701) ('league', 13, 10668) ### +### ('is', 14, 18405) ('stylized', 15, 28460) ('american', 16, 15208) ('civilian', 17, 252) ### +### ('william', 18, 330) ('julian', 19, 51) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('abolitionist', 12660, 0) ('reynolds', 379, 1) ('slavery', 3149, 2) ('resistance', 1303, 3) ### +### ('slave', 12815, 4) ('rebellion', 2142, 5) ('rodriguez', 41, 6) ('brown', 0, 7) ('john', 2, 8) ### +### ('ˈ', 182, 9) ('man', 1487, 10) ('sparked', 10177, 11) ('killed', 2658, 12) ('stumbled', 87, 13) ### +### ('unwilling', 129, 14) ('slaves', 2140, 15) ('hesitated', 95, 16) ('civil', 1, 17) ### +### ('rights', 3, 18) ('##ο', 105, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('brown', 0, 7) ('john', 2, 8) ('civil', 1, 17) ('rights', 3, 18) ('who', 7, 48) ('whose', 8, 87) ### +### ('rodriguez', 41, 6) ('julian', 19, 51) ('reynolds', 379, 1) ('stumbled', 87, 13) ('##α', 38, 34) ### +### ('santiago', 33, 50) ('−', 62, 26) ('angrily', 73, 23) ('hesitated', 95, 16) ('shoved', 46, 45) ### +### ('simon', 26, 84) ('david', 75, 27) ('ˈ', 182, 9) ('unwilling', 129, 14) ### +############################################################################################################ +[2023-10-07 22:52:18,001][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:52:18,001][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:52:18,425][root][INFO] - Epoch: 11: Step: 101/1557, loss[v]=0.066392, lr=0.000009, acc@1[1]=240.0/256=0.9375, acc@1[2]=249.5/256=0.974609375 +[2023-10-07 22:53:34,865][root][INFO] - Train batch 200 +[2023-10-07 22:53:34,865][root][INFO] - Avg. loss per last 100 batches: 0.067377 +[2023-10-07 22:53:35,592][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29155.3/29522=98.76% | mean: 0.01 | max: 5.45 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.25 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] are essential oils in a diffuser safe for cats [SEP] ### +### [P_TEXT]: [CLS] hydrosols ( by - products of essential oil distillation ) are safer to use on cats. ### +### this is because the monoterpene alcohols have an affinity for water and are safe for cats. phenols ### +### and ketones do not appear in hydrosols. there are no known case histories of hydrosols or ### +### monoterpene alcohols causing toxicity in cats. ydrosols ( by - products of essential oil ### +### distillation ) are safer to use on cats. this is because the monoterpene alcohols have an affinity ### +### for water and are safe for cats. phenols and ketones do not appear in hydrosols. [SEP] ### +### ======================================= h_v_q | Gates: 27595 ======================================= ### +### ('cats', 0, 0) ('essential', 1, 9) ('diffuse', 2, 11738) ('oils', 3, 41) ('cat', 4, 7) ### +### ('safe', 5, 13) ('.', 6, 13836) ('##r', 7, 24860) ('oil', 8, 12) ('relating', 9, 23712) ### +### ('familiarity', 10, 26546) ('dangerous', 11, 57) ('dog', 12, 163) ('for', 13, 6836) ### +### ('simon', 14, 68) ('important', 15, 1243) ('julian', 16, 151) ('effective', 17, 312) ### +### ('consisting', 18, 22680) ('basic', 19, 446) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('cats', 0, 0) ('safer', 147, 1) ('toxicity', 17440, 2) ('alcohol', 1868, 3) ### +### ('##llation', 21245, 4) ('hydro', 6597, 5) ('##sol', 14235, 6) ('cat', 4, 7) ('ˈ', 219, 8) ### +### ('essential', 1, 9) ('affinity', 5667, 10) ('mono', 3956, 11) ('oil', 8, 12) ('safe', 5, 13) ### +### ('crashing', 206, 14) ('unwilling', 55, 15) ('−', 37, 16) ('water', 1633, 17) ('gideon', 45, 18) ### +### ('hating', 66, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('cats', 0, 0) ('essential', 1, 9) ('cat', 4, 7) ('oils', 3, 41) ('safe', 5, 13) ('oil', 8, 12) ### +### ('dangerous', 11, 57) ('simon', 14, 68) ('∈', 20, 26) ('diffuse', 2, 11738) ('dog', 12, 163) ### +### ('−', 37, 16) ('safer', 147, 1) ('unwilling', 55, 15) ('julian', 16, 151) ('gideon', 45, 18) ### +### ('annoyance', 30, 42) ('dante', 26, 64) ('hesitated', 42, 34) ('angrily', 56, 24) ### +############################################################################################################ +[2023-10-07 22:53:35,592][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:53:35,593][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:53:36,021][root][INFO] - Epoch: 11: Step: 201/1557, loss[v]=0.067554, lr=0.000009, acc@1[1]=245.5/256=0.958984375, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 22:54:52,856][root][INFO] - Train batch 300 +[2023-10-07 22:54:52,856][root][INFO] - Avg. loss per last 100 batches: 0.068299 +[2023-10-07 22:54:53,578][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29158.5/29522=98.77% | mean: 0.01 | max: 5.43 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.05 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how is cigarette made [SEP] ### +### [P_TEXT]: [CLS] cigarettes are made from four components, each of which is describe below. 1. ### +### filters 2. tobacco 3. additives 4. cigarette wrapper cigarettes today are typically 85 or 100 mm ### +### long, and have diameters of about 8 mm. [SEP] ### +### ======================================= h_v_q | Gates: 26550 ======================================= ### +### ('cigarette', 0, 0) ('made', 1, 9) ('cigarettes', 2, 1) ('.', 3, 11324) ('somehow', 4, 1772) ### +### ('is', 5, 4034) ('produced', 6, 811) ('eager', 7, 189) ('cigar', 8, 58) ('make', 9, 72) ### +### ('manufactured', 10, 78) ('relating', 11, 25023) ('created', 12, 624) ('making', 13, 161) ### +### ('kelly', 14, 396) ('tobacco', 15, 2) ('how', 16, 24379) ('familiarity', 17, 28658) ### +### ('target', 18, 6934) ('plural', 19, 12271) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('cigarette', 0, 0) ('cigarettes', 2, 1) ('tobacco', 15, 2) ('diameter', 1361, 3) ('wrap', 8584, 4) ### +### ('filters', 8782, 5) ('additive', 7906, 6) ('ˈ', 783, 7) ('mm', 8943, 8) ('made', 1, 9) ### +### ('##per', 6726, 10) ('components', 396, 11) ('today', 5199, 12) ('filter', 68, 13) ### +### ('long', 712, 14) ('unwilling', 638, 15) ('crashing', 310, 16) ('wrapping', 24376, 17) ### +### ('size', 6343, 18) ('dimensions', 6167, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('cigarette', 0, 0) ('made', 1, 9) ('cigarettes', 2, 1) ('tobacco', 15, 2) ('cigar', 8, 58) ### +### ('make', 9, 72) ('manufactured', 10, 78) ('eager', 7, 189) ('makes', 22, 40) ('making', 13, 161) ### +### ('somehow', 4, 1772) ('produced', 6, 811) ('filter', 68, 13) ('smoke', 28, 83) ('simon', 38, 64) ### +### ('smoking', 70, 29) ('created', 12, 624) ('kelly', 14, 396) ('30', 80, 49) ('julian', 65, 81) ### +############################################################################################################ +[2023-10-07 22:54:53,578][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:54:53,578][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:54:53,996][root][INFO] - Epoch: 11: Step: 301/1557, loss[v]=0.041594, lr=0.000009, acc@1[1]=249.0/256=0.97265625, acc@1[2]=254.0/256=0.9921875 +[2023-10-07 22:56:10,998][root][INFO] - Train batch 400 +[2023-10-07 22:56:10,999][root][INFO] - Avg. loss per last 100 batches: 0.064698 +[2023-10-07 22:56:11,705][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29097.9/29522=98.56% | mean: 0.01 | max: 5.54 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.14 | max: 6.38 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how many calories can i burn jumping rope [SEP] ### +### [P_TEXT]: [CLS] can't muster the motivation to make it to the gym? skip it! jumping rope burns more ### +### than 10 calories a minute while strengthening your legs, butt, shoulders, and arms. and it doesn't ### +### take long to reap major rewards. you can burn more than 200 calories in two 10 - minute sessions ### +### each day ( that's 1, 000 calories a week )! jumping rope is also a great way to fit in an effective ### +### cardio session when you're on the goajust toss your jump rope in your carry - on! you'll probably ### +### feel completely energized after jumping rope too. [SEP] ### +### ======================================= h_v_q | Gates: 27224 ======================================= ### +### ('rope', 0, 0) ('jumping', 1, 1) ('cal', 2, 33) ('burn', 3, 21) ('##ories', 4, 47) ### +### ('numerous', 5, 139) ('.', 6, 12950) ('six', 7, 3368) ('jump', 8, 3) ('can', 9, 318) ### +### ('multiple', 10, 222) ('и', 11, 131) ('familiarity', 12, 26995) ('many', 13, 223) ### +### ('relating', 14, 25162) ('plural', 15, 20352) ('32', 16, 10609) ('five', 17, 3651) ### +### ('seven', 18, 10186) ('simon', 19, 71) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('rope', 0, 0) ('jumping', 1, 1) ('muster', 8547, 2) ('jump', 8, 3) ('gym', 5173, 4) ### +### ('burns', 31, 5) ('fit', 8178, 6) ('shoulders', 16681, 7) ('ropes', 33, 8) ('ˈ', 94, 9) ### +### ('motivation', 3563, 10) ('skip', 9459, 11) ('##just', 29502, 12) ('reward', 12709, 13) ### +### ('butt', 8323, 14) ('goa', 13753, 15) ('##io', 15464, 16) ('rewards', 18735, 17) ('legs', 4158, 18) ### +### ('unwilling', 89, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('rope', 0, 0) ('jumping', 1, 1) ('cal', 2, 33) ('burn', 3, 21) ('##ories', 4, 47) ('jump', 8, 3) ### +### ('numerous', 5, 139) ('burns', 31, 5) ('ropes', 33, 8) ('и', 11, 131) ('30', 26, 34) ### +### ('jumped', 28, 31) ('simon', 19, 71) ('20', 24, 54) ('multiple', 10, 222) ('couldn', 30, 65) ### +### ('many', 13, 223) ('can', 9, 318) ('burning', 39, 72) ('ˈ', 94, 9) ### +############################################################################################################ +[2023-10-07 22:56:11,705][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:56:11,705][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:56:12,128][root][INFO] - Epoch: 11: Step: 401/1557, loss[v]=0.063132, lr=0.000009, acc@1[1]=242.0/256=0.9453125, acc@1[2]=252.0/256=0.984375 +[2023-10-07 22:57:28,668][root][INFO] - Train batch 500 +[2023-10-07 22:57:28,669][root][INFO] - Avg. loss per last 100 batches: 0.066218 +[2023-10-07 22:57:29,382][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29179.7/29522=98.84% | mean: 0.01 | max: 4.96 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.6/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.16 | max: 6.07 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a lipid panel with fasting [SEP] ### +### [P_TEXT]: [CLS] fasting for a lipid panel. a lipid panel, which is sometimes known as a lipid ### +### profile, is a blood test that measures for high cholesterol or fat. cholesterol is an important ### +### substance that provides support to the outer membrane of the cell, but too much cholesterol can ### +### also be dangerous. [SEP] ### +### ======================================= h_v_q | Gates: 27604 ======================================= ### +### ('lip', 0, 0) ('fast', 1, 1) ('##id', 2, 30) ('panel', 3, 2) ('is', 4, 1721) ('##ing', 5, 51) ### +### ('refers', 6, 10505) ('encompasses', 7, 11) ('definition', 8, 28) ('relating', 9, 22964) ### +### ('.', 10, 21133) ('plural', 11, 17424) ('familiarity', 12, 28808) ('with', 13, 22648) ### +### ('consisting', 14, 25061) ('provides', 15, 822) ('##sam', 16, 27683) ('stands', 17, 7678) ### +### ('term', 18, 9777) ('simon', 19, 73) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('lip', 0, 0) ('fast', 1, 1) ('panel', 3, 2) ('fat', 863, 3) ('ˈ', 79, 4) ('measures', 1361, 5) ### +### ('dangerous', 785, 6) ('measure', 2837, 7) ('unwilling', 44, 8) ('profile', 4103, 9) ### +### ('crashing', 120, 10) ('encompasses', 7, 11) ('wingspan', 211, 12) ('##terol', 28877, 13) ### +### ('##₂', 63, 14) ('hating', 91, 15) ('panels', 31, 16) ('−', 58, 17) ('gideon', 42, 18) ### +### ('define', 10439, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('lip', 0, 0) ('fast', 1, 1) ('panel', 3, 2) ('##id', 2, 30) ('encompasses', 7, 11) ### +### ('definition', 8, 28) ('##ing', 5, 51) ('simon', 19, 73) ('panels', 31, 16) ('sharply', 33, 25) ### +### ('unwilling', 44, 8) ('##α', 32, 31) ('is', 4, 1721) ('ˈ', 79, 4) ('julian', 24, 100) ### +### ('gideon', 42, 18) ('shoved', 38, 39) ('##₂', 63, 14) ('−', 58, 17) ('hesitated', 67, 26) ### +############################################################################################################ +[2023-10-07 22:57:29,382][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:57:29,382][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:57:29,805][root][INFO] - Epoch: 11: Step: 501/1557, loss[v]=0.067693, lr=0.000009, acc@1[1]=242.5/256=0.947265625, acc@1[2]=252.0/256=0.984375 +[2023-10-07 22:58:46,478][root][INFO] - Train batch 600 +[2023-10-07 22:58:46,479][root][INFO] - Avg. loss per last 100 batches: 0.068827 +[2023-10-07 22:58:47,162][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29128.8/29522=98.67% | mean: 0.01 | max: 5.20 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.16 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the biggest lego city set [SEP] ### +### [P_TEXT]: [CLS] top honors goes to 10189 taj mahal as the largest commercially available set lego ### +### has made to date. just 78 more pieces and it would have topped 6, 000 as the set includes 5, 922 ### +### pieces. the finished model is over 20a³ wide and 16a³ tall and is visually impressive. [SEP] ### +### ======================================= h_v_q | Gates: 25597 ======================================= ### +### ('lego', 0, 0) ('city', 1, 23488) ('biggest', 2, 14) ('largest', 3, 3) ('set', 4, 8) ### +### ('.', 5, 12414) ('is', 6, 2101) ('big', 7, 51) ('encompasses', 8, 308) ('familiarity', 9, 28049) ### +### ('relating', 10, 25430) ('giant', 11, 5448) ('plural', 12, 11288) ('refers', 13, 21547) ### +### ('cities', 14, 1943) ('sets', 15, 43) ('definition', 16, 12476) ('simon', 17, 67) ### +### ('urban', 18, 1571) ('stands', 19, 4618) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('lego', 0, 0) ('honors', 12150, 1) ('mahal', 28186, 2) ('largest', 3, 3) ('pieces', 5999, 4) ### +### ('tall', 4908, 5) ('ˈ', 379, 6) ('ta', 6196, 7) ('set', 4, 8) ('##j', 13151, 9) ('model', 421, 10) ### +### ('wide', 696, 11) ('impressive', 4151, 12) ('top', 20, 13) ('biggest', 2, 14) ### +### ('available', 16002, 15) ('##ο', 158, 16) ('wingspan', 874, 17) ('101', 1461, 18) ### +### ('hesitated', 124, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('lego', 0, 0) ('biggest', 2, 14) ('largest', 3, 3) ('set', 4, 8) ('big', 7, 51) ('top', 20, 13) ### +### ('sets', 15, 43) ('simon', 17, 67) ('encompasses', 8, 308) ('julian', 31, 76) ('is', 6, 2101) ### +### ('##α', 66, 34) ('massive', 23, 169) ('sharply', 82, 30) ('hesitated', 124, 19) ('##ο', 158, 16) ### +### ('ˈ', 379, 6) ('fernando', 105, 39) ('bigger', 27, 200) ('gideon', 108, 37) ### +############################################################################################################ +[2023-10-07 22:58:47,162][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 22:58:47,162][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 22:58:47,565][root][INFO] - Epoch: 11: Step: 601/1557, loss[v]=0.056745, lr=0.000009, acc@1[1]=243.5/256=0.951171875, acc@1[2]=252.0/256=0.984375 +[2023-10-07 23:00:04,343][root][INFO] - Train batch 700 +[2023-10-07 23:00:04,344][root][INFO] - Avg. loss per last 100 batches: 0.066230 +[2023-10-07 23:00:05,044][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29133.9/29522=98.69% | mean: 0.01 | max: 5.21 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.26 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] cost for dog c section [SEP] ### +### [P_TEXT]: [CLS] according to dogbreedinfo. com, a dog c - section will cost you between $ 500 and $ ### +### 2, 000. the exact price of the procedure may depend on the vet who will handle the procedure. a ### +### normal easy c - section can be about $ 500 when done during office hours. [SEP] ### +### ======================================= h_v_q | Gates: 26626 ======================================= ### +### ('section', 0, 0) ('c', 1, 7) ('dog', 2, 2) ('$', 3, 11) ('sections', 4, 5) ('##£', 5, 22708) ### +### ('cost', 6, 1) ('part', 7, 919) ('.', 8, 15406) ('division', 9, 140) ('dogs', 10, 4) ### +### ('segment', 11, 274) ('familiarity', 12, 27508) ('plural', 13, 20442) ('costs', 14, 8) ### +### ('stylized', 15, 29052) ('price', 16, 6) ('relating', 17, 28090) ('cents', 18, 1338) ### +### ('fee', 19, 36) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('section', 0, 0) ('cost', 6, 1) ('dog', 2, 2) ('easy', 9687, 3) ('dogs', 10, 4) ('sections', 4, 5) ### +### ('price', 16, 6) ('c', 1, 7) ('costs', 14, 8) ('vet', 23991, 9) ('ˈ', 267, 10) ('$', 3, 11) ### +### ('wingspan', 849, 12) ('normal', 2608, 13) ('unwilling', 145, 14) ('##ο', 162, 15) ### +### ('hesitated', 123, 16) ('prices', 48, 17) ('crashing', 329, 18) ('procedures', 13060, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('section', 0, 0) ('dog', 2, 2) ('c', 1, 7) ('$', 3, 11) ('sections', 4, 5) ('cost', 6, 1) ### +### ('dogs', 10, 4) ('price', 16, 6) ('costs', 14, 8) ('division', 9, 140) ('fee', 19, 36) ### +### ('prices', 48, 17) ('segment', 11, 274) ('##section', 24, 90) ('horse', 20, 185) ('##α', 67, 23) ### +### ('expensive', 44, 66) ('part', 7, 919) ('simon', 39, 92) ('##₂', 89, 24) ### +############################################################################################################ +[2023-10-07 23:00:05,044][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:00:05,044][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:00:05,468][root][INFO] - Epoch: 11: Step: 701/1557, loss[v]=0.069688, lr=0.000009, acc@1[1]=240.5/256=0.939453125, acc@1[2]=251.5/256=0.982421875 +[2023-10-07 23:01:22,514][root][INFO] - Train batch 800 +[2023-10-07 23:01:22,515][root][INFO] - Avg. loss per last 100 batches: 0.067408 +[2023-10-07 23:01:23,226][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29200.8/29522=98.91% | mean: 0.01 | max: 5.06 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.22 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] longest increasing sequence [SEP] ### +### [P_TEXT]: [CLS] let lis ( e ) denote the length of the longest increasing sequence in e. a sequence ### +### eis monotone if e ( 1 ) e ( 2 ) e ( n ). we consider edit operations on a sequence. where a single ### +### edit operation involves deleting a single element of the sequence and inserting it in. a new place. ### +### [SEP] ### +### ======================================= h_v_q | Gates: 26567 ======================================= ### +### ('sequence', 0, 4) ('longest', 1, 3) ('increasing', 2, 13) ('growing', 3, 312) ('increase', 4, 250) ### +### ('.', 5, 16100) ('decreasing', 6, 764) ('increased', 7, 486) ('expanding', 8, 692) ### +### ('familiarity', 9, 27487) ('sequences', 10, 30) ('largest', 11, 38) ('declining', 12, 900) ### +### ('plural', 13, 11741) ('long', 14, 661) ('succession', 15, 1281) ('rising', 16, 132) ### +### ('advancing', 17, 1956) ('relating', 18, 24593) ('stylized', 19, 26112) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('li', 10312, 0) ('edit', 7139, 1) ('mono', 3475, 2) ('longest', 1, 3) ('sequence', 0, 4) ### +### ('##tone', 1210, 5) ('e', 5800, 6) ('length', 55, 7) ('ˈ', 317, 8) ('single', 1502, 9) ### +### ('element', 1479, 10) ('wingspan', 502, 11) ('unwilling', 218, 12) ('increasing', 2, 13) ### +### ('operations', 2377, 14) ('operation', 900, 15) ('##₂', 86, 16) ('denote', 3476, 17) ### +### ('hating', 264, 18) ('crashing', 282, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('sequence', 0, 4) ('longest', 1, 3) ('increasing', 2, 13) ('sequences', 10, 30) ### +### ('growing', 3, 312) ('increase', 4, 250) ('largest', 11, 38) ('increased', 7, 486) ### +### ('decreasing', 6, 764) ('julian', 22, 66) ('expanding', 8, 692) ('length', 55, 7) ### +### ('rising', 16, 132) ('simon', 37, 69) ('##₂', 86, 16) ('##α', 64, 34) ('increases', 26, 169) ### +### ('shoved', 68, 40) ('sharply', 76, 39) ('shortest', 44, 96) ### +############################################################################################################ +[2023-10-07 23:01:23,226][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:01:23,226][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:01:23,632][root][INFO] - Epoch: 11: Step: 801/1557, loss[v]=0.092301, lr=0.000009, acc@1[1]=243.0/256=0.94921875, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 23:02:40,293][root][INFO] - Train batch 900 +[2023-10-07 23:02:40,294][root][INFO] - Avg. loss per last 100 batches: 0.065302 +[2023-10-07 23:02:41,024][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29158.6/29522=98.77% | mean: 0.01 | max: 5.29 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.17 | max: 6.18 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what are the general ranks and stars [SEP] ### +### [P_TEXT]: [CLS] general of the army ( ga ) is a five - star general officer and is the second ### +### highest possible rank in the united states army. a special rank of general of the armies, which ### +### ranks above general of the army, does exist but has only been conferred twice in the history of the ### +### army. [SEP] ### +### ======================================= h_v_q | Gates: 27343 ======================================= ### +### ('ranks', 0, 27) ('stars', 1, 32) ('general', 2, 2) ('rank', 3, 3) ('star', 4, 8) ### +### ('familiarity', 5, 28146) ('are', 6, 3109) ('.', 7, 8887) ('ranked', 8, 896) ('starring', 9, 717) ### +### ('include', 10, 461) ('generals', 11, 7) ('encompasses', 12, 16) ('plural', 13, 13393) ### +### ('stylized', 14, 27940) ('consisting', 15, 18451) ('relating', 16, 24819) ('ranking', 17, 238) ### +### ('what', 18, 122) ('position', 19, 133) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('ga', 2724, 0) ('army', 2058, 1) ('general', 2, 2) ('rank', 3, 3) ('armies', 9343, 4) ### +### ('highest', 931, 5) ('ˈ', 91, 6) ('generals', 11, 7) ('star', 4, 8) ('officers', 298, 9) ### +### ('possible', 1999, 10) ('military', 163, 11) ('crashing', 130, 12) ('hating', 328, 13) ### +### ('##ο', 169, 14) ('##大', 362, 15) ('encompasses', 12, 16) ('wingspan', 389, 17) ('−', 302, 18) ### +### ('officer', 1238, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('general', 2, 2) ('ranks', 0, 27) ('rank', 3, 3) ('stars', 1, 32) ('star', 4, 8) ### +### ('generals', 11, 7) ('encompasses', 12, 16) ('colonel', 30, 23) ('simon', 24, 55) ('ˈ', 91, 6) ### +### ('gideon', 44, 25) ('##α', 57, 22) ('ranked', 8, 896) ('unwilling', 66, 20) ('julian', 36, 63) ### +### ('shoved', 42, 46) ('hesitated', 86, 26) ('what', 18, 122) ('starring', 9, 717) ### +### ('position', 19, 133) ### +############################################################################################################ +[2023-10-07 23:02:41,024][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:02:41,024][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:02:41,454][root][INFO] - Epoch: 11: Step: 901/1557, loss[v]=0.078694, lr=0.000009, acc@1[1]=241.5/256=0.943359375, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 23:03:57,752][root][INFO] - Train batch 1000 +[2023-10-07 23:03:57,753][root][INFO] - Avg. loss per last 100 batches: 0.065777 +[2023-10-07 23:03:58,434][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29162.4/29522=98.78% | mean: 0.01 | max: 5.57 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.31 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the speed of light [SEP] ### +### [P_TEXT]: [CLS] speed of light, speed at which light waves propagate through different materials. ### +### in particular, the value for the speed of light in a vacuum is now defined as exactly 299, 792, 458 ### +### metres per second. play _ circle _ outline. scientists simulating traveling at the speed of light. ### +### contunico a© zdf enterprises gmbh, mainz. [SEP] ### +### ======================================= h_v_q | Gates: 26896 ======================================= ### +### ('light', 0, 4) ('speed', 1, 1) ('plural', 2, 15578) ('familiarity', 3, 28052) ### +### ('relating', 4, 24871) ('speeds', 5, 2) ('stylized', 6, 28234) ('consisting', 7, 23066) ### +### ('is', 8, 1806) ('velocity', 9, 61) ('refers', 10, 7768) ('of', 11, 8587) ### +### ('mathematics', 12, 22980) ('lights', 13, 22) ('.', 14, 11985) ('frequency', 15, 350) ### +### ('encompasses', 16, 28) ('definition', 17, 15) ('gideon', 18, 52) ('heavy', 19, 410) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('vacuum', 5164, 0) ('speed', 1, 1) ('speeds', 5, 2) ('metres', 979, 3) ('light', 0, 4) ### +### ('ˈ', 155, 5) ('lighting', 28, 6) ('gmbh', 19633, 7) ('exactly', 4541, 8) ('circle', 2303, 9) ### +### ('300', 813, 10) ('waves', 7600, 11) ('##df', 27651, 12) ('wingspan', 431, 13) ('fast', 40, 14) ### +### ('definition', 17, 15) ('value', 4778, 16) ('##ο', 104, 17) ('crashing', 215, 18) ### +### ('traveling', 4599, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('speed', 1, 1) ('light', 0, 4) ('speeds', 5, 2) ('lights', 13, 22) ('velocity', 9, 61) ### +### ('definition', 17, 15) ('encompasses', 16, 28) ('lighting', 28, 6) ('gideon', 18, 52) ### +### ('fast', 40, 14) ('simon', 22, 60) ('shoved', 29, 45) ('julian', 30, 51) ('unwilling', 50, 24) ### +### ('hesitated', 61, 21) ('##α', 43, 34) ('sharply', 38, 43) ('ˈ', 155, 5) ('##ο', 104, 17) ### +### ('lama', 60, 44) ### +############################################################################################################ +[2023-10-07 23:03:58,434][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:03:58,434][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:03:58,839][root][INFO] - Epoch: 11: Step: 1001/1557, loss[v]=0.065682, lr=0.000009, acc@1[1]=239.5/256=0.935546875, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 23:05:15,379][root][INFO] - Train batch 1100 +[2023-10-07 23:05:15,379][root][INFO] - Avg. loss per last 100 batches: 0.068815 +[2023-10-07 23:05:16,062][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29099.5/29522=98.57% | mean: 0.01 | max: 5.50 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.17 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is mcafee protection [SEP] ### +### [P_TEXT]: [CLS] mcafee is an antivirus software company that is now owned by the intel corporation. ### +### intel knows a thing or two about protecting devices from threats. mcafee offers excellent antivirus ### +### solutions, a 30 - day money - back guarantee and free support. and itas always improving its ### +### products to make sure they offer best - in - class virus, malware and spyware protection. [SEP] ### +### ======================================= h_v_q | Gates: 27287 ======================================= ### +### ('mca', 0, 1) ('protection', 1, 21) ('##fe', 2, 3) ('##e', 3, 11) ('is', 4, 211) ### +### ('encompasses', 5, 10) ('familiarity', 6, 26978) ('refers', 7, 14112) ('consisting', 8, 24592) ### +### ('relating', 9, 25906) ('stylized', 10, 27837) ('stands', 11, 4607) ('protect', 12, 16) ### +### ('protected', 13, 59) ('plural', 14, 16964) ('protects', 15, 82) ('definition', 16, 127) ### +### ('support', 17, 25) ('mc', 18, 74) ('something', 19, 8525) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('intel', 18873, 0) ('mca', 0, 1) ('##virus', 19457, 2) ('##fe', 2, 3) ('ˈ', 167, 4) ### +### ('threats', 4544, 5) ('virus', 5012, 6) ('##ware', 12331, 7) ('spy', 5830, 8) ### +### ('guarantee', 6184, 9) ('encompasses', 5, 10) ('##e', 3, 11) ('owned', 335, 12) ### +### ('software', 42, 13) ('crashing', 66, 14) ('30', 7894, 15) ('protect', 12, 16) ('##ο', 123, 17) ### +### ('##₂', 246, 18) ('money', 495, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('mca', 0, 1) ('##fe', 2, 3) ('protection', 1, 21) ('##e', 3, 11) ('encompasses', 5, 10) ### +### ('is', 4, 211) ('protect', 12, 16) ('support', 17, 25) ('protected', 13, 59) ('protects', 15, 82) ### +### ('mc', 18, 74) ('software', 42, 13) ('shoved', 24, 46) ('##α', 34, 37) ('unwilling', 39, 32) ### +### ('crashing', 66, 14) ('definition', 16, 127) ('hesitated', 41, 40) ('simon', 33, 55) ### +### ('julian', 30, 68) ### +############################################################################################################ +[2023-10-07 23:05:16,063][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:05:16,063][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:05:16,487][root][INFO] - Epoch: 11: Step: 1101/1557, loss[v]=0.052209, lr=0.000009, acc@1[1]=242.0/256=0.9453125, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 23:06:33,484][root][INFO] - Train batch 1200 +[2023-10-07 23:06:33,485][root][INFO] - Avg. loss per last 100 batches: 0.069659 +[2023-10-07 23:06:34,208][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29131.0/29522=98.68% | mean: 0.01 | max: 5.39 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.17 | max: 6.26 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] micrometeoroid meaning [SEP] ### +### [P_TEXT]: [CLS] freebase ( 0. 00 / 0 votes ) rate this definition : micrometeoroid. a ### +### micrometeoroid is a tiny meteoroid ; a small particle of rock in space, usually weighing less than ### +### a gram. a micrometeorite is such a particle that survives passage through the earth's atmosphere ### +### and reaches the earth's surface. [SEP] ### +### ======================================= h_v_q | Gates: 27683 ======================================= ### +### ('micro', 0, 1) ('##eo', 1, 4) ('##roid', 2, 12) ('##met', 3, 10) ('noun', 4, 21812) ### +### ('definition', 5, 11) ('meaning', 6, 29) ('familiarity', 7, 28119) ('consisting', 8, 17146) ### +### ('stylized', 9, 25187) ('mini', 10, 73) ('something', 11, 6849) ('relating', 12, 26600) ### +### ('plural', 13, 13109) ('sense', 14, 8798) ('or', 15, 20182) ('.', 16, 13825) ('symbol', 17, 2143) ### +### ('latin', 18, 5541) (';', 19, 5107) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('meteor', 8961, 0) ('micro', 0, 1) ('##base', 26074, 2) ('particle', 871, 3) ('##eo', 1, 4) ### +### ('survives', 19562, 5) ('definitions', 281, 6) ('ˈ', 85, 7) ('tiny', 182, 8) ('particles', 3993, 9) ### +### ('##met', 3, 10) ('definition', 5, 11) ('##roid', 2, 12) ('survive', 529, 13) ('crashing', 74, 14) ### +### ('define', 7449, 15) ('##₂', 53, 16) ('wingspan', 290, 17) ('hating', 66, 18) ('space', 1239, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('micro', 0, 1) ('##eo', 1, 4) ('##roid', 2, 12) ('##met', 3, 10) ('definition', 5, 11) ### +### ('meaning', 6, 29) ('mini', 10, 73) ('unwilling', 41, 21) ('hugh', 37, 41) ('##₂', 53, 16) ### +### ('ˈ', 85, 7) ('shoved', 43, 40) ('julian', 29, 78) ('defined', 31, 77) ('crashing', 74, 14) ### +### ('simon', 36, 66) ('hating', 66, 18) ('nano', 27, 99) ('##α', 46, 44) ('sharply', 54, 31) ### +############################################################################################################ +[2023-10-07 23:06:34,209][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:06:34,209][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:06:34,633][root][INFO] - Epoch: 11: Step: 1201/1557, loss[v]=0.109986, lr=0.000009, acc@1[1]=239.0/256=0.93359375, acc@1[2]=245.0/256=0.95703125 +[2023-10-07 23:07:50,403][root][INFO] - Train batch 1300 +[2023-10-07 23:07:50,404][root][INFO] - Avg. loss per last 100 batches: 0.071500 +[2023-10-07 23:07:51,115][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29074.0/29522=98.48% | mean: 0.01 | max: 5.40 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.15 | max: 6.18 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a single focus lens? [SEP] ### +### [P_TEXT]: [CLS] a single vision intraocular lens focuses for a single distance. if you choose a ### +### single vision iol set for distance, you would be likely be able to drive and watch tv without ### +### glasses, but would need glasses to focus for intermediate and near tasks. [SEP] ### +### ======================================= h_v_q | Gates: 25930 ======================================= ### +### ('lens', 0, 0) ('single', 1, 7) ('focus', 2, 3) ('refers', 3, 23494) ('encompasses', 4, 281) ### +### ('focusing', 5, 182) ('is', 6, 13375) ('definition', 7, 341) ('familiarity', 8, 25849) ### +### ('relating', 9, 25279) ('lenses', 10, 18) ('noun', 11, 27757) ('concentrate', 12, 222) ### +### ('stands', 13, 6834) ('consisting', 14, 24485) ('singles', 15, 64) ('.', 16, 15198) ### +### ('focused', 17, 53) ('encyclopedia', 18, 10189) ('one', 19, 679) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('lens', 0, 0) ('vision', 142, 1) ('distance', 4581, 2) ('focus', 2, 3) ('ˈ', 313, 4) ### +### ('focuses', 37, 5) ('io', 24672, 6) ('single', 1, 7) ('glasses', 1490, 8) ('distances', 23458, 9) ### +### ('crashing', 226, 10) ('##ular', 7716, 11) ('intermediate', 5944, 12) ('##₂', 152, 13) ### +### ('##ο', 338, 14) ('abandon', 289, 15) ('hesitated', 132, 16) ('unwilling', 208, 17) ### +### ('lenses', 10, 18) ('hating', 295, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('lens', 0, 0) ('focus', 2, 3) ('single', 1, 7) ('lenses', 10, 18) ('focusing', 5, 182) ### +### ('singles', 15, 64) ('encompasses', 4, 281) ('focused', 17, 53) ('definition', 7, 341) ### +### ('focuses', 37, 5) ('concentrate', 12, 222) ('vision', 142, 1) ('sharply', 46, 20) ### +### ('simon', 38, 44) ('julian', 29, 115) ('hesitated', 132, 16) ('camera', 42, 96) ('##α', 68, 52) ### +### ('##₂', 152, 13) ('ˈ', 313, 4) ### +############################################################################################################ +[2023-10-07 23:07:51,116][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:07:51,116][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:07:51,516][root][INFO] - Epoch: 11: Step: 1301/1557, loss[v]=0.099928, lr=0.000009, acc@1[1]=234.5/256=0.916015625, acc@1[2]=248.5/256=0.970703125 +[2023-10-07 23:09:08,401][root][INFO] - Train batch 1400 +[2023-10-07 23:09:08,401][root][INFO] - Avg. loss per last 100 batches: 0.066008 +[2023-10-07 23:09:09,085][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29185.5/29522=98.86% | mean: 0.01 | max: 5.85 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.50 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] mosaic hbo series cast [SEP] ### +### [P_TEXT]: [CLS] the cast and crew of mosaic - sharon stone, beau bridges, garrett hedlund, ### +### frederick weller, paul reubens, jennifer ferrin, devin ratray, michael cerveris, james ransone, ### +### jeremy bobb and maya kazan - talk about the murder mystery experience that lets you determine the ### +### outcome of the story. [SEP] ### +### ======================================= h_v_q | Gates: 27454 ======================================= ### +### ('mosaic', 0, 0) ('hbo', 1, 5105) ('cast', 2, 1) ('series', 3, 4739) ('starring', 4, 990) ### +### ('.', 5, 9702) ('portrayed', 6, 241) ('film', 7, 2510) ('television', 8, 4278) ### +### ('familiarity', 9, 28026) ('program', 10, 2506) ('drama', 11, 2167) ('actor', 12, 200) ### +### ('network', 13, 2009) ('smith', 14, 3505) ('heritage', 15, 476) ('show', 16, 1681) ### +### ('fabric', 17, 2694) ('presenting', 18, 553) ('tapestry', 19, 73) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('mosaic', 0, 0) ('cast', 2, 1) ('kazan', 26469, 2) ('beau', 8805, 3) ('bridges', 13749, 4) ### +### ('stone', 340, 5) ('sharon', 2266, 6) ('reuben', 4277, 7) ('##ray', 23836, 8) ('mystery', 2306, 9) ### +### ('outcome', 14669, 10) ('murder', 2359, 11) ('garrett', 1723, 12) ('ˈ', 602, 13) ### +### ('experience', 9374, 14) ('casts', 207, 15) ('crew', 680, 16) ('jennifer', 2621, 17) ### +### ('crashing', 1386, 18) ('jeremy', 1879, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('mosaic', 0, 0) ('cast', 2, 1) ('hbo', 1, 5105) ('portrayed', 6, 241) ('starring', 4, 990) ### +### ('tapestry', 19, 73) ('series', 3, 4739) ('actor', 12, 200) ('stone', 340, 5) ('actors', 61, 63) ### +### ('casts', 207, 15) ('film', 7, 2510) ('roles', 35, 213) ('complex', 29, 244) ('heritage', 15, 476) ### +### ('.', 5, 9702) ('sanctuary', 113, 105) ('ˈ', 602, 13) ('pyramid', 72, 136) ('presenter', 180, 71) ### +############################################################################################################ +[2023-10-07 23:09:09,085][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:09:09,085][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:09:09,491][root][INFO] - Epoch: 11: Step: 1401/1557, loss[v]=0.084091, lr=0.000009, acc@1[1]=239.5/256=0.935546875, acc@1[2]=249.5/256=0.974609375 +[2023-10-07 23:10:26,894][root][INFO] - Train batch 1500 +[2023-10-07 23:10:26,895][root][INFO] - Avg. loss per last 100 batches: 0.067558 +[2023-10-07 23:10:27,593][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29193.8/29522=98.89% | mean: 0.01 | max: 5.71 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.52 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] do monoecious plants prevent self pollination [SEP] ### +### [P_TEXT]: [CLS] both hermaphrodite and monoecious species have the potential for self - pollination ### +### leading to self - fertilization unless there is a mechanism to avoid it. eighty percent of all ### +### flowering plants are hermaphroditic, meaning they contain both sexes in the same flower, while 5 ### +### percent of plant species are monoecious. [SEP] ### +### ======================================= h_v_q | Gates: 28142 ======================================= ### +### ('mono', 0, 9) ('poll', 1, 10) ('##ious', 2, 5) ('self', 3, 4) ('prevent', 4, 207) ### +### ('##ination', 5, 42) ('##ec', 6, 33) ('plants', 7, 12) ('.', 8, 15685) ('familiarity', 9, 26617) ### +### ('do', 10, 2864) ('consisting', 11, 22347) ('plant', 12, 60) ('stylized', 13, 26947) ### +### ('relating', 14, 23638) ('reduce', 15, 1006) ('simon', 16, 65) ('avoid', 17, 70) ### +### ('sharply', 18, 19) ('stop', 19, 1408) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('flowering', 578, 0) ('##rod', 23689, 1) ('ˈ', 41, 2) ('species', 87, 3) ('self', 3, 4) ### +### ('##ious', 2, 5) ('flower', 2751, 6) ('hating', 44, 7) ('potential', 9746, 8) ('mono', 0, 9) ### +### ('poll', 1, 10) ('##lization', 9619, 11) ('plants', 7, 12) ('##ο', 30, 13) ('unwilling', 29, 14) ### +### ('sexes', 27148, 15) ('percent', 6597, 16) ('wingspan', 124, 17) ('##itic', 25472, 18) ### +### ('sharply', 18, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('mono', 0, 9) ('##ious', 2, 5) ('poll', 1, 10) ('self', 3, 4) ('plants', 7, 12) ### +### ('##ination', 5, 42) ('##ec', 6, 33) ('prevent', 4, 207) ('sharply', 18, 19) ('plant', 12, 60) ### +### ('unwilling', 29, 14) ('simon', 16, 65) ('##ο', 30, 13) ('avoid', 17, 70) ('ˈ', 41, 2) ### +### ('−', 28, 24) ('hugh', 26, 41) ('##₂', 35, 20) ('hating', 44, 7) ('##α', 36, 26) ### +############################################################################################################ +[2023-10-07 23:10:27,594][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:10:27,594][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:10:28,018][root][INFO] - Epoch: 11: Step: 1501/1557, loss[v]=0.077853, lr=0.000008, acc@1[1]=239.5/256=0.935546875, acc@1[2]=251.5/256=0.982421875 +[2023-10-07 23:11:11,010][root][INFO] - rank=1; last iteration 1557 +[2023-10-07 23:11:11,010][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 23:11:11,011][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-07 23:11:11,013][root][INFO] - rank=0; last iteration 1557 +[2023-10-07 23:11:11,013][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 23:11:11,013][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-07 23:11:11,013][root][INFO] - rank=3; last iteration 1557 +[2023-10-07 23:11:11,014][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 23:11:11,014][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-07 23:11:11,016][root][INFO] - rank=2; last iteration 1557 +[2023-10-07 23:11:11,016][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 23:11:11,016][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-07 23:11:11,018][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 23:11:11,019][root][INFO] - Epoch finished on 1 +[2023-10-07 23:11:11,021][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 23:11:11,021][root][INFO] - Epoch finished on 0 +[2023-10-07 23:11:11,023][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 23:11:11,023][root][INFO] - Epoch finished on 3 +[2023-10-07 23:11:11,024][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 23:11:11,024][root][INFO] - Epoch finished on 2 +[2023-10-07 23:11:21,657][root][INFO] - Saved checkpoint at ./vdr_11 +[2023-10-07 23:11:21,657][root][INFO] - Saved checkpoint at ./vdr_11 +[2023-10-07 23:11:21,658][root][INFO] - Av Loss per epoch=0.067453 +[2023-10-07 23:11:21,658][root][INFO] - epoch total (1) correct predictions=378527 +[2023-10-07 23:11:21,658][root][INFO] - Av Loss per epoch=0.067453 +[2023-10-07 23:11:21,658][root][INFO] - epoch total (2) correct predictions=390819 +[2023-10-07 23:11:21,658][root][INFO] - epoch total (1) correct predictions=378527 +[2023-10-07 23:11:21,658][root][INFO] - epoch total (2) correct predictions=390819 +[2023-10-07 23:11:21,659][root][INFO] - Saved checkpoint at ./vdr_11 +[2023-10-07 23:11:21,660][root][INFO] - Av Loss per epoch=0.067453 +[2023-10-07 23:11:21,660][root][INFO] - epoch total (1) correct predictions=378527 +[2023-10-07 23:11:21,660][root][INFO] - epoch total (2) correct predictions=390819 +[2023-10-07 23:11:21,659][root][INFO] - Saved checkpoint at ./vdr_11 +[2023-10-07 23:11:21,660][root][INFO] - Av Loss per epoch=0.067453 +[2023-10-07 23:11:21,661][root][INFO] - epoch total (1) correct predictions=378527 +[2023-10-07 23:11:21,661][root][INFO] - epoch total (2) correct predictions=390819 +[2023-10-07 23:11:21,662][root][INFO] - ***** Epoch 12 ***** +[2023-10-07 23:11:21,664][root][INFO] - ***** Epoch 12 ***** +[2023-10-07 23:11:21,664][root][INFO] - ***** Epoch 12 ***** +[2023-10-07 23:11:21,669][root][INFO] - rank=3; Iteration start +[2023-10-07 23:11:21,670][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 23:11:21,670][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 23:11:21,670][root][INFO] - rank=0; Iteration start +[2023-10-07 23:11:21,670][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 23:11:21,671][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 23:11:21,671][root][INFO] - rank=1; Iteration start +[2023-10-07 23:11:21,671][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 23:11:21,671][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 23:11:21,672][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-07 23:11:21,670][root][INFO] - ***** Epoch 12 ***** +[2023-10-07 23:11:21,672][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-07 23:11:21,673][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-07 23:11:21,677][root][INFO] - rank=2; Iteration start +[2023-10-07 23:11:21,677][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 23:11:21,677][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 23:11:21,679][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-07 23:11:22,689][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29147.9/29522=98.73% | mean: 0.01 | max: 5.36 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.23 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what causes blood in urine with no infection [SEP] ### +### [P_TEXT]: [CLS] since then i have had 2 urine tests within 24 hours of having the first symptoms. ### +### both tests ive done show blood in the urine, but no infection. so my doctor today has reffered me ### +### for an ultra sound. this can test for a number of things - kidney stones, kidney infection, kidney ### +### cancer and bladder cancer etc. none of which are more likely the the other. [SEP] ### +### ======================================= h_v_q | Gates: 27471 ======================================= ### +### ('urine', 0, 2) ('blood', 1, 35) ('infection', 2, 30) ('no', 3, 36) ('.', 4, 15321) ### +### ('causes', 5, 726) ('caused', 6, 9406) ('cause', 7, 16137) ('minimal', 8, 144) ### +### ('familiarity', 9, 24977) ('nothing', 10, 107) ('stylized', 11, 28629) ('simon', 12, 83) ### +### ('neither', 13, 280) ('consisting', 14, 23038) ('nobody', 15, 1461) ('none', 16, 205) ### +### ('with', 17, 12713) ('∈', 18, 19) ('without', 19, 546) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('ultra', 5803, 0) ('ˈ', 72, 1) ('urine', 0, 2) ('kidney', 1362, 3) ('etc', 20307, 4) ### +### ('iv', 2172, 5) ('wingspan', 132, 6) ('##ο', 58, 7) ('cyrillic', 339, 8) ('sound', 1902, 9) ### +### ('unwilling', 37, 10) ('crashing', 196, 11) ('hating', 60, 12) ('##大', 70, 13) ('sharply', 29, 14) ### +### ('##ང', 179, 15) ('hesitated', 97, 16) ('##₂', 63, 17) ('−', 53, 18) ('∈', 18, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('urine', 0, 2) ('blood', 1, 35) ('infection', 2, 30) ('no', 3, 36) ('∈', 18, 19) ### +### ('nothing', 10, 107) ('minimal', 8, 144) ('simon', 12, 83) ('causes', 5, 726) ('sharply', 29, 14) ### +### ('unwilling', 37, 10) ('ˈ', 72, 1) ('angrily', 39, 32) ('infections', 22, 150) ('##ο', 58, 7) ### +### ('none', 16, 205) ('hugh', 38, 50) ('anton', 34, 75) ('−', 53, 18) ('hating', 60, 12) ### +############################################################################################################ +[2023-10-07 23:11:22,690][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:11:22,690][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:11:23,106][root][INFO] - Epoch: 12: Step: 1/1557, loss[v]=0.086864, lr=0.000008, acc@1[1]=245.0/256=0.95703125, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 23:12:39,642][root][INFO] - Train batch 100 +[2023-10-07 23:12:39,643][root][INFO] - Avg. loss per last 100 batches: 0.063689 +[2023-10-07 23:12:40,348][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29174.8/29522=98.82% | mean: 0.01 | max: 5.39 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.20 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] did pacquiao retire [SEP] ### +### [P_TEXT]: [CLS] lance pugmirecontact reporter. manny pacquiao has stepped into retirement alongside ### +### floyd mayweather jr., so now comes the matter of waiting to see whether it represents the dawn of a ### +### new era in boxing or whether the temptations of their rematch, massive cash and their lifelong ### +### pursuits will prove irresistible. [SEP] ### +### ======================================= h_v_q | Gates: 27876 ======================================= ### +### ('pac', 0, 6) ('##qui', 1, 7) ('##ao', 2, 21) ('retirement', 3, 5) ('retire', 4, 32) ### +### ('retired', 5, 29) ('did', 6, 7118) ('retiring', 7, 71) ('was', 8, 6151) ('familiarity', 9, 27417) ### +### ('knew', 10, 341) ('.', 11, 12847) ('stylized', 12, 26245) ('consisting', 13, 24293) ### +### ('died', 14, 702) ('became', 15, 3686) ('relating', 16, 27439) ('aired', 17, 21075) ### +### ('simon', 18, 94) ('remained', 19, 401) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('lance', 6429, 0) ('boxing', 2525, 1) ('##weather', 22702, 2) ('reporter', 11000, 3) ### +### ('floyd', 3881, 4) ('retirement', 3, 5) ('pac', 0, 6) ('##qui', 1, 7) ('manny', 15875, 8) ### +### ('ˈ', 32, 9) ('dawn', 5571, 10) ('pursuits', 1673, 11) ('boxers', 20630, 12) ('era', 4870, 13) ### +### ('alongside', 642, 14) ('pu', 15880, 15) ('rematch', 24973, 16) ('temptation', 7163, 17) ### +### ('wingspan', 58, 18) ('##ο', 41, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('pac', 0, 6) ('##qui', 1, 7) ('retirement', 3, 5) ('##ao', 2, 21) ('retire', 4, 32) ### +### ('retired', 5, 29) ('retiring', 7, 71) ('ˈ', 32, 9) ('##α', 22, 53) ('unwilling', 29, 33) ### +### ('##ο', 41, 19) ('simon', 18, 94) ('−', 26, 57) ('sharply', 36, 46) ('hesitated', 37, 47) ### +### ('hugh', 34, 58) ('wingspan', 58, 18) ('knew', 10, 341) ('crashed', 52, 35) ('##₂', 57, 36) ### +############################################################################################################ +[2023-10-07 23:12:40,349][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:12:40,349][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:12:40,765][root][INFO] - Epoch: 12: Step: 101/1557, loss[v]=0.056781, lr=0.000008, acc@1[1]=246.5/256=0.962890625, acc@1[2]=252.0/256=0.984375 +[2023-10-07 23:13:57,739][root][INFO] - Train batch 200 +[2023-10-07 23:13:57,740][root][INFO] - Avg. loss per last 100 batches: 0.062666 +[2023-10-07 23:13:58,438][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29145.8/29522=98.73% | mean: 0.01 | max: 5.66 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.46 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] todd rundgren net worth [SEP] ### +### [P_TEXT]: [CLS] todd rundgren net worth is $ 10 million. todd rundgren net worth : todd rundgren is ### +### an american musician and producer with a net worth of $ 10 million dollars. born in upper darby, ### +### pennsylvania, todd rundgren, began playing in garage rock and psychede [SEP] ### +### ======================================= h_v_q | Gates: 27613 ======================================= ### +### ('##gren', 0, 1) ('todd', 1, 0) ('worth', 2, 2) ('run', 3, 4) ('net', 4, 19) ('##d', 5, 41) ### +### ('$', 6, 8) ('.', 7, 5388) ('ran', 8, 85) ('familiarity', 9, 27971) ('running', 10, 65) ### +### ('stylized', 11, 27454) ('##£', 12, 23161) ('relating', 13, 26519) ('consisting', 14, 21255) ### +### ('wealth', 15, 59) ('runs', 16, 51) ('currency', 17, 108) ('simon', 18, 49) ('value', 19, 111) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('todd', 1, 0) ('##gren', 0, 1) ('worth', 2, 2) ('musician', 6944, 3) ('run', 3, 4) ### +### ('producer', 583, 5) ('ˈ', 141, 6) ('darby', 21803, 7) ('$', 6, 8) ('pennsylvania', 1499, 9) ### +### ('upper', 6586, 10) ('garage', 7395, 11) ('##ο', 32, 12) ('wingspan', 131, 13) ### +### ('psyche', 18888, 14) ('unwilling', 39, 15) ('crashing', 115, 16) ('stumbled', 124, 17) ### +### ('##大', 150, 18) ('net', 4, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##gren', 0, 1) ('todd', 1, 0) ('worth', 2, 2) ('run', 3, 4) ('net', 4, 19) ('$', 6, 8) ### +### ('##d', 5, 41) ('ran', 8, 85) ('running', 10, 65) ('wealth', 15, 59) ('runs', 16, 51) ### +### ('##ο', 32, 12) ('simon', 18, 49) ('unwilling', 39, 15) ('currency', 17, 108) ('sharply', 28, 31) ### +### ('##α', 35, 26) ('ˈ', 141, 6) ('money', 24, 86) ('value', 19, 111) ### +############################################################################################################ +[2023-10-07 23:13:58,439][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:13:58,439][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:13:58,842][root][INFO] - Epoch: 12: Step: 201/1557, loss[v]=0.094815, lr=0.000008, acc@1[1]=245.5/256=0.958984375, acc@1[2]=248.0/256=0.96875 +[2023-10-07 23:15:14,602][root][INFO] - Train batch 300 +[2023-10-07 23:15:14,602][root][INFO] - Avg. loss per last 100 batches: 0.066318 +[2023-10-07 23:15:15,275][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29131.6/29522=98.68% | mean: 0.01 | max: 5.32 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.14 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] does a screened in porch add value [SEP] ### +### [P_TEXT]: [CLS] 1 adding a screened porch or enclosed porch will add value to your home and adds ### +### value to it. 2 most homeowners are not aware that building an enclosed porch does not only help in ### +### increasing the resale value of any property, but also do wonders to enhancing the curb appeal. it ### +### is very useful especially on wet rainy days or during severe weather conditions. 2 even toddlers ### +### love playing in an enclosed porch. 3 adding a screened porch or enclosed porch will add value to ### +### your home and adds value to it. [SEP] ### +### ======================================= h_v_q | Gates: 26641 ======================================= ### +### ('porch', 0, 0) ('screened', 1, 1) ('value', 2, 4) ('add', 3, 6) ('screening', 4, 167) ### +### ('.', 5, 7845) ('##£', 6, 24901) ('$', 7, 262) ('screen', 8, 136) ('adding', 9, 3) ### +### ('added', 10, 28) ('familiarity', 11, 27996) ('does', 12, 1123) ('relating', 13, 25498) ### +### ('doesn', 14, 71) ('door', 15, 368) ('price', 16, 138) ('deck', 17, 366) ('a', 18, 18356) ### +### ('validity', 19, 10516) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('porch', 0, 0) ('screened', 1, 1) ('enclosed', 4414, 2) ('adding', 9, 3) ('value', 2, 4) ### +### ('ˈ', 328, 5) ('add', 3, 6) ('useful', 269, 7) ('crashing', 114, 8) ('unwilling', 126, 9) ### +### ('hating', 413, 10) ('##ο', 57, 11) ('wonders', 11158, 12) ('appeal', 3428, 13) ('curb', 4876, 14) ### +### ('##ale', 6699, 15) ('sharply', 53, 16) ('−', 247, 17) ('building', 91, 18) ('cyrillic', 611, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('porch', 0, 0) ('screened', 1, 1) ('value', 2, 4) ('add', 3, 6) ('adding', 9, 3) ### +### ('screening', 4, 167) ('added', 10, 28) ('screen', 8, 136) ('$', 7, 262) ('doesn', 14, 71) ### +### ('increase', 23, 61) ('house', 30, 40) ('simon', 26, 55) ('price', 16, 138) ('∈', 28, 67) ### +### ('##ο', 57, 11) ('sharply', 53, 16) ('adds', 33, 105) ('crashing', 114, 8) ('unwilling', 126, 9) ### +############################################################################################################ +[2023-10-07 23:15:15,276][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:15:15,276][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:15:15,696][root][INFO] - Epoch: 12: Step: 301/1557, loss[v]=0.078604, lr=0.000008, acc@1[1]=245.5/256=0.958984375, acc@1[2]=248.5/256=0.970703125 +[2023-10-07 23:16:32,057][root][INFO] - Train batch 400 +[2023-10-07 23:16:32,058][root][INFO] - Avg. loss per last 100 batches: 0.065156 +[2023-10-07 23:16:32,753][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29106.0/29522=98.59% | mean: 0.01 | max: 5.23 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.15 | max: 6.18 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] define lunacy [SEP] ### +### [P_TEXT]: [CLS] lunacy ( countable and uncountable, plural lunacies ) ( 1 of a person or group of ### +### people ) the state of being mad, insanity a cyclical mental disease, apparently linked to the lunar ### +### phases. 2 something deeply misguided. [SEP] ### +### ======================================= h_v_q | Gates: 26165 ======================================= ### +### ('luna', 0, 0) ('##cy', 1, 10) ('definition', 2, 9) ('noun', 3, 21195) ('refers', 4, 6429) ### +### ('relating', 5, 20365) ('defined', 6, 57) ('term', 7, 7188) ('familiarity', 8, 27370) ### +### ('encyclopedia', 9, 11611) ('##º', 10, 28027) ('consisting', 11, 16459) ('something', 12, 634) ### +### ('.', 13, 8256) ('or', 14, 16910) ('stylized', 15, 15343) ('especially', 16, 13297) (';', 17, 3716) ### +### ('plural', 18, 94) ('development', 19, 6639) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('luna', 0, 0) ('insanity', 4658, 1) ('mad', 5722, 2) ('##guide', 25881, 3) ('define', 67, 4) ### +### ('##able', 10600, 5) ('lunar', 1142, 6) ('count', 10624, 7) ('ˈ', 146, 8) ('definition', 2, 9) ### +### ('##cy', 1, 10) ('crashing', 64, 11) ('cyclic', 5281, 12) ('##cies', 3143, 13) ('##ο', 89, 14) ### +### ('synonym', 83, 15) ('##unt', 21941, 16) ('meaning', 24, 17) ('madness', 7662, 18) ### +### ('unwilling', 59, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('luna', 0, 0) ('##cy', 1, 10) ('definition', 2, 9) ('defined', 6, 57) ('definitions', 21, 20) ### +### ('meaning', 24, 17) ('encompasses', 20, 43) ('plural', 18, 94) ('define', 67, 4) ### +### ('crashing', 64, 11) ('person', 48, 28) ('unwilling', 59, 19) ('ˈ', 146, 8) ('simon', 40, 48) ### +### ('sharply', 54, 30) ('−', 66, 21) ('synonym', 83, 15) ('##ο', 89, 14) ('##₂', 77, 26) ### +### ('##α', 49, 52) ### +############################################################################################################ +[2023-10-07 23:16:32,754][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:16:32,754][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:16:33,170][root][INFO] - Epoch: 12: Step: 401/1557, loss[v]=0.050809, lr=0.000008, acc@1[1]=245.0/256=0.95703125, acc@1[2]=250.5/256=0.978515625 +[2023-10-07 23:17:48,900][root][INFO] - Train batch 500 +[2023-10-07 23:17:48,901][root][INFO] - Avg. loss per last 100 batches: 0.064294 +[2023-10-07 23:17:49,579][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29131.4/29522=98.68% | mean: 0.01 | max: 5.17 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.09 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what airport code is iad [SEP] ### +### [P_TEXT]: [CLS] transportation and parking at iad. when you have a flight from washington dulles ### +### intl. ( iad ), take advantage of the various transportation options available. if youave bribed ### +### your friend to drive you the airport, use the convenient curbside drop - off or the short - term ### +### parking at iad. [SEP] ### +### ======================================= h_v_q | Gates: 27547 ======================================= ### +### ('ia', 0, 0) ('airport', 1, 2) ('code', 2, 14226) ('##d', 3, 5) ('codes', 4, 10292) ### +### ('familiarity', 5, 26200) ('relating', 6, 27180) ('is', 7, 21533) ('iowa', 8, 5660) ### +### ('letter', 9, 4524) ('encompasses', 10, 414) ('d', 11, 396) ('refers', 12, 23196) ### +### ('government', 13, 3511) ('plural', 14, 16521) ('consisting', 15, 19511) ('.', 16, 8748) ### +### ('stadium', 17, 857) ('aviation', 18, 177) ('stylized', 19, 25738) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('ia', 0, 0) ('parking', 1253, 1) ('airport', 1, 2) ('dull', 12804, 3) ('washington', 1024, 4) ### +### ('##d', 3, 5) ('##side', 20366, 6) ('transportation', 914, 7) ('ˈ', 145, 8) ('flight', 109, 9) ### +### ('crashing', 185, 10) ('unwilling', 62, 11) ('airports', 166, 12) ('transport', 580, 13) ### +### ('drop', 3116, 14) ('bribe', 28704, 15) ('−', 711, 16) ('##ave', 25697, 17) ('##ο', 132, 18) ### +### ('crashed', 283, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ia', 0, 0) ('airport', 1, 2) ('##d', 3, 5) ('code', 2, 14226) ('unwilling', 62, 11) ### +### ('##大', 39, 34) ('flight', 109, 9) ('airfield', 41, 48) ('aviation', 18, 177) ('simon', 40, 60) ### +### ('ˈ', 145, 8) ('encompasses', 10, 414) ('crashing', 185, 10) ('airports', 166, 12) ('d', 11, 396) ### +### ('ছ', 94, 36) ('##ο', 132, 18) ('aircraft', 24, 199) ('##α', 61, 66) ('sharply', 90, 44) ### +############################################################################################################ +[2023-10-07 23:17:49,579][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:17:49,579][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:17:49,999][root][INFO] - Epoch: 12: Step: 501/1557, loss[v]=0.070081, lr=0.000008, acc@1[1]=247.0/256=0.96484375, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 23:19:05,918][root][INFO] - Train batch 600 +[2023-10-07 23:19:05,919][root][INFO] - Avg. loss per last 100 batches: 0.063097 +[2023-10-07 23:19:06,624][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29157.3/29522=98.76% | mean: 0.01 | max: 5.43 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.16 | max: 6.02 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] define vector function [SEP] ### +### [P_TEXT]: [CLS] vector function. noun, mathematics. 1. a function that assigns a vector to each ### +### point in a given set. [SEP] ### +### ======================================= h_v_q | Gates: 26259 ======================================= ### +### ('vector', 0, 0) ('function', 1, 1) ('definition', 2, 6) ('functions', 3, 3) ('noun', 4, 479) ### +### ('.', 5, 7920) ('refers', 6, 4933) ('relating', 7, 21541) ('something', 8, 4227) ('or', 9, 9728) ### +### ('defined', 10, 55) ('latin', 11, 1048) ('familiarity', 12, 26670) (';', 13, 2260) ### +### ('term', 14, 1652) ('matrix', 15, 73) ('encyclopedia', 16, 11633) ('##º', 17, 27221) ### +### ('plural', 18, 12755) ('vectors', 19, 2) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('vector', 0, 0) ('function', 1, 1) ('vectors', 19, 2) ('functions', 3, 3) ('define', 222, 4) ### +### ('ˈ', 697, 5) ('definition', 2, 6) ('mathematics', 40, 7) ('assigns', 17037, 8) ### +### ('crashing', 155, 9) ('meaning', 32, 10) ('points', 6735, 11) ('mathematical', 184, 12) ### +### ('encompasses', 58, 13) ('point', 53, 14) ('unwilling', 277, 15) ('stumbled', 720, 16) ### +### ('##ο', 195, 17) ('wingspan', 1157, 18) ('−', 796, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('vector', 0, 0) ('function', 1, 1) ('definition', 2, 6) ('functions', 3, 3) ('vectors', 19, 2) ### +### ('defined', 10, 55) ('tensor', 20, 26) ('meaning', 32, 10) ('matrix', 15, 73) ### +### ('mathematics', 40, 7) ('noun', 4, 479) ('definitions', 38, 23) ('point', 53, 14) ### +### ('encompasses', 58, 13) ('means', 59, 28) ('##α', 66, 40) ('define', 222, 4) ('crashing', 155, 9) ### +### ('latin', 11, 1048) ('sharply', 114, 27) ### +############################################################################################################ +[2023-10-07 23:19:06,625][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:19:06,625][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:19:07,048][root][INFO] - Epoch: 12: Step: 601/1557, loss[v]=0.046273, lr=0.000008, acc@1[1]=245.5/256=0.958984375, acc@1[2]=254.0/256=0.9921875 +[2023-10-07 23:20:23,615][root][INFO] - Train batch 700 +[2023-10-07 23:20:23,615][root][INFO] - Avg. loss per last 100 batches: 0.064600 +[2023-10-07 23:20:24,296][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29239.4/29522=99.04% | mean: 0.01 | max: 5.42 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29522.0/29522=100.00% | mean: 0.17 | max: 6.08 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] where are your 5 sinuses located [SEP] ### +### [P_TEXT]: [CLS] the paranasal sinuses. there are four pairs of paranasal sinuses, the frontal ### +### sinuses are located above the eyes, in the forehead bone. the maxillary sinuses ( antra of highmore ### +### ) are located in the cheekbones, under the eyes. the ethmoid sinuses, also called ethmoid labyrinth ### +### are located between the eyes and the nose. here are four pairs of paranasal sinuses, the frontal ### +### sinuses are located above the eyes, in the forehead bone. the maxillary sinuses ( antra of highmore ### +### ) are located in the cheekbones, under the eyes. [SEP] ### +### ======================================= h_v_q | Gates: 28111 ======================================= ### +### ('sin', 0, 4) ('##uses', 1, 6) ('located', 2, 36) ('5', 3, 3111) ('five', 4, 1609) ('where', 5, 64) ### +### ('familiarity', 6, 28160) ('location', 7, 79) ('situated', 8, 117) ('consisting', 9, 23821) ### +### ('your', 10, 18828) ('america', 11, 25128) ('relating', 12, 24318) ('california', 13, 15702) ### +### ('australia', 14, 1572) ('5th', 15, 14587) ('africa', 16, 16042) ('china', 17, 9900) ### +### ('england', 18, 15925) ('stylized', 19, 28103) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('labyrinth', 9121, 0) ('cheekbones', 7626, 1) ('parana', 6165, 2) ('eyes', 6746, 3) ('sin', 0, 4) ### +### ('ˈ', 157, 5) ('##uses', 1, 6) ('forehead', 2045, 7) ('pairs', 8475, 8) ('frontal', 19980, 9) ### +### ('nose', 8414, 10) ('unwilling', 144, 11) ('ant', 13358, 12) ('maxi', 10453, 13) ### +### ('hating', 457, 14) ('crashing', 125, 15) ('gideon', 116, 16) ('##ο', 139, 17) ### +### ('hesitated', 266, 18) ('##ང', 301, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('sin', 0, 4) ('##uses', 1, 6) ('located', 2, 36) ('where', 5, 64) ('location', 7, 79) ### +### ('situated', 8, 117) ('somewhere', 24, 51) ('five', 4, 1609) ('5', 3, 3111) ('locate', 36, 87) ### +### ('are', 21, 168) ('ˈ', 157, 5) ('gideon', 116, 16) ('crashing', 125, 15) ('unwilling', 144, 11) ### +### ('##use', 39, 154) ('locations', 25, 258) ('##ο', 139, 17) ('##α', 97, 55) ('angrily', 113, 33) ### +############################################################################################################ +[2023-10-07 23:20:24,296][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:20:24,296][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:20:24,716][root][INFO] - Epoch: 12: Step: 701/1557, loss[v]=0.072796, lr=0.000008, acc@1[1]=247.5/256=0.966796875, acc@1[2]=253.0/256=0.98828125 +[2023-10-07 23:21:40,775][root][INFO] - Train batch 800 +[2023-10-07 23:21:40,776][root][INFO] - Avg. loss per last 100 batches: 0.065062 +[2023-10-07 23:21:41,480][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29174.9/29522=98.82% | mean: 0.01 | max: 5.62 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.34 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how much does it cost to attend kapiolani community college [SEP] ### +### [P_TEXT]: [CLS] undergraduate tuition. kapiolani community college tuition is $ 2, 736 per year for ### +### in - state residents. this is 16 % cheaper than the national average public two year tuition of $ ### +### 3, 263. the cost is $ 1, 537 and 36 % cheaper than the average hawaii tuition of $ 4, 273 for 2 ### +### year colleges. tuition ranks 2nd in hawaii amongst 2 year colleges for affordability and is the 6th ### +### most expensive 2 year college in the state. if attending from out - of - state, the tuition is $ 7, ### +### 584 which represents a 177 % premium. 2 % of kapiolani community college students received grant ### +### aid in 2013 / 2014. the average total aid amount was $ 4, 276. 31 percent of students received aid ### +### in the form of pell grants from the u. s. federal government. the average pell grant awarded for ### +### 2013 / 2014 was $ 4, 447. [SEP] ### +### ======================================= h_v_q | Gates: 27673 ======================================= ### +### ('$', 0, 12) ('##£', 1, 22863) ('ka', 2, 3) ('##pio', 3, 2) ('##lani', 4, 8) ('college', 5, 13) ### +### ('community', 6, 15) ('cost', 7, 9) ('attend', 8, 6536) ('familiarity', 9, 28314) ### +### ('join', 10, 12307) ('pounds', 11, 1755) ('stylized', 12, 28500) ('.', 13, 7519) ('costs', 14, 19) ### +### ('attended', 15, 3869) ('attending', 16, 92) ('graduate', 17, 299) ('relating', 18, 27230) ### +### ('430', 19, 28447) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('tuition', 3737, 0) ('hawaii', 2581, 1) ('##pio', 3, 2) ('ka', 2, 3) ('ˈ', 91, 4) ### +### ('undergraduate', 83, 5) ('colleges', 144, 6) ('expensive', 53, 7) ('##lani', 4, 8) ('cost', 7, 9) ### +### ('cheaper', 3056, 10) ('amongst', 7859, 11) ('$', 0, 12) ('college', 5, 13) ('grant', 164, 14) ### +### ('community', 6, 15) ('##ο', 36, 16) ('price', 31, 17) ('unwilling', 30, 18) ('costs', 14, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('$', 0, 12) ('ka', 2, 3) ('##pio', 3, 2) ('##lani', 4, 8) ('college', 5, 13) ('community', 6, 15) ### +### ('cost', 7, 9) ('costs', 14, 19) ('##α', 25, 26) ('attending', 16, 92) ('price', 31, 17) ### +### ('unwilling', 30, 18) ('##ο', 36, 16) ('expensive', 53, 7) ('ˈ', 91, 4) ('undergraduate', 83, 5) ### +### ('average', 28, 51) ('ko', 24, 77) ('hating', 57, 20) ('colleges', 144, 6) ### +############################################################################################################ +[2023-10-07 23:21:41,481][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:21:41,481][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:21:41,882][root][INFO] - Epoch: 12: Step: 801/1557, loss[v]=0.055876, lr=0.000008, acc@1[1]=244.5/256=0.955078125, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 23:22:58,246][root][INFO] - Train batch 900 +[2023-10-07 23:22:58,247][root][INFO] - Avg. loss per last 100 batches: 0.063972 +[2023-10-07 23:22:58,936][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29159.3/29522=98.77% | mean: 0.01 | max: 5.54 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.09 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is nurse engagement [SEP] ### +### [P_TEXT]: [CLS] however nurse engagement is defined, it refers to buy -. in, and in our case buy - ### +### in by nurses for the catheter - associated. urinary tract infection ( cauti ) prevention program. ### +### getting nurses to buy - in to any new initiative can be challenging, but especially an initiative ### +### aimed at changing nursing practice. [SEP] ### +### ======================================= h_v_q | Gates: 26280 ======================================= ### +### ('engagement', 0, 0) ('nurse', 1, 1) ('refers', 2, 163) ('definition', 3, 127) ### +### ('relating', 4, 23815) ('is', 5, 3368) ('engaged', 6, 333) ('encompasses', 7, 175) ### +### ('noun', 8, 28542) ('##sam', 9, 23537) ('nurses', 10, 2) ('term', 11, 13102) ('.', 12, 16082) ### +### ('stands', 13, 6801) ('encyclopedia', 14, 12213) ('wedding', 15, 1877) ('nursing', 16, 7) ### +### ('engage', 17, 834) ('consisting', 18, 25333) ('designed', 19, 5109) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('engagement', 0, 0) ('nurse', 1, 1) ('nurses', 10, 2) ('buy', 6958, 3) ('ˈ', 1833, 4) ### +### ('##uti', 27360, 5) ('initiative', 2759, 6) ('nursing', 16, 7) ('programs', 4490, 8) ### +### ('program', 527, 9) ('hating', 321, 10) ('tract', 7170, 11) ('##ο', 369, 12) ### +### ('prevention', 3011, 13) ('crashing', 313, 14) ('unwilling', 588, 15) ('infection', 6301, 16) ### +### ('−', 188, 17) ('##₂', 437, 18) ('ca', 5045, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('engagement', 0, 0) ('nurse', 1, 1) ('nurses', 10, 2) ('nursing', 16, 7) ('refers', 2, 163) ### +### ('definition', 3, 127) ('encompasses', 7, 175) ('engaged', 6, 333) ('defined', 22, 28) ### +### ('is', 5, 3368) ('fernando', 70, 38) ('sharply', 110, 30) ('engage', 17, 834) ('##α', 100, 41) ### +### ('ছ', 137, 31) ('−', 188, 17) ('hated', 186, 23) ('hospital', 38, 241) ('engagements', 25, 565) ### +### ('hating', 321, 10) ### +############################################################################################################ +[2023-10-07 23:22:58,936][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:22:58,937][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:22:59,357][root][INFO] - Epoch: 12: Step: 901/1557, loss[v]=0.077114, lr=0.000008, acc@1[1]=240.5/256=0.939453125, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 23:24:15,257][root][INFO] - Train batch 1000 +[2023-10-07 23:24:15,257][root][INFO] - Avg. loss per last 100 batches: 0.067822 +[2023-10-07 23:24:15,986][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29165.4/29522=98.79% | mean: 0.01 | max: 5.73 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.24 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the difference between skeletal and cardiac muscle [SEP] ### +### [P_TEXT]: [CLS] 2. the skeletal muscles are attached to the bone, and the cardiac muscle is found ### +### in the heart. 3. skeletal muscle cells are cylindrical in shape, whereas the cardiac muscle cells ### +### are semi - spindle in shape. 4. the skeletal muscle cells are longer than the cells of the cardiac ### +### muscle. [SEP] ### +### ======================================= h_v_q | Gates: 27128 ======================================= ### +### ('cardiac', 0, 4) ('skeletal', 1, 2) ('difference', 2, 17639) ('muscle', 3, 1) ('whereas', 4, 6) ### +### ('differences', 5, 13522) ('.', 6, 14540) ('is', 7, 5599) ('heart', 8, 36) ('between', 9, 23844) ### +### ('stands', 10, 8545) ('familiarity', 11, 28036) ('encompasses', 12, 57) ('refers', 13, 23340) ### +### ('consisting', 14, 21540) ('muscles', 15, 0) ('stylized', 16, 27974) ('contrast', 17, 2784) ### +### ('definition', 18, 4209) ('relating', 19, 19063) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('muscles', 15, 0) ('muscle', 3, 1) ('skeletal', 1, 2) ('shape', 6794, 3) ('cardiac', 0, 4) ### +### ('ˈ', 510, 5) ('whereas', 4, 6) ('hesitated', 151, 7) ('unwilling', 67, 8) ('hating', 31, 9) ### +### ('##dle', 19629, 10) ('stumbled', 161, 11) ('attached', 5518, 12) ('##ο', 53, 13) ### +### ('crashing', 306, 14) ('semi', 8017, 15) ('bone', 43, 16) ('gideon', 130, 17) ('##₂', 56, 18) ### +### ('hugh', 44, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('skeletal', 1, 2) ('cardiac', 0, 4) ('muscle', 3, 1) ('whereas', 4, 6) ('muscles', 15, 0) ### +### ('heart', 8, 36) ('encompasses', 12, 57) ('simon', 22, 45) ('hating', 31, 9) ('sharply', 34, 20) ### +### ('bone', 43, 16) ('hugh', 44, 19) ('hated', 40, 27) ('##ο', 53, 13) ('unwilling', 67, 8) ### +### ('##₂', 56, 18) ('ছ', 41, 38) ('##α', 28, 77) ('julian', 21, 124) ('angrily', 66, 24) ### +############################################################################################################ +[2023-10-07 23:24:15,986][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:24:15,986][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:24:16,403][root][INFO] - Epoch: 12: Step: 1001/1557, loss[v]=0.039126, lr=0.000008, acc@1[1]=246.5/256=0.962890625, acc@1[2]=254.5/256=0.994140625 +[2023-10-07 23:25:33,118][root][INFO] - Train batch 1100 +[2023-10-07 23:25:33,119][root][INFO] - Avg. loss per last 100 batches: 0.065856 +[2023-10-07 23:25:33,797][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29121.8/29522=98.64% | mean: 0.01 | max: 5.36 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.15 | max: 6.16 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] cost to install air conditioner [SEP] ### +### [P_TEXT]: [CLS] the cost to install a window air conditioner averages between $ 150 and $ 300, ### +### depending on the size you need. installing a window air conditioning unit can bring added comfort ### +### for a reasonable price, but it will be less powerful than a central air conditioning system. [SEP] ### +### ======================================= h_v_q | Gates: 27124 ======================================= ### +### ('air', 0, 5) ('$', 1, 3) ('condition', 2, 19) ('##£', 3, 15039) ('##er', 4, 40) ('install', 5, 70) ### +### ('cost', 6, 2) ('familiarity', 7, 26547) ('relating', 8, 28374) ('stylized', 9, 28967) ### +### ('.', 10, 12660) ('consisting', 11, 24432) ('costs', 12, 6) ('installed', 13, 154) ('=', 14, 11491) ### +### ('launch', 15, 3097) ('##α', 16, 24) ('plural', 17, 20345) ('fee', 18, 43) ('sharply', 19, 22) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('window', 7883, 0) ('price', 27, 1) ('cost', 6, 2) ('$', 1, 3) ('ˈ', 71, 4) ('air', 0, 5) ### +### ('costs', 12, 6) ('conditioning', 850, 7) ('prices', 130, 8) ('##ο', 59, 9) ('unwilling', 47, 10) ### +### ('hating', 101, 11) ('300', 342, 12) ('depending', 4982, 13) ('powerful', 4250, 14) ### +### ('hesitated', 165, 15) ('crashing', 25, 16) ('stumbled', 194, 17) ('wingspan', 309, 18) ### +### ('condition', 2, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('air', 0, 5) ('$', 1, 3) ('condition', 2, 19) ('cost', 6, 2) ('##er', 4, 40) ('install', 5, 70) ### +### ('costs', 12, 6) ('price', 27, 1) ('##α', 16, 24) ('sharply', 19, 22) ('crashing', 25, 16) ### +### ('fee', 18, 43) ('angrily', 26, 27) ('ˈ', 71, 4) ('unwilling', 47, 10) ('##ο', 59, 9) ### +### ('installing', 21, 55) ('simon', 22, 59) ('installed', 13, 154) ('hating', 101, 11) ### +############################################################################################################ +[2023-10-07 23:25:33,797][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:25:33,797][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:25:34,200][root][INFO] - Epoch: 12: Step: 1101/1557, loss[v]=0.059758, lr=0.000008, acc@1[1]=244.5/256=0.955078125, acc@1[2]=248.5/256=0.970703125 +[2023-10-07 23:26:50,359][root][INFO] - Train batch 1200 +[2023-10-07 23:26:50,359][root][INFO] - Avg. loss per last 100 batches: 0.063785 +[2023-10-07 23:26:51,061][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29139.2/29522=98.70% | mean: 0.01 | max: 5.57 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.55 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is picea [SEP] ### +### [P_TEXT]: [CLS] definition of picea. : a genus of temperate and arctic evergreen trees ( family ### +### pinaceae ) having acicular leaves that are keeled on both surfaces and borne individually on ### +### persistent peg - shaped bases and cones that become pendulous and have reflexed scales a see ### +### spruce. ew latin, from latin, pitch pine, from pic -, pix pitch. this word doesn't usually appear ### +### in our free dictionary, but the definition from our premium unabridged dictionary is offered here ### +### on a limited basis. note that some information is displayed differently in the unabridged. [SEP] ### +### ======================================= h_v_q | Gates: 27311 ======================================= ### +### ('pic', 0, 1) ('##ea', 1, 16) ('is', 2, 304) ('encompasses', 3, 28) ('##sam', 4, 25631) ### +### ('definition', 5, 18) ('relating', 6, 25594) ('familiarity', 7, 28292) ('consisting', 8, 22165) ### +### ('refers', 9, 8821) ('stylized', 10, 27256) ('noun', 11, 21237) ('stands', 12, 4395) ### +### ('plural', 13, 7697) ('provides', 14, 6649) ('encyclopedia', 15, 2734) ('an', 16, 11255) ### +### ('language', 17, 6716) ('genus', 18, 29) ('a', 19, 14786) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##aceae', 6022, 0) ('pic', 0, 1) ('pin', 33, 2) ('peg', 18161, 3) ('latin', 27, 4) ### +### ('evergreen', 12639, 5) ('scales', 5092, 6) ('spruce', 24050, 7) ('definitions', 626, 8) ### +### ('keel', 13589, 9) ('trees', 8015, 10) ('cones', 19458, 11) ('pitch', 3985, 12) ### +### ('arctic', 6895, 13) ('pi', 140, 14) ('pine', 8548, 15) ('##ea', 1, 16) ('bases', 10241, 17) ### +### ('definition', 5, 18) ('temperate', 6309, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('pic', 0, 1) ('##ea', 1, 16) ('encompasses', 3, 28) ('definition', 5, 18) ('is', 2, 304) ### +### ('latin', 27, 4) ('genus', 18, 29) ('pin', 33, 2) ('defined', 20, 66) ('sharply', 42, 62) ### +### ('pi', 140, 14) ('crashing', 78, 32) ('simon', 41, 86) ('angrily', 62, 56) ('prehistoric', 101, 38) ### +### ('ˈ', 214, 24) ('ছ', 83, 53) ('hating', 91, 50) ('##α', 53, 94) ('julian', 40, 133) ### +############################################################################################################ +[2023-10-07 23:26:51,062][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:26:51,062][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:26:51,464][root][INFO] - Epoch: 12: Step: 1201/1557, loss[v]=0.061560, lr=0.000008, acc@1[1]=244.0/256=0.953125, acc@1[2]=254.5/256=0.994140625 +[2023-10-07 23:28:07,474][root][INFO] - Train batch 1300 +[2023-10-07 23:28:07,475][root][INFO] - Avg. loss per last 100 batches: 0.064500 +[2023-10-07 23:28:08,188][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29204.6/29522=98.93% | mean: 0.01 | max: 5.68 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.17 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] age limit for child exam icd coding [SEP] ### +### [P_TEXT]: [CLS] answer : youare correct that you would typically use v20. 2 ( health supervision of ### +### infant or child ; routine infant or child health check ) for a preventive medicine exam of a 17 - ### +### year - old. in fact, the designation that appears in the icd - 9 manual next to v20. 2 supports ### +### using the diagnosis code when the patientas age is between 0 and 17 years. [SEP] ### +### ======================================= h_v_q | Gates: 27561 ======================================= ### +### ('age', 0, 0) ('coding', 1, 418) ('ic', 2, 16) ('child', 3, 18) ('limit', 4, 16127) ('exam', 5, 9) ### +### ('##d', 6, 63) ('familiarity', 7, 27119) ('years', 8, 197) ('.', 9, 17306) ('ages', 10, 401) ### +### ('18', 11, 135) ('relating', 12, 27406) ('examination', 13, 213) ('children', 14, 46) ### +### ('16', 15, 75) ('21', 16, 76) ('old', 17, 4) ('800', 18, 12632) ('older', 19, 48) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('age', 0, 0) ('supervision', 5572, 1) ('infant', 2742, 2) ('medicine', 2030, 3) ('old', 17, 4) ### +### ('v', 3395, 5) ('ˈ', 160, 6) ('##20', 18982, 7) ('manual', 1198, 8) ('exam', 5, 9) ('20', 39, 10) ### +### ('crashing', 99, 11) ('routine', 6424, 12) ('##ο', 55, 13) ('check', 3146, 14) ### +### ('unwilling', 103, 15) ('ic', 2, 16) ('2', 5671, 17) ('child', 3, 18) ('diagnosis', 15364, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('age', 0, 0) ('ic', 2, 16) ('child', 3, 18) ('exam', 5, 9) ('##d', 6, 63) ('coding', 1, 418) ### +### ('old', 17, 4) ('children', 14, 46) ('years', 8, 197) ('16', 15, 75) ('18', 11, 135) ('21', 16, 76) ### +### ('older', 19, 48) ('20', 39, 10) ('17', 25, 37) ('examination', 13, 213) ('ages', 10, 401) ### +### ('##α', 36, 39) ('##ο', 55, 13) ('angrily', 43, 31) ### +############################################################################################################ +[2023-10-07 23:28:08,188][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:28:08,188][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:28:08,612][root][INFO] - Epoch: 12: Step: 1301/1557, loss[v]=0.091118, lr=0.000008, acc@1[1]=244.0/256=0.953125, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 23:29:24,997][root][INFO] - Train batch 1400 +[2023-10-07 23:29:24,997][root][INFO] - Avg. loss per last 100 batches: 0.066484 +[2023-10-07 23:29:25,661][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29204.7/29522=98.93% | mean: 0.01 | max: 5.63 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.14 | max: 6.22 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] age to work in chipotle [SEP] ### +### [P_TEXT]: [CLS] report abuse. check out here http : / / www. glassdoor. com / hourly - pay / ### +### chip... minimum age to work at chipotle : 16 years old ( how old do you have to be to work at ### +### chipotle? ) http : / / www. job - applications. com / chipotle... [SEP] ### +### ======================================= h_v_q | Gates: 27872 ======================================= ### +### ('chip', 0, 1) ('age', 1, 3) ('##le', 2, 104) ('work', 3, 8) ('##ot', 4, 79) ### +### ('familiarity', 5, 24139) ('chips', 6, 24) ('years', 7, 93) ('ages', 8, 1616) ### +### ('stylized', 9, 28702) ('18', 10, 455) ('16', 11, 22) ('relating', 12, 25594) ('working', 13, 55) ### +### ('older', 14, 35) ('old', 15, 9) ('21', 16, 519) ('consisting', 17, 21345) ('.', 18, 7565) ### +### ('##α', 19, 21) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('abuse', 14771, 0) ('chip', 0, 1) ('ˈ', 86, 2) ('age', 1, 3) ('unwilling', 118, 4) ('##ο', 45, 5) ### +### ('crashing', 36, 6) ('##ང', 87, 7) ('work', 3, 8) ('old', 15, 9) ('hating', 161, 10) ### +### ('minimum', 3563, 11) ('ছ', 65, 12) ('−', 59, 13) ('hesitated', 153, 14) ('##大', 120, 15) ### +### ('cyrillic', 193, 16) ('angrily', 46, 17) ('wingspan', 155, 18) ('stumbled', 136, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('chip', 0, 1) ('age', 1, 3) ('work', 3, 8) ('##le', 2, 104) ('##ot', 4, 79) ('chips', 6, 24) ### +### ('16', 11, 22) ('old', 15, 9) ('years', 7, 93) ('older', 14, 35) ('working', 13, 55) ### +### ('##α', 19, 21) ('crashing', 36, 6) ('##ο', 45, 5) ('sharply', 32, 27) ('ˈ', 86, 2) ### +### ('crashed', 35, 29) ('angrily', 46, 17) ('−', 59, 13) ('##ང', 87, 7) ### +############################################################################################################ +[2023-10-07 23:29:25,662][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:29:25,662][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:29:26,081][root][INFO] - Epoch: 12: Step: 1401/1557, loss[v]=0.066096, lr=0.000007, acc@1[1]=244.0/256=0.953125, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 23:30:42,557][root][INFO] - Train batch 1500 +[2023-10-07 23:30:42,558][root][INFO] - Avg. loss per last 100 batches: 0.066343 +[2023-10-07 23:30:43,232][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29073.6/29522=98.48% | mean: 0.01 | max: 5.30 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.25 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] is noscapine narcotic or not [SEP] ### +### [P_TEXT]: [CLS] noscapine is a benzylisoquinoline alkaloid produced in opium poppy ( papaver ### +### somniferum ) and other members of the papaveraceae. it has been used as a cough suppressant and ### +### more recently was shown to possess anticancer activity. [SEP] ### +### ======================================= h_v_q | Gates: 28221 ======================================= ### +### ('nos', 0, 0) ('##cap', 1, 5) ('na', 2, 6882) ('##ine', 3, 31) ('not', 4, 10647) ### +### ('##tic', 5, 22624) ('is', 6, 450) ('familiarity', 7, 28865) ('never', 8, 10686) ### +### ('##rco', 9, 28833) ('cannot', 10, 5516) ('.', 11, 20469) ('nobody', 12, 12577) ### +### ('neither', 13, 5487) ('nothing', 14, 1099) ('insufficient', 15, 12591) ('stylized', 16, 26456) ### +### ('relating', 17, 25462) ('non', 18, 15542) ('consisting', 19, 25532) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('nos', 0, 0) ('opium', 15764, 1) ('papa', 7813, 2) ('poppy', 13442, 3) ('cough', 18416, 4) ### +### ('##cap', 1, 5) ('##ceae', 3221, 6) ('encompasses', 48, 7) ('ˈ', 147, 8) ('##olin', 28451, 9) ### +### ('benz', 16474, 10) ('suppress', 352, 11) ('##vera', 20875, 12) ('−', 45, 13) ### +### ('suppressed', 62, 14) ('stumbled', 182, 15) ('hating', 73, 16) ('unwilling', 43, 17) ### +### ('##ο', 37, 18) ('al', 8372, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('nos', 0, 0) ('##cap', 1, 5) ('##ine', 3, 31) ('sharply', 25, 21) ('is', 6, 450) ### +### ('angrily', 28, 32) ('##α', 26, 44) ('encompasses', 48, 7) ('##ο', 37, 18) ('simon', 29, 49) ### +### ('na', 2, 6882) ('unwilling', 43, 17) ('−', 45, 13) ('suppressed', 62, 14) ('anton', 32, 67) ### +### ('##₂', 52, 30) ('hating', 73, 16) ('hated', 51, 35) ('hesitated', 68, 27) ('crashing', 75, 22) ### +############################################################################################################ +[2023-10-07 23:30:43,233][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:30:43,233][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:30:43,651][root][INFO] - Epoch: 12: Step: 1501/1557, loss[v]=0.051264, lr=0.000007, acc@1[1]=242.0/256=0.9453125, acc@1[2]=254.0/256=0.9921875 +[2023-10-07 23:31:26,525][root][INFO] - rank=2; last iteration 1557 +[2023-10-07 23:31:26,526][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 23:31:26,526][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-07 23:31:26,531][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 23:31:26,531][root][INFO] - Epoch finished on 2 +[2023-10-07 23:31:26,546][root][INFO] - rank=1; last iteration 1557 +[2023-10-07 23:31:26,547][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 23:31:26,547][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-07 23:31:26,550][root][INFO] - rank=3; last iteration 1557 +[2023-10-07 23:31:26,551][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 23:31:26,551][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-07 23:31:26,552][root][INFO] - rank=0; last iteration 1557 +[2023-10-07 23:31:26,553][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 23:31:26,553][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-07 23:31:26,554][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 23:31:26,554][root][INFO] - Epoch finished on 1 +[2023-10-07 23:31:26,560][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 23:31:26,560][root][INFO] - Epoch finished on 3 +[2023-10-07 23:31:26,560][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 23:31:26,561][root][INFO] - Epoch finished on 0 +[2023-10-07 23:31:42,971][root][INFO] - Saved checkpoint at ./vdr_12 +[2023-10-07 23:31:42,971][root][INFO] - Saved checkpoint at ./vdr_12 +[2023-10-07 23:31:42,971][root][INFO] - Saved checkpoint at ./vdr_12 +[2023-10-07 23:31:42,972][root][INFO] - Av Loss per epoch=0.065026 +[2023-10-07 23:31:42,972][root][INFO] - Av Loss per epoch=0.065026 +[2023-10-07 23:31:42,972][root][INFO] - Av Loss per epoch=0.065026 +[2023-10-07 23:31:42,972][root][INFO] - epoch total (1) correct predictions=378952 +[2023-10-07 23:31:42,972][root][INFO] - epoch total (1) correct predictions=378952 +[2023-10-07 23:31:42,972][root][INFO] - epoch total (1) correct predictions=378952 +[2023-10-07 23:31:42,972][root][INFO] - epoch total (2) correct predictions=391096 +[2023-10-07 23:31:42,972][root][INFO] - epoch total (2) correct predictions=391096 +[2023-10-07 23:31:42,972][root][INFO] - epoch total (2) correct predictions=391096 +[2023-10-07 23:31:42,974][root][INFO] - Saved checkpoint at ./vdr_12 +[2023-10-07 23:31:42,974][root][INFO] - Av Loss per epoch=0.065026 +[2023-10-07 23:31:42,975][root][INFO] - epoch total (1) correct predictions=378952 +[2023-10-07 23:31:42,975][root][INFO] - epoch total (2) correct predictions=391096 +[2023-10-07 23:31:42,976][root][INFO] - ***** Epoch 13 ***** +[2023-10-07 23:31:42,976][root][INFO] - ***** Epoch 13 ***** +[2023-10-07 23:31:42,978][root][INFO] - ***** Epoch 13 ***** +[2023-10-07 23:31:42,982][root][INFO] - rank=1; Iteration start +[2023-10-07 23:31:42,980][root][INFO] - ***** Epoch 13 ***** +[2023-10-07 23:31:42,983][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 23:31:42,983][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 23:31:42,985][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-07 23:31:42,985][root][INFO] - rank=3; Iteration start +[2023-10-07 23:31:42,985][root][INFO] - rank=0; Iteration start +[2023-10-07 23:31:42,985][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 23:31:42,985][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 23:31:42,985][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 23:31:42,985][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 23:31:42,987][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-07 23:31:42,987][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-07 23:31:42,988][root][INFO] - rank=2; Iteration start +[2023-10-07 23:31:42,988][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 23:31:42,988][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 23:31:42,990][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-07 23:31:44,020][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29117.8/29522=98.63% | mean: 0.01 | max: 5.23 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.04 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what four states border missouri [SEP] ### +### [P_TEXT]: [CLS] missouri's bordering states. missouri is surrounded by eight different states. ### +### starting north of missouri and going clockwise, missouri is bordered by iowa, illinois, kentucky, ### +### tennessee, arkansas, oklahoma, kansas, and nebraska. click on the following link to find missouri ### +### and the eight states that border it. [SEP] ### +### ======================================= h_v_q | Gates: 27467 ======================================= ### +### ('missouri', 0, 0) ('border', 1, 8) ('four', 2, 165) ('states', 3, 2) ('borders', 4, 28) ### +### ('4', 5, 6718) ('familiarity', 6, 23612) ('.', 7, 4904) ('six', 8, 349) ('frontier', 9, 176) ### +### ('hampshire', 10, 14323) ('america', 11, 19193) ('pennsylvania', 12, 2121) ('boundary', 13, 80) ### +### ('stylized', 14, 28043) ('4th', 15, 4986) ('ohio', 16, 2031) ('1790', 17, 14012) ### +### ('coastal', 18, 5316) ('relating', 19, 23952) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('missouri', 0, 0) ('surrounded', 1318, 1) ('states', 3, 2) ('clockwise', 28209, 3) ('ˈ', 729, 4) ### +### ('bordering', 331, 5) ('arkansas', 90, 6) ('bordered', 52, 7) ('border', 1, 8) ('kentucky', 234, 9) ### +### ('tennessee', 40, 10) ('oklahoma', 718, 11) ('kansas', 65, 12) ('state', 26, 13) ('iowa', 220, 14) ### +### ('crashing', 566, 15) ('unwilling', 236, 16) ('stumbled', 655, 17) ('##ο', 478, 18) ### +### ('encompasses', 1168, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('missouri', 0, 0) ('border', 1, 8) ('states', 3, 2) ('borders', 4, 28) ('four', 2, 165) ### +### ('state', 26, 13) ('bordered', 52, 7) ('tennessee', 40, 10) ('eight', 32, 35) ('arkansas', 90, 6) ### +### ('boundary', 13, 80) ('kansas', 65, 12) ('illinois', 55, 26) ('frontier', 9, 176) ### +### ('numerous', 45, 51) ('six', 8, 349) ('kentucky', 234, 9) ('mo', 81, 48) ('nebraska', 142, 22) ### +### ('bordering', 331, 5) ### +############################################################################################################ +[2023-10-07 23:31:44,020][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:31:44,020][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:31:44,419][root][INFO] - Epoch: 13: Step: 1/1557, loss[v]=0.084853, lr=0.000007, acc@1[1]=234.5/256=0.916015625, acc@1[2]=247.5/256=0.966796875 +[2023-10-07 23:33:00,704][root][INFO] - Train batch 100 +[2023-10-07 23:33:00,705][root][INFO] - Avg. loss per last 100 batches: 0.062377 +[2023-10-07 23:33:01,382][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29125.7/29522=98.66% | mean: 0.01 | max: 5.14 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.9/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.20 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] apple stock price per share [SEP] ### +### [P_TEXT]: [CLS] apple stock has soared 123 percent from the launch of the ipod in 2001 to the most ### +### recent period. prior to the end of 2001, apple stock was stuck trading around $ 1. 50 per share ( ### +### except for a short blip in 2000 when shares rose to $ 5. 00 before falling again ). it took about ### +### two and half years after the ipod launch for the stock to start its spectacular upward ascent. ### +### [SEP] ### +### ======================================= h_v_q | Gates: 26302 ======================================= ### +### ('apple', 0, 1) ('share', 1, 34) ('stock', 2, 4) ('$', 3, 11) ('price', 4, 13) ('##£', 5, 23544) ### +### ('per', 6, 117) ('shares', 7, 47) ('familiarity', 8, 27487) ('prices', 9, 28) ('.', 10, 10557) ### +### ('stylized', 11, 28713) ('sharing', 12, 1932) ('apples', 13, 53) ('relating', 14, 28165) ### +### ('plural', 15, 16543) ('shared', 16, 1718) ('=', 17, 13931) ('cost', 18, 71) ('800', 19, 12005) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('ipod', 14400, 0) ('apple', 0, 1) ('stuck', 10663, 2) ('ˈ', 350, 3) ('stock', 2, 4) ### +### ('percent', 28, 5) ('recent', 17586, 6) ('launch', 998, 7) ('stocks', 24, 8) ('trading', 2172, 9) ### +### ('##ο', 110, 10) ('$', 3, 11) ('except', 5857, 12) ('price', 4, 13) ('soared', 28778, 14) ### +### ('stumbled', 326, 15) ('##lip', 21939, 16) ('recently', 21455, 17) ('upward', 10852, 18) ### +### ('percentage', 971, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('apple', 0, 1) ('share', 1, 34) ('stock', 2, 4) ('$', 3, 11) ('price', 4, 13) ('per', 6, 117) ### +### ('shares', 7, 47) ('prices', 9, 28) ('apples', 13, 53) ('stocks', 24, 8) ('percent', 28, 5) ### +### ('cost', 18, 71) ('##α', 25, 35) ('##ο', 110, 10) ('sharply', 55, 33) ('simon', 41, 52) ### +### ('angrily', 61, 31) ('crashing', 71, 26) ('hated', 79, 21) ('ˈ', 350, 3) ### +############################################################################################################ +[2023-10-07 23:33:01,383][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:33:01,383][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:33:01,806][root][INFO] - Epoch: 13: Step: 101/1557, loss[v]=0.071305, lr=0.000007, acc@1[1]=243.0/256=0.94921875, acc@1[2]=250.0/256=0.9765625 +[2023-10-07 23:34:18,779][root][INFO] - Train batch 200 +[2023-10-07 23:34:18,780][root][INFO] - Avg. loss per last 100 batches: 0.063986 +[2023-10-07 23:34:19,485][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29153.8/29522=98.75% | mean: 0.01 | max: 5.46 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.30 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how long should i expect a dishwasher to last [SEP] ### +### [P_TEXT]: [CLS] the typical dishwasher can be expected to last between seven and 12 years, with the ### +### average working expectancy being nine to 10 years. certain factors, such as cost, quality, brand, ### +### care, maintenance and frequency of usage, can affect the life expectancy of a dishwasher. hile your ### +### particular model may last for a longer or shorter duration than the average expectancy, it is good ### +### to have an idea of the average time you can expect your appliance to last. this will give you an ### +### idea of when the dishwasher will need to be replaced and help you to budget accordingly. [SEP] ### +### ======================================= h_v_q | Gates: 27709 ======================================= ### +### ('dish', 0, 4) ('##wash', 1, 3) ('expect', 2, 46) ('last', 3, 70) ('##er', 4, 34) ('days', 5, 1737) ### +### ('minutes', 6, 32) ('years', 7, 63) ('hours', 8, 15183) ('weeks', 9, 42) ('familiarity', 10, 27035) ### +### ('should', 11, 21003) ('expected', 12, 7) ('lasts', 13, 254) ('months', 14, 2210) ('long', 15, 2) ### +### ('.', 16, 15179) ('и', 17, 385) ('expecting', 18, 18) ('length', 19, 457) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('duration', 44, 0) ('ˈ', 407, 1) ('long', 15, 2) ('##wash', 1, 3) ('dish', 0, 4) ### +### ('hating', 129, 5) ('hi', 7713, 6) ('expected', 12, 7) ('model', 1197, 8) ('life', 379, 9) ### +### ('crashing', 69, 10) ('cyrillic', 646, 11) ('unwilling', 63, 12) ('app', 9544, 13) ('##ο', 121, 14) ### +### ('##₂', 211, 15) ('typical', 234, 16) ('hated', 117, 17) ('expecting', 18, 18) ('dishes', 31, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##wash', 1, 3) ('dish', 0, 4) ('expect', 2, 46) ('##er', 4, 34) ('last', 3, 70) ### +### ('minutes', 6, 32) ('expected', 12, 7) ('years', 7, 63) ('long', 15, 2) ('weeks', 9, 42) ### +### ('expecting', 18, 18) ('time', 20, 38) ('30', 22, 41) ('lasts', 13, 254) ('duration', 44, 0) ### +### ('dishes', 31, 19) ('days', 5, 1737) ('longer', 37, 36) ('и', 17, 385) ('hoped', 25, 80) ### +############################################################################################################ +[2023-10-07 23:34:19,485][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:34:19,485][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:34:19,906][root][INFO] - Epoch: 13: Step: 201/1557, loss[v]=0.043310, lr=0.000007, acc@1[1]=243.5/256=0.951171875, acc@1[2]=254.0/256=0.9921875 +[2023-10-07 23:35:36,364][root][INFO] - Train batch 300 +[2023-10-07 23:35:36,365][root][INFO] - Avg. loss per last 100 batches: 0.064729 +[2023-10-07 23:35:37,104][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29165.8/29522=98.79% | mean: 0.01 | max: 5.05 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.26 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] yumi definition [SEP] ### +### [P_TEXT]: [CLS] is the japanese term for a bow. as used in english, yumi refers more specifically ### +### to traditional japanese asymmetrical bows, and includes the longer and the shorter used in the ### +### practice of kyudo and kyujutsu, or japanese archery. the yumi was an important weapon of the ### +### samurai warrior during the feudal period of japan. see more at wikipedia. org... this is a list of ### +### utilities for creating live usb. [SEP] ### +### ======================================= h_v_q | Gates: 26467 ======================================= ### +### ('yu', 0, 0) ('##mi', 1, 3) ('definition', 2, 48) ('noun', 3, 19262) ('familiarity', 4, 26713) ### +### ('relating', 5, 23258) ('something', 6, 7014) ('defined', 7, 291) ('term', 8, 66) ### +### ('plural', 9, 16371) ('or', 10, 16252) ('latin', 11, 6191) ('stylized', 12, 25614) ### +### ('encyclopedia', 13, 10447) (';', 14, 5507) ('refers', 15, 278) ('##yu', 16, 516) ### +### ('sense', 17, 9697) ('consisting', 18, 25569) ('meaning', 19, 89) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('yu', 0, 0) ('bow', 8515, 1) ('archery', 14466, 2) ('##mi', 1, 3) ('samurai', 8255, 4) ### +### ('japanese', 372, 5) ('bows', 29128, 6) ('usb', 27132, 7) ('feudal', 2134, 8) ('japan', 71, 9) ### +### ('ˈ', 340, 10) ('weapon', 253, 11) ('practice', 1393, 12) ('crashing', 284, 13) ('##ο', 285, 14) ### +### ('ky', 7810, 15) ('encompasses', 250, 16) ('stumbled', 225, 17) ('warrior', 915, 18) ### +### ('hating', 432, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('yu', 0, 0) ('##mi', 1, 3) ('definition', 2, 48) ('term', 8, 66) ('defined', 7, 291) ### +### ('definitions', 24, 44) ('meaning', 19, 89) ('japan', 71, 9) ('ya', 28, 104) ('means', 31, 129) ### +### ('refers', 15, 278) ('sharply', 73, 27) ('julian', 50, 64) ('##α', 69, 45) ('angrily', 68, 51) ### +### ('japanese', 372, 5) ('weapon', 253, 11) ('ˈ', 340, 10) ('ছ', 152, 28) ('encompasses', 250, 16) ### +############################################################################################################ +[2023-10-07 23:35:37,104][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:35:37,104][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:35:37,533][root][INFO] - Epoch: 13: Step: 301/1557, loss[v]=0.041842, lr=0.000007, acc@1[1]=243.5/256=0.951171875, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 23:36:53,900][root][INFO] - Train batch 400 +[2023-10-07 23:36:53,901][root][INFO] - Avg. loss per last 100 batches: 0.061876 +[2023-10-07 23:36:54,601][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29248.3/29522=99.07% | mean: 0.01 | max: 5.55 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.16 | max: 6.38 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] fix skype for business certifacate error [SEP] ### +### [P_TEXT]: [CLS] 1 delete the certificate and try to sign in to skype for business online. 2 if you ### +### can't sign in to skype for business online, go to step 2. 3 if you're running windows 7, remove the ### +### useras stored credentials in windows credential manager. 4 to do this, follow these steps : 5 open ### +### control panel, and then click credential manager. dditionally, when you try to sign in to lync ### +### after a network outage or a skype for business online service outage, you receive the following ### +### error message : there was a problem acquiring a personal certificate required to sign in. if the ### +### problem continues, please contact your support team. cannot sign into lync. [SEP] ### +### ======================================= h_v_q | Gates: 28522 ======================================= ### +### ('sky', 0, 7) ('error', 1, 23) ('##pe', 2, 3) ('fix', 3, 2240) ('##rti', 4, 28209) ### +### ('business', 5, 16) ('ce', 6, 15237) ('##fa', 7, 18511) ('##cate', 8, 23005) ### +### ('familiarity', 9, 27189) ('.', 10, 15737) ('stylized', 11, 28464) ('##α', 12, 35) ### +### ('for', 13, 2701) ('relating', 14, 24965) ('simon', 15, 54) ('consisting', 16, 24488) ### +### ('crashed', 17, 36) ('fixing', 18, 9264) ('##ο', 19, 8) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('credentials', 19564, 0) ('online', 5657, 1) ('certificate', 11201, 2) ('##pe', 2, 3) ('ˈ', 60, 4) ### +### ('unwilling', 32, 5) ('crashing', 22, 6) ('sky', 0, 7) ('##ο', 19, 8) ('dd', 22758, 9) ### +### ('hating', 40, 10) ('signing', 8878, 11) ('manager', 963, 12) ('cyrillic', 125, 13) ### +### ('remove', 668, 14) ('sign', 12175, 15) ('business', 5, 16) ('sharply', 35, 17) ('##ང', 108, 18) ### +### ('##大', 80, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('sky', 0, 7) ('##pe', 2, 3) ('error', 1, 23) ('business', 5, 16) ('##α', 12, 35) ('fix', 3, 2240) ### +### ('##ο', 19, 8) ('crashing', 22, 6) ('crashed', 17, 36) ('simon', 15, 54) ('unwilling', 32, 5) ### +### ('##₂', 21, 32) ('angrily', 29, 24) ('sharply', 35, 17) ('hating', 40, 10) ('−', 34, 22) ### +### ('⟩', 23, 72) ('ˈ', 60, 4) ('hesitated', 43, 21) ('gideon', 42, 38) ### +############################################################################################################ +[2023-10-07 23:36:54,602][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:36:54,602][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:36:55,007][root][INFO] - Epoch: 13: Step: 401/1557, loss[v]=0.065072, lr=0.000007, acc@1[1]=241.5/256=0.943359375, acc@1[2]=252.0/256=0.984375 +[2023-10-07 23:38:12,132][root][INFO] - Train batch 500 +[2023-10-07 23:38:12,133][root][INFO] - Avg. loss per last 100 batches: 0.061783 +[2023-10-07 23:38:12,818][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29101.8/29522=98.58% | mean: 0.01 | max: 5.52 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.16 | max: 6.20 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] do bylaws need to be signed [SEP] ### +### [P_TEXT]: [CLS] 0 attorneys agreed. re : corporation bylaws. by - laws do not need to be prepared ### +### or signed by an attonrey. most pre - printed by laws are adequate. it naturally depends on the type ### +### of business you conduct. please contact my office at 714 363 0220 to set up an appointment. i would ### +### like to review your written documentation including any letters. [SEP] ### +### ======================================= h_v_q | Gates: 27746 ======================================= ### +### ('##law', 0, 1) ('signed', 1, 12) ('by', 2, 56) ('##s', 3, 241) ('signing', 4, 52) ('need', 5, 60) ### +### ('sign', 6, 151) ('do', 7, 499) ('law', 8, 13) ('familiarity', 9, 23951) ('must', 10, 452) ### +### ('be', 11, 219) ('required', 12, 142) ('.', 13, 15000) ('requires', 14, 99) ('stylized', 15, 27611) ### +### ('signature', 16, 917) ('being', 17, 2206) ('relating', 18, 26123) ('needs', 19, 272) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##rey', 1284, 0) ('##law', 0, 1) ('appointment', 332, 2) ('laws', 41, 3) ('printed', 5181, 4) ### +### ('corporation', 609, 5) ('adequate', 54, 6) ('appointments', 5068, 7) ('ˈ', 451, 8) ### +### ('depends', 4416, 9) ('pre', 6425, 10) ('naturally', 6811, 11) ('signed', 1, 12) ('law', 8, 13) ### +### ('documentation', 962, 14) ('##ton', 193, 15) ('##ο', 103, 16) ('prepare', 4040, 17) ### +### ('depending', 18153, 18) ('lawyers', 6547, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##law', 0, 1) ('signed', 1, 12) ('law', 8, 13) ('by', 2, 56) ('signing', 4, 52) ('need', 5, 60) ### +### ('##s', 3, 241) ('sign', 6, 151) ('do', 7, 499) ('required', 12, 142) ('requires', 14, 99) ### +### ('be', 11, 219) ('laws', 41, 3) ('must', 10, 452) ('requirements', 29, 57) ('written', 23, 81) ### +### ('needed', 27, 84) ('adequate', 54, 6) ('seek', 20, 159) ('##α', 35, 43) ### +############################################################################################################ +[2023-10-07 23:38:12,819][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:38:12,819][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:38:13,242][root][INFO] - Epoch: 13: Step: 501/1557, loss[v]=0.052897, lr=0.000007, acc@1[1]=246.0/256=0.9609375, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 23:39:29,492][root][INFO] - Train batch 600 +[2023-10-07 23:39:29,493][root][INFO] - Avg. loss per last 100 batches: 0.061105 +[2023-10-07 23:39:40,304][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29175.7/29522=98.83% | mean: 0.01 | max: 5.29 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.16 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what county is baldwin city ks [SEP] ### +### [P_TEXT]: [CLS] aerial view of baldwin city. baldwin city is a city in douglas county, kansas, ### +### united states about 12 miles ( 19 km ) south of lawrence and 15 miles ( 24 km ) west of gardner. as ### +### of the 2010 census, the city population was 4, 515. the city is home to baker university, the ### +### oldest four - year university in the state. [SEP] ### +### ======================================= h_v_q | Gates: 26766 ======================================= ### +### ('baldwin', 0, 0) ('county', 1, 16) ('city', 2, 5) ('kansas', 3, 4) ('familiarity', 4, 25270) ### +### ('missouri', 5, 251) ('.', 6, 3842) ('##sam', 7, 26995) ('town', 8, 30) ('stylized', 9, 27380) ### +### ('minnesota', 10, 2547) ('relating', 11, 25836) ('pan', 12, 367) ('urban', 13, 61) ### +### ('consisting', 14, 22824) ('colorado', 15, 2914) ('is', 16, 308) ('nebraska', 17, 3360) ### +### ('group', 18, 2489) ('iowa', 19, 2824) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('baldwin', 0, 0) ('baker', 228, 1) ('gardner', 7572, 2) ('douglas', 262, 3) ('kansas', 3, 4) ### +### ('city', 2, 5) ('lawrence', 1403, 6) ('ˈ', 1784, 7) ('aerial', 9991, 8) ('crashing', 356, 9) ### +### ('cities', 33, 10) ('##ο', 656, 11) ('oldest', 7813, 12) ('where', 1215, 13) ('view', 1977, 14) ### +### ('stumbled', 902, 15) ('county', 1, 16) ('encompasses', 20, 17) ('unwilling', 288, 18) ### +### ('population', 780, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('baldwin', 0, 0) ('city', 2, 5) ('county', 1, 16) ('kansas', 3, 4) ('town', 8, 30) ### +### ('encompasses', 20, 17) ('missouri', 5, 251) ('urban', 13, 61) ('cities', 33, 10) ### +### ('counties', 31, 22) ('ks', 27, 36) ('baker', 228, 1) ('douglas', 262, 3) ('downtown', 30, 91) ### +### ('states', 25, 103) ('state', 35, 80) ('university', 100, 37) ('simon', 115, 34) ### +### ('angrily', 95, 42) ('pan', 12, 367) ### +############################################################################################################ +[2023-10-07 23:39:40,304][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:39:40,304][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:39:40,718][root][INFO] - Epoch: 13: Step: 601/1557, loss[v]=0.063801, lr=0.000007, acc@1[1]=246.5/256=0.962890625, acc@1[2]=252.0/256=0.984375 +[2023-10-07 23:40:57,358][root][INFO] - Train batch 700 +[2023-10-07 23:40:57,359][root][INFO] - Avg. loss per last 100 batches: 0.068660 +[2023-10-07 23:40:58,084][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29178.5/29522=98.84% | mean: 0.01 | max: 5.32 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.14 | max: 6.22 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] whose quote the meaning of life is to find your gift [SEP] ### +### [P_TEXT]: [CLS] athe purpose of life is to discover your gift ; the work of life is to develop it ; ### +### and the meaning of life is to give your gift away. a a david viscott. a note on the wall says, athe ### +### meaning of life is to find your gift ; the purpose of life is to give it away. a. at one point, ### +### santini summarizes his philosophy : the meaning of life is to find your gift, and the purpose of ### +### life is to give it away. also in 2006 a book titled ajust do it! : the power of positive livinga ### +### used the saying as an epigraph for a chapter. [SEP] ### +### ======================================= h_v_q | Gates: 27331 ======================================= ### +### ('gift', 0, 4) ('life', 1, 0) ('quote', 2, 2695) ('whose', 3, 166) ('find', 4, 27) ('your', 5, 294) ### +### ('meaning', 6, 7) ('gifts', 7, 21) ('familiarity', 8, 27101) ('.', 9, 6367) ('lives', 10, 115) ### +### ('found', 11, 1139) ('relating', 12, 24661) ('stylized', 13, 27807) ('finding', 14, 1370) ### +### ('noun', 15, 23739) ('definition', 16, 155) ('seek', 17, 109) ('quoted', 18, 16181) ### +### ('search', 19, 9197) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('life', 1, 0) ('purpose', 63, 1) ('sant', 12698, 2) ('aj', 7169, 3) ('gift', 0, 4) ### +### ('##cott', 25546, 5) ('##ust', 17979, 6) ('meaning', 6, 7) ('living', 59, 8) ('positive', 5159, 9) ### +### ('ˈ', 696, 10) ('hating', 158, 11) ('david', 487, 12) ('away', 5372, 13) ('vis', 19633, 14) ### +### ('##ο', 120, 15) ('unwilling', 248, 16) ('##ini', 23430, 17) ('crashing', 80, 18) ### +### ('meanings', 417, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('gift', 0, 4) ('life', 1, 0) ('meaning', 6, 7) ('find', 4, 27) ('whose', 3, 166) ('gifts', 7, 21) ### +### ('your', 5, 294) ('lives', 10, 115) ('quote', 2, 2695) ('purpose', 63, 1) ('living', 59, 8) ### +### ('means', 22, 51) ('seek', 17, 109) ('encompasses', 32, 52) ('mean', 46, 35) ### +### ('definition', 16, 155) ('##α', 65, 25) ('crashing', 80, 18) ('sharply', 74, 26) ### +### ('crashed', 69, 42) ### +############################################################################################################ +[2023-10-07 23:40:58,085][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:40:58,085][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:40:58,507][root][INFO] - Epoch: 13: Step: 701/1557, loss[v]=0.048239, lr=0.000007, acc@1[1]=247.0/256=0.96484375, acc@1[2]=254.0/256=0.9921875 +[2023-10-07 23:42:15,163][root][INFO] - Train batch 800 +[2023-10-07 23:42:15,163][root][INFO] - Avg. loss per last 100 batches: 0.064058 +[2023-10-07 23:42:15,889][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29216.7/29522=98.97% | mean: 0.01 | max: 5.53 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.14 | max: 6.24 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what kind of insurance has deductibles [SEP] ### +### [P_TEXT]: [CLS] if you know health insurance. not take a beating. a fee - for - service plan is ### +### what most americans think of as traditional health insurance. the deductibles and co - pays for ### +### this type of plan are generally higher than those of a managed health care provider. these fees can ### +### add up over the course of a year, but they also buy more options. [SEP] ### +### ======================================= h_v_q | Gates: 28421 ======================================= ### +### ('insurance', 0, 3) ('de', 1, 16) ('##ible', 2, 100) ('has', 3, 16225) ('types', 4, 44) ### +### ('familiarity', 5, 26669) ('##s', 6, 571) ('stylized', 7, 27250) ('.', 8, 14350) ('type', 9, 327) ### +### ('kind', 10, 284) ('having', 11, 4984) ('consisting', 12, 22679) ('relating', 13, 24168) ### +### ('sort', 14, 868) ('variety', 15, 1978) ('##ct', 16, 2415) ('##du', 17, 1715) ('class', 18, 5738) ### +### ('##स', 19, 282) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('fee', 16048, 0) ('fees', 18781, 1) ('ˈ', 96, 2) ('insurance', 0, 3) ('beating', 15217, 4) ### +### ('hating', 47, 5) ('unwilling', 43, 6) ('plan', 4941, 7) ('health', 1080, 8) ('##ο', 38, 9) ### +### ('traditional', 3541, 10) ('americans', 547, 11) ('stumbled', 40, 12) ('hesitated', 34, 13) ### +### ('−', 49, 14) ('crashing', 20, 15) ('de', 1, 16) ('service', 1199, 17) ('know', 11900, 18) ### +### ('encompasses', 125, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('insurance', 0, 3) ('de', 1, 16) ('types', 4, 44) ('##ible', 2, 100) ('crashing', 20, 15) ### +### ('crashed', 22, 38) ('##₂', 25, 26) ('hesitated', 34, 13) ('##α', 28, 30) ('unwilling', 43, 6) ### +### ('##ο', 38, 9) ('hating', 47, 5) ('stumbled', 40, 12) ('kind', 10, 284) ('type', 9, 327) ### +### ('sharply', 41, 25) ('−', 49, 14) ('ˈ', 96, 2) ('simon', 33, 48) ('angrily', 35, 42) ### +############################################################################################################ +[2023-10-07 23:42:15,889][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:42:15,890][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:42:16,294][root][INFO] - Epoch: 13: Step: 801/1557, loss[v]=0.057655, lr=0.000007, acc@1[1]=244.5/256=0.955078125, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 23:43:33,780][root][INFO] - Train batch 900 +[2023-10-07 23:43:33,781][root][INFO] - Avg. loss per last 100 batches: 0.063936 +[2023-10-07 23:43:34,480][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29152.6/29522=98.75% | mean: 0.01 | max: 5.62 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.22 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] meaning for the name raines [SEP] ### +### [P_TEXT]: [CLS] buy jpg image a ». raines is a name that was brought to england by the ancestors of ### +### the raines family when they emigrated following the norman conquest of 1066. the name raines comes ### +### from the short forms of various germanic personal names containing the element ragin, meaning ### +### counsel. it it thought that the name could also have been derived from rennes, in brittany. ### +### citation [ close ] lowe, mark anthony, patronymica britannica, a dictionary of family names of the ### +### united kingdom. [SEP] ### +### ======================================= h_v_q | Gates: 27091 ======================================= ### +### ('rain', 0, 0) ('##es', 1, 2) ('name', 2, 18) ('meaning', 3, 33) ('familiarity', 4, 27427) ### +### ('noun', 5, 20139) ('means', 6, 106) ('genus', 7, 4163) ('surname', 8, 41) ('stylized', 9, 27132) ### +### ('definition', 10, 156) ('relating', 11, 24758) ('language', 12, 8220) ('consisting', 13, 22663) ### +### ('plural', 14, 12627) ('latin', 15, 2807) ('sense', 16, 12005) ('named', 17, 107) ### +### ('mathematics', 18, 24466) ('something', 19, 7172) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('rain', 0, 0) ('rag', 9801, 1) ('##es', 1, 2) ('»', 13035, 3) ('ˈ', 177, 4) ('germanic', 981, 5) ### +### ('counsel', 11565, 6) ('names', 26, 7) ('patron', 4578, 8) ('jp', 25327, 9) ('##ο', 215, 10) ### +### ('lowe', 15674, 11) ('ren', 3347, 12) ('ancestors', 15500, 13) ('conquest', 1576, 14) ### +### ('encompasses', 219, 15) ('crashing', 54, 16) ('derived', 1381, 17) ('name', 2, 18) ### +### ('norman', 2149, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('rain', 0, 0) ('##es', 1, 2) ('name', 2, 18) ('meaning', 3, 33) ('means', 6, 106) ### +### ('surname', 8, 41) ('names', 26, 7) ('definition', 10, 156) ('named', 17, 107) ('nickname', 28, 48) ### +### ('crashing', 54, 16) ('crashed', 43, 27) ('prehistoric', 38, 46) ('mean', 24, 94) ('−', 63, 22) ### +### ('sharply', 65, 29) ('rains', 55, 50) ('ˈ', 177, 4) ('precipitation', 29, 113) ('angrily', 74, 35) ### +############################################################################################################ +[2023-10-07 23:43:34,480][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:43:34,480][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:43:34,902][root][INFO] - Epoch: 13: Step: 901/1557, loss[v]=0.054663, lr=0.000007, acc@1[1]=241.5/256=0.943359375, acc@1[2]=252.5/256=0.986328125 +[2023-10-07 23:44:51,227][root][INFO] - Train batch 1000 +[2023-10-07 23:44:51,228][root][INFO] - Avg. loss per last 100 batches: 0.062842 +[2023-10-07 23:44:51,932][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29134.8/29522=98.69% | mean: 0.01 | max: 5.59 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.52 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is tess domain [SEP] ### +### [P_TEXT]: [CLS] t - tess is the texas teacher evaluation and support system. it is a new teacher ### +### evaluation. system for the state of texas designed to support teachers in their professional ### +### development. and help them grow and improve as educators. [SEP] ### +### ======================================= h_v_q | Gates: 26850 ======================================= ### +### ('tess', 0, 0) ('domain', 1, 12577) ('is', 2, 377) ('##sam', 3, 28486) ('familiarity', 4, 24502) ### +### ('domains', 5, 20290) ('encompasses', 6, 8) ('relating', 7, 23872) ('stylized', 8, 26669) ### +### ('plural', 9, 21839) ('.', 10, 8932) ('stands', 11, 4169) ('designed', 12, 68) ### +### ('definition', 13, 32) ('refers', 14, 12393) ('encyclopedia', 15, 5942) ('consisting', 16, 21917) ### +### ('language', 17, 8595) ('mathematics', 18, 15183) ('provides', 19, 4514) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('tess', 0, 0) ('teacher', 2511, 1) ('t', 1884, 2) ('texas', 349, 3) ('teachers', 4709, 4) ### +### ('evaluation', 9832, 5) ('support', 2330, 6) ('ˈ', 1885, 7) ('encompasses', 6, 8) ('##ο', 961, 9) ### +### ('##₂', 95, 10) ('unwilling', 398, 11) ('crashing', 519, 12) ('system', 538, 13) ('−', 308, 14) ### +### ('helps', 11021, 15) ('tx', 27461, 16) ('hating', 785, 17) ('##α', 393, 18) ('define', 11860, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('tess', 0, 0) ('encompasses', 6, 8) ('is', 2, 377) ('definition', 13, 32) ('designed', 12, 68) ### +### ('domain', 1, 12577) ('anton', 20, 90) ('nash', 24, 111) ('##₂', 95, 10) ('kelly', 23, 124) ### +### ('texas', 349, 3) ('simon', 101, 43) ('−', 308, 14) ('sharply', 235, 20) ('development', 106, 52) ### +### ('unwilling', 398, 11) ('angrily', 181, 40) ('crashing', 519, 12) ('##α', 393, 18) ### +### ('ryan', 27, 346) ### +############################################################################################################ +[2023-10-07 23:44:51,932][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:44:51,932][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:44:52,341][root][INFO] - Epoch: 13: Step: 1001/1557, loss[v]=0.064940, lr=0.000007, acc@1[1]=244.0/256=0.953125, acc@1[2]=249.0/256=0.97265625 +[2023-10-07 23:46:08,220][root][INFO] - Train batch 1100 +[2023-10-07 23:46:08,221][root][INFO] - Avg. loss per last 100 batches: 0.062754 +[2023-10-07 23:46:08,916][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29178.0/29522=98.83% | mean: 0.01 | max: 5.65 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.28 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is known as the droste effect? [SEP] ### +### [P_TEXT]: [CLS] multiple droste effects in a mirror shop in the tunis medina. the droste effect ( ### +### dutch pronunciation : [ dreste ] ) aknown as mise en abyme in artais the effect of a picture ### +### appearing within itself, in a place where a similar picture would realistically be expected to ### +### appear. the appearance is recursive : the smaller version contains an even smaller version of the ### +### picture, and so on. [SEP] ### +### ======================================= h_v_q | Gates: 27981 ======================================= ### +### ('dr', 0, 4) ('##ost', 1, 11) ('##e', 2, 29) ('known', 3, 6001) ('effect', 4, 1) ### +### ('familiarity', 5, 27611) ('stylized', 6, 26255) ('is', 7, 7800) ('nicknamed', 8, 106) ### +### ('##sam', 9, 26646) ('encompasses', 10, 76) ('famous', 11, 14891) ('relating', 12, 22468) ### +### ('refers', 13, 5457) ('effects', 14, 7) ('consisting', 15, 24729) ('.', 16, 16025) ### +### ('developed', 17, 11963) ('plural', 18, 9657) ('noted', 19, 754) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('tunisia', 1144, 0) ('effect', 4, 1) ('##yme', 22686, 2) ('picture', 5468, 3) ('dr', 0, 4) ### +### ('mirror', 870, 5) ('##ais', 25045, 6) ('effects', 14, 7) ('tunis', 24194, 8) ('medina', 14662, 9) ### +### ('multiple', 1093, 10) ('##ost', 1, 11) ('dutch', 6770, 12) ('ˈ', 352, 13) ('crashing', 29, 14) ### +### ('shop', 4877, 15) ('ak', 12606, 16) ('mis', 21521, 17) ('##est', 3124, 18) ### +### ('appearance', 7397, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('dr', 0, 4) ('##ost', 1, 11) ('effect', 4, 1) ('##e', 2, 29) ('effects', 14, 7) ### +### ('encompasses', 10, 76) ('nicknamed', 8, 106) ('definition', 22, 35) ('crashing', 29, 14) ### +### ('crashed', 31, 26) ('−', 80, 23) ('annoyance', 63, 33) ('known', 3, 6001) ('##α', 48, 66) ### +### ('sharply', 75, 43) ('##大', 76, 42) ('simon', 50, 74) ('stumbled', 103, 28) ('angrily', 68, 65) ### +### ('hating', 105, 32) ### +############################################################################################################ +[2023-10-07 23:46:08,916][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:46:08,917][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:46:09,340][root][INFO] - Epoch: 13: Step: 1101/1557, loss[v]=0.047567, lr=0.000007, acc@1[1]=244.0/256=0.953125, acc@1[2]=250.5/256=0.978515625 +[2023-10-07 23:47:26,761][root][INFO] - Train batch 1200 +[2023-10-07 23:47:26,762][root][INFO] - Avg. loss per last 100 batches: 0.063365 +[2023-10-07 23:47:27,444][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29138.7/29522=98.70% | mean: 0.01 | max: 5.50 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.16 | max: 6.32 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is leprechauns t in irish folklore [SEP] ### +### [P_TEXT]: [CLS] a leprechaun in irish folklore was, a little sprite, or goblin. the name leprechaun ### +### may have derived from the irish leath brogan or shoemaker, although its origins may lie in ### +### luacharma'n irish for pygmy. these apparently aged, diminutive men are frequently to be found in an ### +### intoxicated state, caused by home - brew poteen. [SEP] ### +### ======================================= h_v_q | Gates: 28005 ======================================= ### +### ('le', 0, 1) ('irish', 1, 2) ('##un', 2, 21) ('folklore', 3, 4) ('##cha', 4, 12) ('##pre', 5, 14) ### +### ('familiarity', 6, 27306) ('t', 7, 11333) ('is', 8, 4636) ('ireland', 9, 5) ('encompasses', 10, 80) ### +### ('##s', 11, 24451) ('stylized', 12, 27525) ('.', 13, 17675) ('relating', 14, 25729) ### +### ('##sam', 15, 24895) ('consisting', 16, 23102) ('refers', 17, 21447) ('stands', 18, 3414) ### +### ('definition', 19, 56) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('goblin', 8625, 0) ('le', 0, 1) ('irish', 1, 2) ('##utive', 21797, 3) ('folklore', 3, 4) ### +### ('ireland', 9, 5) ('##maker', 13272, 6) ('lea', 20847, 7) ('##gan', 1227, 8) ('##cated', 22199, 9) ### +### ('##rite', 25262, 10) ('shoe', 12663, 11) ('##cha', 4, 12) ('crashing', 28, 13) ('##pre', 5, 14) ### +### ('ˈ', 233, 15) ('##ο', 41, 16) ('little', 15432, 17) ('##een', 23518, 18) ('stumbled', 72, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('le', 0, 1) ('irish', 1, 2) ('folklore', 3, 4) ('##un', 2, 21) ('##cha', 4, 12) ('##pre', 5, 14) ### +### ('ireland', 9, 5) ('encompasses', 10, 80) ('crashing', 28, 13) ('crashed', 23, 24) ('##α', 24, 37) ### +### ('definition', 19, 56) ('sharply', 34, 28) ('angrily', 26, 50) ('##ο', 41, 16) ('stumbled', 72, 19) ### +### ('−', 57, 29) ('simon', 31, 109) ('ruined', 54, 48) ('##₂', 50, 53) ### +############################################################################################################ +[2023-10-07 23:47:27,444][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:47:27,444][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:47:27,848][root][INFO] - Epoch: 13: Step: 1201/1557, loss[v]=0.046093, lr=0.000007, acc@1[1]=248.0/256=0.96875, acc@1[2]=253.5/256=0.990234375 +[2023-10-07 23:48:44,461][root][INFO] - Train batch 1300 +[2023-10-07 23:48:44,462][root][INFO] - Avg. loss per last 100 batches: 0.064252 +[2023-10-07 23:48:45,178][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29110.2/29522=98.61% | mean: 0.01 | max: 5.42 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.31 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] who is harry potter's main antagonist [SEP] ### +### [P_TEXT]: [CLS] harry potter is the protagonist while lord voldemort is the antagonist. voldemort ### +### is also known as tom riddle, the dark lord, you - know - who, or he - who - must - not - bea¦ - ### +### named. [SEP] ### +### ======================================= h_v_q | Gates: 26759 ======================================= ### +### ('potter', 0, 1) ('antagonist', 1, 3) ('harry', 2, 6) ('main', 3, 3152) ('.', 4, 16527) ### +### ('familiarity', 5, 25667) ('american', 6, 24639) ('who', 7, 36) ('whose', 8, 111) ### +### ('primary', 9, 4086) ('is', 10, 1091) ('stylized', 11, 25290) ('principal', 12, 2744) ### +### ('born', 13, 9262) ('relating', 14, 18587) ('was', 15, 6567) ('league', 16, 12978) ### +### ('##sam', 17, 26610) ('consisting', 18, 23502) ('primarily', 19, 3775) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('riddle', 13630, 0) ('potter', 0, 1) ('protagonist', 78, 2) ('antagonist', 1, 3) ('lord', 1680, 4) ### +### ('bea', 20210, 5) ('harry', 2, 6) ('tom', 881, 7) ('vol', 15051, 8) ('##¦', 29506, 9) ### +### ('##rt', 17483, 10) ('dark', 2164, 11) ('cyrillic', 373, 12) ('named', 3226, 13) ('ˈ', 769, 14) ### +### ('##ο', 137, 15) ('nicknamed', 5110, 16) ('crashed', 46, 17) ('hating', 80, 18) ### +### ('crashing', 66, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('potter', 0, 1) ('antagonist', 1, 3) ('harry', 2, 6) ('who', 7, 36) ('whose', 8, 111) ### +### ('main', 3, 3152) ('protagonist', 78, 2) ('character', 22, 34) ('crashed', 46, 17) ### +### ('crashing', 66, 19) ('##α', 34, 52) ('hating', 80, 18) ('annoyance', 64, 29) ('knew', 31, 67) ### +### ('sharply', 77, 26) ('##ο', 137, 15) ('hated', 98, 27) ('simon', 50, 65) ('julian', 32, 89) ### +### ('−', 86, 39) ### +############################################################################################################ +[2023-10-07 23:48:45,179][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:48:45,179][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:48:45,603][root][INFO] - Epoch: 13: Step: 1301/1557, loss[v]=0.044825, lr=0.000006, acc@1[1]=246.5/256=0.962890625, acc@1[2]=252.5/256=0.986328125 +[2023-10-07 23:50:01,971][root][INFO] - Train batch 1400 +[2023-10-07 23:50:01,972][root][INFO] - Avg. loss per last 100 batches: 0.067047 +[2023-10-07 23:50:02,695][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29149.2/29522=98.74% | mean: 0.01 | max: 5.61 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.28 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] kai meaning in japanese [SEP] ### +### [P_TEXT]: [CLS] kaidan ( æªe « ) ( sometimes transliterated kwaidan ) is a japanese word consisting ### +### of two kanji : æª ( kai ) meaning astrange, mysterious, rare or bewitching apparition and e « ( dan ### +### ) meaning atalka or arecited narrative. a. 1 1 overall meaning and usage. 2 hyakumonogatari ### +### kaidankai and kaidanshu. [SEP] ### +### ======================================= h_v_q | Gates: 25950 ======================================= ### +### ('kai', 0, 0) ('japanese', 1, 5) ('meaning', 2, 17) ('noun', 3, 9159) ('definition', 4, 32) ### +### ('familiarity', 5, 25689) ('means', 6, 29) ('chinese', 7, 421) ('japan', 8, 14) ('spanish', 9, 620) ### +### ('sense', 10, 6885) ('.', 11, 9178) ('something', 12, 7320) ('relating', 13, 21373) ### +### ('german', 14, 639) ('latin', 15, 4329) ('stylized', 16, 22078) ('refers', 17, 14750) ### +### ('plural', 18, 4702) ('expression', 19, 1204) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('kai', 0, 0) ('##dan', 5581, 1) ('«', 13023, 2) ('kw', 23404, 3) ('ata', 20679, 4) ### +### ('japanese', 1, 5) ('##ª', 1232, 6) ('word', 40, 7) ('mysterious', 4531, 8) ('narrative', 7965, 9) ### +### ('app', 8029, 10) ('##aid', 26323, 11) ('rare', 2045, 12) ('##cite', 26631, 13) ('japan', 8, 14) ### +### ('æ', 25958, 15) ('##an', 2242, 16) ('meaning', 2, 17) ('translit', 22968, 18) ('ˈ', 1555, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('kai', 0, 0) ('japanese', 1, 5) ('meaning', 2, 17) ('definition', 4, 32) ('japan', 8, 14) ### +### ('means', 6, 29) ('word', 40, 7) ('chinese', 7, 421) ('tokyo', 38, 31) ('spanish', 9, 620) ### +### ('term', 21, 148) ('italian', 26, 137) ('words', 61, 50) ('defined', 39, 118) ('mean', 81, 42) ### +### ('symbol', 29, 172) ('german', 14, 639) ('##α', 138, 24) ('noun', 3, 9159) ('meant', 88, 71) ### +############################################################################################################ +[2023-10-07 23:50:02,695][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:50:02,695][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:50:03,123][root][INFO] - Epoch: 13: Step: 1401/1557, loss[v]=0.049523, lr=0.000006, acc@1[1]=241.5/256=0.943359375, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 23:51:19,746][root][INFO] - Train batch 1500 +[2023-10-07 23:51:19,747][root][INFO] - Avg. loss per last 100 batches: 0.065061 +[2023-10-07 23:51:20,427][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29190.5/29522=98.88% | mean: 0.01 | max: 5.39 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.18 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a knowledge broker [SEP] ### +### [P_TEXT]: [CLS] a new study describes the need for aknowledge brokersa a a new kind of ### +### communication professional that specializes in bridging the science and policy worlds. a knowledge ### +### brokeras job is to engage scientists in the regulatory process and to engage regulators in the ### +### scientific process. the ideal knowledge broker has phd - level training in the scientific area plus ### +### extensive knowledge of the regulatory process. [SEP] ### +### ======================================= h_v_q | Gates: 26331 ======================================= ### +### ('broker', 0, 0) ('knowledge', 1, 3) ('definition', 2, 102) ('familiarity', 3, 23202) ### +### ('relating', 4, 21569) ('or', 5, 21368) ('encompasses', 6, 47) ('noun', 7, 25059) ('is', 8, 1922) ### +### ('refers', 9, 19141) ('encyclopedia', 10, 9103) ('##sam', 11, 21426) ('plural', 12, 21131) ### +### ('a', 13, 14957) ('language', 14, 4990) ('stands', 15, 4171) (';', 16, 7423) ### +### ('stylized', 17, 28659) ('consisting', 18, 27021) ('mathematics', 19, 16107) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('broker', 0, 0) ('ak', 5828, 1) ('regulatory', 8133, 2) ('knowledge', 1, 3) ('##ge', 14084, 4) ### +### ('##sa', 8489, 5) ('regulators', 25594, 6) ('ˈ', 359, 7) ('professional', 167, 8) ### +### ('##ging', 14111, 9) ('communication', 192, 10) ('ideal', 3347, 11) ('job', 3127, 12) ### +### ('unwilling', 277, 13) ('##ο', 397, 14) ('hating', 189, 15) ('##now', 22578, 16) ### +### ('regulator', 22762, 17) ('process', 82, 18) ('stumbled', 106, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('broker', 0, 0) ('knowledge', 1, 3) ('definition', 2, 102) ('encompasses', 6, 47) ### +### ('learning', 24, 68) ('crashing', 75, 20) ('scientific', 70, 25) ('knowing', 49, 45) ### +### ('process', 82, 18) ('sharply', 67, 32) ('science', 85, 21) ('is', 8, 1922) ('crashed', 79, 40) ### +### ('professional', 167, 8) ('stumbled', 106, 19) ('communication', 192, 10) ('##₂', 145, 24) ### +### ('hating', 189, 15) ('describes', 35, 165) ('defined', 29, 287) ### +############################################################################################################ +[2023-10-07 23:51:20,427][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:51:20,427][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:51:20,851][root][INFO] - Epoch: 13: Step: 1501/1557, loss[v]=0.055299, lr=0.000006, acc@1[1]=243.5/256=0.951171875, acc@1[2]=252.0/256=0.984375 +[2023-10-07 23:52:04,691][root][INFO] - rank=1; last iteration 1557 +[2023-10-07 23:52:04,692][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 23:52:04,692][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-07 23:52:04,693][root][INFO] - rank=3; last iteration 1557 +[2023-10-07 23:52:04,694][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 23:52:04,694][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-07 23:52:04,695][root][INFO] - rank=2; last iteration 1557 +[2023-10-07 23:52:04,696][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 23:52:04,696][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-07 23:52:04,698][root][INFO] - rank=0; last iteration 1557 +[2023-10-07 23:52:04,699][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-07 23:52:04,699][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-07 23:52:04,700][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 23:52:04,701][root][INFO] - Epoch finished on 1 +[2023-10-07 23:52:04,703][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 23:52:04,703][root][INFO] - Epoch finished on 3 +[2023-10-07 23:52:04,704][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 23:52:04,704][root][INFO] - Epoch finished on 2 +[2023-10-07 23:52:04,705][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-07 23:52:04,706][root][INFO] - Epoch finished on 0 +[2023-10-07 23:52:41,818][root][INFO] - Saved checkpoint at ./vdr_13 +[2023-10-07 23:52:41,819][root][INFO] - Av Loss per epoch=0.063981 +[2023-10-07 23:52:41,819][root][INFO] - epoch total (1) correct predictions=379118 +[2023-10-07 23:52:41,819][root][INFO] - epoch total (2) correct predictions=391321 +[2023-10-07 23:52:41,820][root][INFO] - Saved checkpoint at ./vdr_13 +[2023-10-07 23:52:41,821][root][INFO] - Av Loss per epoch=0.063981 +[2023-10-07 23:52:41,821][root][INFO] - epoch total (1) correct predictions=379118 +[2023-10-07 23:52:41,821][root][INFO] - epoch total (2) correct predictions=391321 +[2023-10-07 23:52:41,820][root][INFO] - Saved checkpoint at ./vdr_13 +[2023-10-07 23:52:41,821][root][INFO] - Av Loss per epoch=0.063981 +[2023-10-07 23:52:41,822][root][INFO] - epoch total (1) correct predictions=379118 +[2023-10-07 23:52:41,822][root][INFO] - epoch total (2) correct predictions=391321 +[2023-10-07 23:52:41,823][root][INFO] - ***** Epoch 14 ***** +[2023-10-07 23:52:41,825][root][INFO] - Saved checkpoint at ./vdr_13 +[2023-10-07 23:52:41,825][root][INFO] - Av Loss per epoch=0.063981 +[2023-10-07 23:52:41,826][root][INFO] - epoch total (1) correct predictions=379118 +[2023-10-07 23:52:41,826][root][INFO] - epoch total (2) correct predictions=391321 +[2023-10-07 23:52:41,825][root][INFO] - ***** Epoch 14 ***** +[2023-10-07 23:52:41,829][root][INFO] - rank=3; Iteration start +[2023-10-07 23:52:41,830][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 23:52:41,830][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 23:52:41,827][root][INFO] - ***** Epoch 14 ***** +[2023-10-07 23:52:41,832][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-07 23:52:41,832][root][INFO] - rank=0; Iteration start +[2023-10-07 23:52:41,832][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 23:52:41,832][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 23:52:41,834][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-07 23:52:41,833][root][INFO] - ***** Epoch 14 ***** +[2023-10-07 23:52:41,835][root][INFO] - rank=1; Iteration start +[2023-10-07 23:52:41,835][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 23:52:41,835][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 23:52:41,837][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-07 23:52:41,839][root][INFO] - rank=2; Iteration start +[2023-10-07 23:52:41,839][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-07 23:52:41,839][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-07 23:52:41,841][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-07 23:52:42,815][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29153.1/29522=98.75% | mean: 0.01 | max: 5.12 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.15 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how long to boil parsnips [SEP] ### +### [P_TEXT]: [CLS] prep. cook. ready in. 1 in a large saucepan cover parsnips with water, cover and ### +### boil over medium - high heat until tender, about 10 minutes. 2 drain. 3 in a plastic bag combine ### +### flour and seasoning salt. 4 dip parsnips in butter and place them in the bag. 5 shake bag to coat ### +### parsnips with the seasoned flour. [SEP] ### +### ======================================= h_v_q | Gates: 27959 ======================================= ### +### ('##ni', 0, 17) ('par', 1, 3) ('minutes', 2, 16) ('boil', 3, 8) ('##ps', 4, 14) ### +### ('familiarity', 5, 26405) ('hours', 6, 17671) ('.', 7, 5288) ('stylized', 8, 27151) ('##s', 9, 284) ### +### ('to', 10, 309) ('days', 11, 14054) ('onto', 12, 136) ('boiling', 13, 81) ('weeks', 14, 246) ### +### ('minute', 15, 325) ('approximately', 16, 5648) ('##acker', 17, 61) ('consisting', 18, 25115) ### +### ('##と', 19, 129) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('flour', 3111, 0) ('ready', 3731, 1) ('prep', 2350, 2) ('par', 1, 3) ('hating', 44, 4) ### +### ('ˈ', 160, 5) ('drain', 940, 6) ('##ο', 75, 7) ('boil', 3, 8) ('unwilling', 69, 9) ### +### ('until', 30, 10) ('tender', 14088, 11) ('sauce', 2976, 12) ('crashing', 63, 13) ('##ps', 4, 14) ### +### ('##pan', 21716, 15) ('minutes', 2, 16) ('##ni', 0, 17) ('−', 25, 18) ('cover', 6467, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('par', 1, 3) ('##ni', 0, 17) ('boil', 3, 8) ('minutes', 2, 16) ('##ps', 4, 14) ('30', 20, 42) ### +### ('boiling', 13, 81) ('##acker', 17, 61) ('until', 30, 10) ('−', 25, 18) ('onto', 12, 136) ### +### ('##s', 9, 284) ('hating', 44, 4) ('##と', 19, 129) ('##α', 33, 36) ('##大', 34, 35) ### +### ('angrily', 31, 49) ('to', 10, 309) ('weeks', 14, 246) ('##ο', 75, 7) ### +############################################################################################################ +[2023-10-07 23:52:42,815][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:52:42,815][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:52:43,214][root][INFO] - Epoch: 14: Step: 1/1557, loss[v]=0.047351, lr=0.000006, acc@1[1]=243.5/256=0.951171875, acc@1[2]=253.0/256=0.98828125 +[2023-10-07 23:54:00,182][root][INFO] - Train batch 100 +[2023-10-07 23:54:00,182][root][INFO] - Avg. loss per last 100 batches: 0.060791 +[2023-10-07 23:54:00,893][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29149.1/29522=98.74% | mean: 0.01 | max: 5.55 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.14 | max: 6.30 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how far is richmond va to williamsburg va [SEP] ### +### [P_TEXT]: [CLS] distance, gas consumption and emission notes. distance from williamsburg, va to ### +### richmond, va is 52miles or 83 km. you can get this distance about 53 mins. if you want to planning ### +### travel with plane for 45 miles or 72 km, you can get this distance about 35 mins. a car with an ### +### average mpg will needs 2. 41 gallons of gas to get the route between these points. the estimated ### +### cost of gas to get between williamsburg, va and richmond, va is $ 5. 47. during the route, an ### +### average car will release 47. 22 pounds of co2 to the atmosphere. your carbon footprint is 0. 91 ### +### pounds of co2 per mile. [SEP] ### +### ======================================= h_v_q | Gates: 26460 ======================================= ### +### ('williamsburg', 0, 0) ('richmond', 1, 1) ('virginia', 2, 45) ('miles', 3, 53) ('distance', 4, 4) ### +### ('va', 5, 2) ('inches', 6, 15416) ('far', 7, 44) ('familiarity', 8, 25074) ('.', 9, 11347) ### +### ('=', 10, 18148) ('hampshire', 11, 1672) ('washington', 12, 306) ('plural', 13, 14929) ### +### ('carolina', 14, 475) ('stylized', 15, 28451) ('pennsylvania', 16, 1134) ### +### ('approximately', 17, 1888) ('430', 18, 28581) ('brooklyn', 19, 73) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('williamsburg', 0, 0) ('richmond', 1, 1) ('va', 5, 2) ('ˈ', 304, 3) ('distance', 4, 4) ### +### ('gas', 6694, 5) ('45', 35, 6) ('gallons', 3985, 7) ('cost', 1806, 8) ('mile', 22, 9) ### +### ('hating', 530, 10) ('unwilling', 206, 11) ('$', 1734, 12) ('##ο', 1439, 13) ('costs', 5437, 14) ### +### ('travel', 1399, 15) ('footprint', 8941, 16) ('distances', 118, 17) ('car', 2121, 18) ### +### ('crashing', 350, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('williamsburg', 0, 0) ('richmond', 1, 1) ('va', 5, 2) ('distance', 4, 4) ('virginia', 2, 45) ### +### ('miles', 3, 53) ('far', 7, 44) ('mile', 22, 9) ('brooklyn', 19, 73) ('45', 35, 6) ('30', 26, 85) ### +### ('farther', 28, 89) ('washington', 12, 306) ('carolina', 14, 475) ('distances', 118, 17) ### +### ('shoved', 92, 43) ('##と', 51, 87) ('ˈ', 304, 3) ('pounds', 102, 49) ('unwilling', 206, 11) ### +############################################################################################################ +[2023-10-07 23:54:00,893][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:54:00,893][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:54:01,313][root][INFO] - Epoch: 14: Step: 101/1557, loss[v]=0.045857, lr=0.000006, acc@1[1]=244.5/256=0.955078125, acc@1[2]=254.0/256=0.9921875 +[2023-10-07 23:55:17,451][root][INFO] - Train batch 200 +[2023-10-07 23:55:17,452][root][INFO] - Avg. loss per last 100 batches: 0.065129 +[2023-10-07 23:55:18,146][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29139.6/29522=98.70% | mean: 0.01 | max: 5.70 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.14 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] jupiter number of moons [SEP] ### +### [P_TEXT]: [CLS] some asteroids share the same names as moons of jupiter : 9 metis, 38 leda, 52 ### +### europa, 85 io, 113 amalthea, 239 adrastea. two more asteroids previously shared the names of jovian ### +### moons until spelling differences were made permanent by the iau : ganymede and asteroid 1036 ### +### ganymed ; and callisto and asteroid 204 kallisto. [SEP] ### +### ======================================= h_v_q | Gates: 26125 ======================================= ### +### ('jupiter', 0, 1) ('moons', 1, 2) ('number', 2, 1051) ('.', 3, 14056) ('moon', 4, 3) ### +### ('familiarity', 5, 25237) ('800', 6, 12487) ('earth', 7, 585) ('144', 8, 23450) ('total', 9, 16282) ### +### ('000', 10, 27608) ('volume', 11, 21444) ('phoenix', 12, 151) ('million', 13, 22628) ### +### ('neptune', 14, 128) ('relating', 15, 24311) ('stylized', 16, 28731) ('145', 17, 10645) ### +### ('inches', 18, 25654) ('numerous', 19, 284) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('asteroids', 7102, 0) ('jupiter', 0, 1) ('moons', 1, 2) ('moon', 4, 3) ('metis', 15465, 4) ### +### ('##tea', 28240, 5) ('asteroid', 8475, 6) ('names', 3418, 7) ('spelling', 8448, 8) ### +### ('gan', 10113, 9) ('europa', 3692, 10) ('shared', 8831, 11) ('cyrillic', 609, 12) ('##ο', 125, 13) ### +### ('jo', 10389, 14) ('ˈ', 398, 15) ('##yme', 25260, 16) ('hesitated', 388, 17) ('hating', 111, 18) ### +### ('crashing', 265, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('jupiter', 0, 1) ('moons', 1, 2) ('moon', 4, 3) ('number', 2, 1051) ('neptune', 14, 128) ### +### ('phoenix', 12, 151) ('diameter', 27, 88) ('angrily', 44, 42) ('satellites', 39, 53) ### +### ('##α', 65, 39) ('ruined', 70, 38) ('##ο', 125, 13) ('−', 92, 27) ('hating', 111, 18) ### +### ('hated', 72, 46) ('crashed', 101, 25) ('30', 77, 51) ('.', 3, 14056) ('##ང', 145, 21) ### +### ('earth', 7, 585) ### +############################################################################################################ +[2023-10-07 23:55:18,147][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:55:18,147][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:55:18,566][root][INFO] - Epoch: 14: Step: 201/1557, loss[v]=0.056266, lr=0.000006, acc@1[1]=245.0/256=0.95703125, acc@1[2]=252.5/256=0.986328125 +[2023-10-07 23:56:36,463][root][INFO] - Train batch 300 +[2023-10-07 23:56:36,464][root][INFO] - Avg. loss per last 100 batches: 0.059871 +[2023-10-07 23:56:37,181][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29148.9/29522=98.74% | mean: 0.01 | max: 5.46 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.21 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] foods to eat for colon health [SEP] ### +### [P_TEXT]: [CLS] foods for a healthy colon. mangoes : although seasonal, this tropical fruit is ### +### great for your colon. mangoes have a high fiber a one mango has about 5 grams of fiber a and water ### +### content. both of these can help promote regularity as well as relieve constipation. [SEP] ### +### ======================================= h_v_q | Gates: 27331 ======================================= ### +### ('colon', 0, 1) ('foods', 1, 8) ('health', 2, 3805) ('eat', 3, 206) ('.', 4, 18489) ### +### ('familiarity', 5, 26168) ('fish', 6, 7867) ('for', 7, 745) ('vegetables', 8, 363) ### +### ('products', 9, 4080) ('mushrooms', 10, 443) ('healthcare', 11, 4385) ('simon', 12, 28) ### +### ('stylized', 13, 28962) ('grapes', 14, 873) ('relating', 15, 26844) ('food', 16, 20) ### +### ('onto', 17, 824) ('meat', 18, 331) ('goods', 19, 13613) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('mango', 8681, 0) ('colon', 0, 1) ('ˈ', 145, 2) ('healthy', 1640, 3) ('fiber', 3140, 4) ### +### ('seasonal', 13647, 5) ('fruit', 60, 6) ('tropical', 3369, 7) ('foods', 1, 8) ('hating', 29, 9) ### +### ('fruits', 163, 10) ('−', 28, 11) ('##ο', 144, 12) ('sharply', 61, 13) ('crashing', 141, 14) ### +### ('cyrillic', 308, 15) ('hesitated', 73, 16) ('wingspan', 409, 17) ('unwilling', 93, 18) ### +### ('hated', 48, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('colon', 0, 1) ('foods', 1, 8) ('eat', 3, 206) ('simon', 12, 28) ('food', 16, 20) ### +### ('health', 2, 3805) ('hating', 29, 9) ('−', 28, 11) ('fruit', 60, 6) ('crashed', 26, 21) ### +### ('hated', 48, 19) ('ˈ', 145, 2) ('sharply', 61, 13) ('shoved', 42, 30) ('vegetables', 8, 363) ### +### ('angrily', 50, 31) ('hesitated', 73, 16) ('for', 7, 745) ('##₂', 63, 26) ('ruined', 52, 41) ### +############################################################################################################ +[2023-10-07 23:56:37,181][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:56:37,181][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:56:37,588][root][INFO] - Epoch: 14: Step: 301/1557, loss[v]=0.084725, lr=0.000006, acc@1[1]=238.5/256=0.931640625, acc@1[2]=248.5/256=0.970703125 +[2023-10-07 23:57:55,287][root][INFO] - Train batch 400 +[2023-10-07 23:57:55,288][root][INFO] - Avg. loss per last 100 batches: 0.065852 +[2023-10-07 23:57:56,025][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29191.6/29522=98.88% | mean: 0.01 | max: 5.38 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 5.87 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is ct sales tax [SEP] ### +### [P_TEXT]: [CLS] the connecticut state sales tax rate is 6. 35 %, and the average ct sales tax after ### +### local surtaxes is 6. 35 % 1. groceries, prescription drugs and non - prescription drugs are exempt ### +### from the connecticut sales tax. 2 counties and cities are not allowed to collect local sales taxes. ### +### [SEP] ### +### ======================================= h_v_q | Gates: 26684 ======================================= ### +### ('ct', 0, 0) ('sales', 1, 24) ('tax', 2, 2) ('connecticut', 3, 1) ('familiarity', 4, 26695) ### +### ('is', 5, 493) ('relating', 6, 25567) ('encompasses', 7, 590) ('refers', 8, 26398) ### +### ('##sam', 9, 22681) ('.', 10, 14831) ('stylized', 11, 29250) ('massachusetts', 12, 2071) ### +### ('hampshire', 13, 828) ('plural', 14, 18532) ('california', 15, 2897) ('definition', 16, 12234) ### +### ('noun', 17, 28981) ('taxes', 18, 4) ('designed', 19, 16637) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('ct', 0, 0) ('connecticut', 3, 1) ('tax', 2, 2) ('ˈ', 153, 3) ('taxes', 18, 4) ('##xes', 27447, 5) ### +### ('hating', 159, 6) ('exempt', 8974, 7) ('##ο', 530, 8) ('hesitated', 207, 9) ### +### ('groceries', 11436, 10) ('rate', 1331, 11) ('−', 108, 12) ('hated', 92, 13) ('unwilling', 422, 14) ### +### ('wingspan', 305, 15) ('percentage', 6395, 16) ('##ང', 117, 17) ('rates', 10084, 18) ### +### ('cyrillic', 529, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ct', 0, 0) ('tax', 2, 2) ('connecticut', 3, 1) ('sales', 1, 24) ('taxes', 18, 4) ### +### ('taxation', 20, 30) ('is', 5, 493) ('crashed', 41, 29) ('crashing', 43, 26) ### +### ('encompasses', 7, 590) ('simon', 50, 40) ('hated', 92, 13) ('ˈ', 153, 3) ('−', 108, 12) ### +### ('hartford', 64, 44) ('hating', 159, 6) ('$', 90, 32) ('##ང', 117, 17) ('##α', 121, 22) ### +### ('angrily', 113, 23) ### +############################################################################################################ +[2023-10-07 23:57:56,025][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:57:56,025][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:57:56,437][root][INFO] - Epoch: 14: Step: 401/1557, loss[v]=0.053368, lr=0.000006, acc@1[1]=244.0/256=0.953125, acc@1[2]=251.0/256=0.98046875 +[2023-10-07 23:59:13,342][root][INFO] - Train batch 500 +[2023-10-07 23:59:13,344][root][INFO] - Avg. loss per last 100 batches: 0.060732 +[2023-10-07 23:59:14,051][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29098.3/29522=98.56% | mean: 0.01 | max: 5.33 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.11 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how long should a mother exclusively breastfeed her child for? ( in ideal ### +### circumstances ) [SEP] ### +### [P_TEXT]: [CLS] exclusive breastfeeding ten things you may not know about exclusive breastfeeding ### +### every mother who chooses to exclusively breastfeed her baby in the first 6 months is doing the best ### +### thing she can to help her baby grow and stay healthy not all breastfeeding is equal and the ### +### differences are important : exclusive breastfeeding : breastmilk only - nothing else. 1 mixed ### +### feeding : adding anything else - water, juice, tea, formula, cereals, baby foods or other foods. ### +### mixed feeding can increase the chances of a child getting infections. [SEP] ### +### ======================================= h_v_q | Gates: 27482 ======================================= ### +### ('breast', 0, 4) ('mother', 1, 13) ('ideal', 2, 2887) ('circumstances', 3, 2410) ('child', 4, 17) ### +### ('##fe', 5, 88) ('exclusively', 6, 28) ('her', 7, 216) ('weeks', 8, 45) ('.', 9, 15313) ### +### ('days', 10, 9607) ('months', 11, 122) ('##ed', 12, 39) ('years', 13, 5518) ('primarily', 14, 123) ### +### ('minutes', 15, 6382) ('should', 16, 15558) ('familiarity', 17, 26362) ('length', 18, 9253) ### +### ('miles', 19, 22846) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('exclusive', 137, 0) ('feeding', 6305, 1) ('else', 834, 2) ('chances', 6009, 3) ('breast', 0, 4) ### +### ('baby', 25, 5) ('ˈ', 337, 6) ('mixed', 2232, 7) ('−', 61, 8) ('nothing', 6503, 9) ### +### ('hesitated', 124, 10) ('hating', 80, 11) ('##ο', 162, 12) ('mother', 1, 13) ('mothers', 27, 14) ### +### ('cyrillic', 212, 15) ('unwilling', 121, 16) ('child', 4, 17) ('stumbled', 159, 18) ### +### ('feed', 5365, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('breast', 0, 4) ('mother', 1, 13) ('child', 4, 17) ('exclusively', 6, 28) ('weeks', 8, 45) ### +### ('##fe', 5, 88) ('##ed', 12, 39) ('her', 7, 216) ('months', 11, 122) ('baby', 25, 5) ### +### ('mothers', 27, 14) ('mom', 23, 36) ('primarily', 14, 123) ('circumstances', 3, 2410) ### +### ('children', 24, 51) ('ideal', 2, 2887) ('30', 21, 159) ('−', 61, 8) ('exclusive', 137, 0) ### +### ('hating', 80, 11) ### +############################################################################################################ +[2023-10-07 23:59:14,052][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-07 23:59:14,052][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-07 23:59:14,474][root][INFO] - Epoch: 14: Step: 501/1557, loss[v]=0.096052, lr=0.000006, acc@1[1]=237.5/256=0.927734375, acc@1[2]=250.0/256=0.9765625 +[2023-10-08 00:00:30,635][root][INFO] - Train batch 600 +[2023-10-08 00:00:30,636][root][INFO] - Avg. loss per last 100 batches: 0.063617 +[2023-10-08 00:00:31,340][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29108.3/29522=98.60% | mean: 0.01 | max: 5.14 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.04 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what year was litchfield ct founded [SEP] ### +### [P_TEXT]: [CLS] litchfield is a small town of approximately 8, 000 people located in northwestern ### +### connecticut. founded in 1719, litchfield has a rich history. the town was the home of the first law ### +### school in the united states as well as an early school for girls. at the end of the 19th century, ### +### litchfield residents became leaders in the colonial revival movement. [SEP] ### +### ======================================= h_v_q | Gates: 27616 ======================================= ### +### ('##chfield', 0, 0) ('founded', 1, 7) ('lit', 2, 3) ('connecticut', 3, 1) ('founder', 4, 126) ### +### ('established', 5, 78) ('was', 6, 99) ('familiarity', 7, 24330) ('1964', 8, 18003) ### +### ('opened', 9, 1674) ('1956', 10, 21137) ('founders', 11, 240) ('ct', 12, 4) ('.', 13, 5144) ### +### ('knew', 14, 108) ('stylized', 15, 28384) ('march', 16, 3251) ('april', 17, 4781) ### +### ('1959', 18, 23245) ('november', 19, 10678) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##chfield', 0, 0) ('connecticut', 3, 1) ('revival', 369, 2) ('lit', 2, 3) ('ct', 12, 4) ### +### ('ˈ', 612, 5) ('town', 1939, 6) ('founded', 1, 7) ('encompasses', 785, 8) ('law', 2951, 9) ### +### ('school', 2865, 10) ('stumbled', 355, 11) ('##ο', 759, 12) ('crashing', 170, 13) ### +### ('where', 1545, 14) ('colonial', 441, 15) ('hesitated', 559, 16) ('first', 208, 17) ### +### ('unwilling', 467, 18) ('gideon', 305, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##chfield', 0, 0) ('lit', 2, 3) ('founded', 1, 7) ('connecticut', 3, 1) ('established', 5, 78) ### +### ('founder', 4, 126) ('ct', 12, 4) ('was', 6, 99) ('knew', 14, 108) ('century', 36, 60) ### +### ('early', 51, 48) ('remained', 65, 49) ('founders', 11, 240) ('revival', 369, 2) ### +### ('started', 62, 90) ('crashing', 170, 13) ('became', 64, 121) ('−', 163, 20) ('vermont', 58, 144) ### +### ('##α', 145, 41) ### +############################################################################################################ +[2023-10-08 00:00:31,341][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:00:31,341][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:00:31,764][root][INFO] - Epoch: 14: Step: 601/1557, loss[v]=0.048547, lr=0.000006, acc@1[1]=243.5/256=0.951171875, acc@1[2]=251.5/256=0.982421875 +[2023-10-08 00:01:48,922][root][INFO] - Train batch 700 +[2023-10-08 00:01:48,923][root][INFO] - Avg. loss per last 100 batches: 0.059603 +[2023-10-08 00:01:49,642][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29080.1/29522=98.50% | mean: 0.01 | max: 5.89 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.14 | max: 6.66 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what does title insurance protect [SEP] ### +### [P_TEXT]: [CLS] title insurance is a form of indemnity insurance predominantly found in the united ### +### states which insures against financial loss from defects in title to real property and from the ### +### invalidity or unenforceability of mortgage loans. the vast majority of title insurance policies are ### +### written on land within the united states. [SEP] ### +### ======================================= h_v_q | Gates: 26325 ======================================= ### +### ('title', 0, 0) ('insurance', 1, 1) ('protect', 2, 103) ('titles', 3, 2) ('.', 4, 17222) ### +### ('protection', 5, 194) ('protects', 6, 629) ('familiarity', 7, 26898) ('provides', 8, 9882) ### +### ('relating', 9, 23252) ('is', 10, 325) ('allows', 11, 14965) ('protected', 12, 261) ### +### ('refers', 13, 10228) ('what', 14, 129) ('plural', 15, 6135) ('defense', 16, 5664) ### +### ('##sam', 17, 22465) ('guard', 18, 6661) ('answer', 19, 17807) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('title', 0, 0) ('insurance', 1, 1) ('titles', 3, 2) ('ˈ', 1044, 3) ('##force', 18352, 4) ### +### ('ind', 7740, 5) ('encompasses', 119, 6) ('defects', 10009, 7) ('##ο', 581, 8) ('loss', 1143, 9) ### +### ('crashing', 45, 10) ('##ability', 10355, 11) ('stumbled', 210, 12) ('unwilling', 566, 13) ### +### ('hating', 360, 14) ('−', 62, 15) ('definition', 203, 16) ('define', 10405, 17) ### +### ('hesitated', 452, 18) ('mortgage', 4547, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('title', 0, 0) ('insurance', 1, 1) ('titles', 3, 2) ('protect', 2, 103) ('protection', 5, 194) ### +### ('crashing', 45, 10) ('what', 14, 129) ('protects', 6, 629) ('−', 62, 15) ('is', 10, 325) ### +### ('julian', 38, 57) ('encompasses', 119, 6) ('protected', 12, 261) ('crashed', 78, 25) ### +### ('simon', 47, 61) ('##α', 79, 37) ('beside', 66, 54) ('stumbled', 210, 12) ('angrily', 148, 27) ### +### ('ruined', 125, 34) ### +############################################################################################################ +[2023-10-08 00:01:49,643][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:01:49,643][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:01:50,048][root][INFO] - Epoch: 14: Step: 701/1557, loss[v]=0.087430, lr=0.000006, acc@1[1]=238.5/256=0.931640625, acc@1[2]=250.0/256=0.9765625 +[2023-10-08 00:03:06,699][root][INFO] - Train batch 800 +[2023-10-08 00:03:06,699][root][INFO] - Avg. loss per last 100 batches: 0.058903 +[2023-10-08 00:03:07,383][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 28982.3/29522=98.17% | mean: 0.01 | max: 5.68 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.14 | max: 6.49 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the longest word in english [SEP] ### +### [P_TEXT]: [CLS] pneumonoaultraamicroscopicasilicoavolcanoaconiosis succeeded ### +### electrophotomicrographically as the longest word in the english language recognized by the national ### +### puzzlers'league at the opening session of the organization's 103rd semi - annual meeting held ### +### yesterday at the hotel new yorker. [SEP] ### +### ======================================= h_v_q | Gates: 24883 ======================================= ### +### ('longest', 0, 0) ('word', 1, 4) ('english', 2, 79) ('long', 3, 851) ('words', 4, 23) ### +### ('.', 5, 12379) ('noun', 6, 17972) ('language', 7, 39) ('familiarity', 8, 27222) ('is', 9, 14568) ### +### ('##sam', 10, 27036) ('largest', 11, 8) ('letter', 12, 552) ('refers', 13, 25100) ### +### ('term', 14, 1021) ('phrase', 15, 121) ('longer', 16, 276) ('plural', 17, 13382) ### +### ('british', 18, 7038) ('relating', 19, 26582) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('longest', 0, 0) ('yorker', 25982, 1) ('puzzle', 665, 2) ('yesterday', 5479, 3) ('word', 1, 4) ### +### ('ˈ', 328, 5) ('electro', 7395, 6) ('semi', 7079, 7) ('largest', 11, 8) ('meeting', 1663, 9) ### +### ('##ο', 1286, 10) ('puzzles', 20428, 11) ('succeeded', 7096, 12) ('##ono', 14557, 13) ### +### ('p', 3378, 14) ('recognized', 3834, 15) ('hotel', 721, 16) ('crashing', 304, 17) ### +### ('hesitated', 906, 18) ('##graphic', 22499, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('longest', 0, 0) ('word', 1, 4) ('english', 2, 79) ('words', 4, 23) ('language', 7, 39) ### +### ('largest', 11, 8) ('shortest', 25, 33) ('phrase', 15, 121) ('long', 3, 851) ('longer', 16, 276) ### +### ('encompasses', 24, 224) ('letter', 12, 552) ('biggest', 110, 22) ('most', 48, 122) ### +### ('oldest', 57, 116) ('ˈ', 328, 5) ('term', 14, 1021) ('−', 216, 24) ('##大', 179, 37) ### +### ('crashing', 304, 17) ### +############################################################################################################ +[2023-10-08 00:03:07,383][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:03:07,383][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:03:07,805][root][INFO] - Epoch: 14: Step: 801/1557, loss[v]=0.061013, lr=0.000006, acc@1[1]=242.0/256=0.9453125, acc@1[2]=250.5/256=0.978515625 +[2023-10-08 00:04:24,204][root][INFO] - Train batch 900 +[2023-10-08 00:04:24,205][root][INFO] - Avg. loss per last 100 batches: 0.060849 +[2023-10-08 00:04:24,902][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29085.2/29522=98.52% | mean: 0.01 | max: 5.51 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.14 | max: 6.34 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is ncr [SEP] ### +### [P_TEXT]: [CLS] the national capital region ( ncr ) is not an operational entity because emergency ### +### response is a local function. [SEP] ### +### ======================================= h_v_q | Gates: 27045 ======================================= ### +### ('nc', 0, 1) ('##r', 1, 4) ('plural', 2, 16351) ('is', 3, 367) ('##sam', 4, 26886) ### +### ('carolina', 5, 314) ('encompasses', 6, 36) ('familiarity', 7, 26474) ('relating', 8, 20824) ### +### ('definition', 9, 318) ('refers', 10, 14017) ('stylized', 11, 28746) ('designed', 12, 12644) ### +### ('consisting', 13, 23895) ('##rs', 14, 59) ('noun', 15, 25370) ('language', 16, 16763) ### +### ('term', 17, 3456) ('national', 18, 7) ('globe', 19, 2070) ### +### ======================================= h_v_p | Gates: 29520 ======================================= ### +### ('capital', 518, 0) ('nc', 0, 1) ('emergency', 6168, 2) ('operational', 3576, 3) ('##r', 1, 4) ### +### ('region', 231, 5) ('entity', 282, 6) ('national', 18, 7) ('ˈ', 618, 8) ('regions', 849, 9) ### +### ('response', 5116, 10) ('crashing', 246, 11) ('unwilling', 568, 12) ('hating', 409, 13) ### +### ('sharply', 370, 14) ('stumbled', 206, 15) ('##ο', 2000, 16) ('wingspan', 1351, 17) ### +### ('crashed', 84, 18) ('hesitated', 649, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('nc', 0, 1) ('##r', 1, 4) ('encompasses', 6, 36) ('national', 18, 7) ('##rs', 14, 59) ### +### ('carolina', 5, 314) ('is', 3, 367) ('definition', 9, 318) ('region', 231, 5) ('crashed', 84, 18) ### +### ('entity', 282, 6) ('nationale', 81, 32) ('capital', 518, 0) ('henderson', 34, 93) ### +### ('crashing', 246, 11) ('stumbled', 206, 15) ('−', 251, 21) ('julian', 128, 50) ('beside', 51, 112) ### +### ('ছ', 236, 29) ### +############################################################################################################ +[2023-10-08 00:04:24,903][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:04:24,903][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:04:25,325][root][INFO] - Epoch: 14: Step: 901/1557, loss[v]=0.041635, lr=0.000006, acc@1[1]=246.5/256=0.962890625, acc@1[2]=253.5/256=0.990234375 +[2023-10-08 00:05:42,155][root][INFO] - Train batch 1000 +[2023-10-08 00:05:42,156][root][INFO] - Avg. loss per last 100 batches: 0.059639 +[2023-10-08 00:05:42,833][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29125.1/29522=98.66% | mean: 0.01 | max: 5.63 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.34 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] who plays little simba [SEP] ### +### [P_TEXT]: [CLS] exists and is an alternate of. cub nala is voiced by niketa calame and cub simba is ### +### voiced by jonathan taylor thomas. cub nala is voiced by niketa calame and cub simba is voiced by ### +### jonathan taylor thomas. [SEP] ### +### ======================================= h_v_q | Gates: 27542 ======================================= ### +### ('sim', 0, 1) ('##ba', 1, 8) ('little', 2, 2068) ('played', 3, 132) ('plays', 4, 332) ### +### ('portrayed', 5, 48) ('playing', 6, 880) ('starring', 7, 2063) ('play', 8, 682) ('role', 9, 2620) ### +### ('familiarity', 10, 24536) ('actor', 11, 783) ('.', 12, 15979) ('tiny', 13, 3796) ### +### ('small', 14, 6054) ('drama', 15, 4378) ('stylized', 16, 28180) ('relating', 17, 24482) ### +### ('consisting', 18, 24239) ('ba', 19, 501) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('cub', 14325, 0) ('sim', 0, 1) ('nike', 8542, 2) ('##la', 180, 3) ('na', 2862, 4) ### +### ('voiced', 40, 5) ('ˈ', 150, 6) ('alternate', 202, 7) ('##ba', 1, 8) ('hesitated', 114, 9) ### +### ('crashing', 82, 10) ('cubs', 2293, 11) ('cyrillic', 241, 12) ('##ο', 168, 13) ('##ame', 25848, 14) ### +### ('thomas', 985, 15) ('sharply', 128, 16) ('taylor', 132, 17) ('##ང', 157, 18) ('hating', 152, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('sim', 0, 1) ('##ba', 1, 8) ('portrayed', 5, 48) ('played', 3, 132) ('plays', 4, 332) ### +### ('little', 2, 2068) ('voiced', 40, 5) ('playing', 6, 880) ('play', 8, 682) ('##α', 36, 22) ### +### ('crashed', 47, 21) ('actor', 11, 783) ('−', 57, 29) ('crashing', 82, 10) ('##la', 180, 3) ### +### ('starring', 7, 2063) ('julian', 32, 112) ('who', 25, 164) ('hesitated', 114, 9) ('ˈ', 150, 6) ### +############################################################################################################ +[2023-10-08 00:05:42,833][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:05:42,833][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:05:43,235][root][INFO] - Epoch: 14: Step: 1001/1557, loss[v]=0.067693, lr=0.000006, acc@1[1]=240.5/256=0.939453125, acc@1[2]=251.0/256=0.98046875 +[2023-10-08 00:06:58,796][root][INFO] - Train batch 1100 +[2023-10-08 00:06:58,797][root][INFO] - Avg. loss per last 100 batches: 0.061498 +[2023-10-08 00:06:59,517][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29145.2/29522=98.72% | mean: 0.01 | max: 5.72 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.23 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what are the conditions associated with sleep [SEP] ### +### [P_TEXT]: [CLS] sleep apnea related diseases & conditions. medical conditions are often related to ### +### other diseases and conditions. our doctors have compiled a list of ailments related to the topic of ### +### sleep apnea. these conditions may be a cause or symptom of sleep apnea or be a condition for which ### +### you may be at increased risk. 1 sinus infection ( sinusitis ) sinus infection ( sinusitis ) is ### +### caused by allergies, infection, and chemicals or other irritants of sinuses. signs and symptoms... ### +### learn more a ». [SEP] ### +### ======================================= h_v_q | Gates: 27244 ======================================= ### +### ('sleep', 0, 0) ('conditions', 1, 6) ('associated', 2, 210) ('relating', 3, 6989) ('.', 4, 13675) ### +### ('familiarity', 5, 25593) ('include', 6, 1221) ('are', 7, 7319) ('surrounding', 8, 8276) ### +### ('condition', 9, 33) ('stylized', 10, 28754) ('consisting', 11, 26012) ('sleeping', 12, 34) ### +### ('plural', 13, 17861) ('linked', 14, 273) ('connected', 15, 124) ('united', 16, 21275) ### +### ('correlated', 17, 795) ('mathematics', 18, 24533) ('related', 19, 4) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('sleep', 0, 0) ('diseases', 573, 1) ('ˈ', 161, 2) ('crashing', 32, 3) ('related', 19, 4) ### +### ('hating', 126, 5) ('conditions', 1, 6) ('sharply', 69, 7) ('hesitated', 160, 8) ('»', 18490, 9) ### +### ('ap', 5227, 10) ('−', 24, 11) ('unwilling', 106, 12) ('crashed', 34, 13) ('wingspan', 424, 14) ### +### ('##ο', 118, 15) ('##us', 8732, 16) ('compiled', 7594, 17) ('sin', 9980, 18) ('stumbled', 216, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('sleep', 0, 0) ('conditions', 1, 6) ('associated', 2, 210) ('condition', 9, 33) ### +### ('sleeping', 12, 34) ('related', 19, 4) ('−', 24, 11) ('crashing', 32, 3) ('crashed', 34, 13) ### +### ('ruined', 39, 31) ('sharply', 69, 7) ('connected', 15, 124) ('hating', 126, 5) ('simon', 29, 77) ### +### ('unwilling', 106, 12) ('##₂', 60, 36) ('julian', 27, 85) ('angrily', 68, 32) ('ˈ', 161, 2) ### +### ('##ο', 118, 15) ### +############################################################################################################ +[2023-10-08 00:06:59,518][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:06:59,518][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:06:59,924][root][INFO] - Epoch: 14: Step: 1101/1557, loss[v]=0.061468, lr=0.000006, acc@1[1]=244.5/256=0.955078125, acc@1[2]=250.0/256=0.9765625 +[2023-10-08 00:08:17,009][root][INFO] - Train batch 1200 +[2023-10-08 00:08:17,009][root][INFO] - Avg. loss per last 100 batches: 0.064129 +[2023-10-08 00:08:17,695][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29071.7/29522=98.47% | mean: 0.01 | max: 5.18 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.23 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] dahlia where does the name come from [SEP] ### +### [P_TEXT]: [CLS] dahlia / dah - lia / [ 3 sylls. ] as a girls'name is pronounced dal - yah. it is of ### +### swedish and scandinavian origin, and the meaning of dahlia is valley. also the flower named for ### +### 18th - century swedish botanist anders dahl. kreatif forms : daihlia, dehlia, dohlia. ahlia / dah - ### +### lia / [ 3 sylls. ] as a girls'name is pronounced dal - yah. it is of swedish and scandinavian ### +### origin, and the meaning of dahlia is valley. also the flower named for 18th - century swedish ### +### botanist anders dahl. kreatif forms : daihlia, dehlia, dohlia. [SEP] ### +### ======================================= h_v_q | Gates: 27687 ======================================= ### +### ('dahl', 0, 0) ('##ia', 1, 4) ('name', 2, 35) ('originated', 3, 147) ('comes', 4, 15922) ### +### ('surname', 5, 112) ('familiarity', 6, 25374) ('##ュ', 7, 2415) ('from', 8, 14654) ### +### ('noun', 9, 23714) ('word', 10, 383) ('come', 11, 1273) ('origin', 12, 43) ('.', 13, 7208) ### +### ('born', 14, 3508) ('where', 15, 547) ('originates', 16, 8147) ('relating', 17, 26424) ### +### ('stylized', 18, 18798) ('plural', 19, 13267) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('dahl', 0, 0) ('ya', 2232, 1) ('dal', 13125, 2) ('botanist', 7020, 3) ('##ia', 1, 4) ### +### ('scandinavian', 18172, 5) ('ahl', 18204, 6) ('da', 851, 7) ('swedish', 1746, 8) ('lia', 11176, 9) ### +### ('##hli', 25390, 10) ('named', 46, 11) ('flower', 5271, 12) ('ˈ', 769, 13) ('meaning', 86, 14) ### +### ('sweden', 295, 15) ('valley', 938, 16) ('crashing', 278, 17) ('dai', 13993, 18) ('##ο', 628, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('dahl', 0, 0) ('##ia', 1, 4) ('name', 2, 35) ('originated', 3, 147) ('origin', 12, 43) ### +### ('surname', 5, 112) ('named', 46, 11) ('nickname', 24, 51) ('pronounced', 32, 39) ('names', 63, 23) ### +### ('meaning', 86, 14) ('word', 10, 383) ('whose', 27, 169) ('pseudonym', 51, 93) ### +### ('somewhere', 21, 263) ('where', 15, 547) ('sweden', 295, 15) ('crashing', 278, 17) ### +### ('##α', 206, 25) ('−', 195, 27) ### +############################################################################################################ +[2023-10-08 00:08:17,696][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:08:17,696][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:08:18,116][root][INFO] - Epoch: 14: Step: 1201/1557, loss[v]=0.050467, lr=0.000006, acc@1[1]=246.0/256=0.9609375, acc@1[2]=251.5/256=0.982421875 +[2023-10-08 00:09:34,743][root][INFO] - Train batch 1300 +[2023-10-08 00:09:34,747][root][INFO] - Avg. loss per last 100 batches: 0.063220 +[2023-10-08 00:09:35,421][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29117.2/29522=98.63% | mean: 0.01 | max: 5.07 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.14 | max: 6.12 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] microsoft incident definition [SEP] ### +### [P_TEXT]: [CLS] microsoft's definition of an incident is : a single support issue and the ### +### reasonable effort ( s ) needed to resolve it. a single support issue is a problem that cannot be ### +### broken down into subordinate problems. if a problem consists of subordinate problems, each shall be ### +### considered a separate incident. before microsoft provides support for an incident, you and ### +### microsoft's designated support engineer ( s ) must agree on what the problem is and the parameters ### +### for an acceptable solution. [SEP] ### +### ======================================= h_v_q | Gates: 25312 ======================================= ### +### ('microsoft', 0, 2) ('incident', 1, 0) ('definition', 2, 7) ('defined', 3, 44) ### +### ('relating', 4, 24350) ('refers', 5, 10346) ('noun', 6, 20875) ('event', 7, 97) ('.', 8, 8683) ### +### ('software', 9, 240) ('encyclopedia', 10, 7363) ('##º', 11, 28220) ('something', 12, 3328) ### +### ('or', 13, 16219) ('plural', 14, 10523) ('familiarity', 15, 26281) ('accident', 16, 40) ### +### ('definitions', 17, 3) ('latin', 18, 3671) ('specified', 19, 11173) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('incident', 1, 0) ('subordinate', 1914, 1) ('microsoft', 0, 2) ('definitions', 17, 3) ### +### ('support', 1063, 4) ('problem', 490, 5) ('incidents', 21, 6) ('definition', 2, 7) ### +### ('single', 1610, 8) ('define', 673, 9) ('issue', 173, 10) ('broken', 1404, 11) ('ˈ', 887, 12) ### +### ('separate', 3123, 13) ('crashing', 233, 14) ('problems', 3419, 15) ('##ο', 1035, 16) ### +### ('meaning', 39, 17) ('hating', 1299, 18) ('−', 151, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('incident', 1, 0) ('microsoft', 0, 2) ('definition', 2, 7) ('defined', 3, 44) ### +### ('definitions', 17, 3) ('incidents', 21, 6) ('event', 7, 97) ('accident', 16, 40) ### +### ('meaning', 39, 17) ('software', 9, 240) ('means', 32, 74) ('ibm', 24, 136) ('issue', 173, 10) ### +### ('encompasses', 78, 45) ('crisis', 40, 150) ('−', 151, 19) ('an', 35, 194) ('crashing', 233, 14) ### +### ('problem', 490, 5) ('mean', 155, 30) ### +############################################################################################################ +[2023-10-08 00:09:35,421][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:09:35,421][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:09:35,840][root][INFO] - Epoch: 14: Step: 1301/1557, loss[v]=0.050674, lr=0.000005, acc@1[1]=242.0/256=0.9453125, acc@1[2]=250.5/256=0.978515625 +[2023-10-08 00:10:51,209][root][INFO] - Train batch 1400 +[2023-10-08 00:10:51,211][root][INFO] - Avg. loss per last 100 batches: 0.062269 +[2023-10-08 00:10:51,906][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29112.7/29522=98.61% | mean: 0.01 | max: 5.50 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.14 | max: 6.12 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] is there an addiction gene [SEP] ### +### [P_TEXT]: [CLS] myth no. 1 : there is an addiction gene. there is no single gene, or set of genes, ### +### that determines whether or not a person will become an addict. and even if a person's parents are ### +### addicts, it doesn't mean they will be too. current addiction research shows that roughly 50 % of ### +### addiction tendencies are attributable to genes. [SEP] ### +### ======================================= h_v_q | Gates: 27090 ======================================= ### +### ('addiction', 0, 1) ('gene', 1, 4) ('there', 2, 25) ('.', 3, 11556) ('addicted', 4, 26) ### +### ('genes', 5, 2) ('an', 6, 185) ('familiarity', 7, 25877) ('relating', 8, 21525) ### +### ('network', 9, 3396) ('consisting', 10, 24419) ('genetic', 11, 53) ('nearby', 12, 181) ### +### ('addict', 13, 0) ('is', 14, 1061) ('stylized', 15, 28699) ('##sam', 16, 24807) ### +### ('protein', 17, 1987) ('person', 18, 42) ('ancestry', 19, 343) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('addict', 13, 0) ('addiction', 0, 1) ('genes', 5, 2) ('myth', 11947, 3) ('gene', 1, 4) ### +### ('ˈ', 930, 5) ('##ο', 205, 6) ('parents', 6344, 7) ('−', 33, 8) ('cyrillic', 1137, 9) ### +### ('crashing', 247, 10) ('wingspan', 1319, 11) ('hating', 316, 12) ('doesn', 784, 13) ### +### ('sharply', 182, 14) ('hesitated', 171, 15) ('unwilling', 118, 16) ('whether', 11034, 17) ### +### ('single', 1364, 18) ('##ང', 620, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('addiction', 0, 1) ('gene', 1, 4) ('there', 2, 25) ('genes', 5, 2) ('addicted', 4, 26) ### +### ('addict', 13, 0) ('genetic', 11, 53) ('an', 6, 185) ('−', 33, 8) ('person', 18, 42) ### +### ('nearby', 12, 181) ('mutation', 23, 104) ('unwilling', 118, 16) ('somewhere', 22, 135) ### +### ('julian', 47, 68) ('angrily', 120, 21) ('##ο', 205, 6) ('simon', 45, 81) ('##₂', 125, 24) ### +### ('sharply', 182, 14) ### +############################################################################################################ +[2023-10-08 00:10:51,906][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:10:51,907][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:10:52,311][root][INFO] - Epoch: 14: Step: 1401/1557, loss[v]=0.035907, lr=0.000005, acc@1[1]=240.5/256=0.939453125, acc@1[2]=255.0/256=0.99609375 +[2023-10-08 00:12:09,198][root][INFO] - Train batch 1500 +[2023-10-08 00:12:09,201][root][INFO] - Avg. loss per last 100 batches: 0.058367 +[2023-10-08 00:12:09,921][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29037.4/29522=98.36% | mean: 0.01 | max: 5.32 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.14 | max: 6.09 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] which president lost both the popular vote and the electoral college [SEP] ### +### [P_TEXT]: [CLS] 1 in 1876 samuel tilden won the popular vote but lost the election when rutherford ### +### b. hayes got 185 electoral votes to tildenas 184. 2 in 1888 grover cleveland won the popular vote ### +### but lost the election when benjamin harrison got 233 electoral votes to clevelandas 168. [SEP] ### +### ======================================= h_v_q | Gates: 26779 ======================================= ### +### ('lost', 0, 17) ('president', 1, 1660) ('electoral', 2, 1) ('vote', 3, 4) ('both', 4, 2828) ### +### ('.', 5, 16934) ('lose', 6, 30) ('popular', 7, 13) ('college', 8, 7354) ('familiarity', 9, 25577) ### +### ('losing', 10, 20) ('loss', 11, 56) ('stylized', 12, 29256) ('gone', 13, 74) ('won', 14, 58) ### +### ('relating', 15, 26524) ('simon', 16, 68) ('consisting', 17, 25591) ('plural', 18, 13624) ### +### ('julian', 19, 84) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('grover', 15667, 0) ('electoral', 2, 1) ('hayes', 1512, 2) ('cleveland', 864, 3) ('vote', 3, 4) ### +### ('harrison', 70, 5) ('til', 23127, 6) ('election', 280, 7) ('ˈ', 45, 8) ('rutherford', 10736, 9) ### +### ('samuel', 86, 10) ('##den', 5227, 11) ('##ο', 49, 12) ('popular', 7, 13) ('cyrillic', 118, 14) ### +### ('votes', 21, 15) ('benjamin', 3635, 16) ('lost', 0, 17) ('−', 23, 18) ('stumbled', 93, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('electoral', 2, 1) ('vote', 3, 4) ('lost', 0, 17) ('popular', 7, 13) ('lose', 6, 30) ### +### ('losing', 10, 20) ('loss', 11, 56) ('president', 1, 1660) ('votes', 21, 15) ('won', 14, 58) ### +### ('both', 4, 2828) ('−', 23, 18) ('gone', 13, 74) ('simon', 16, 68) ('ˈ', 45, 8) ('harrison', 70, 5) ### +### ('##α', 24, 38) ('voter', 20, 76) ('julian', 19, 84) ('##ο', 49, 12) ### +############################################################################################################ +[2023-10-08 00:12:09,921][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:12:09,921][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:12:10,367][root][INFO] - Epoch: 14: Step: 1501/1557, loss[v]=0.059145, lr=0.000005, acc@1[1]=243.0/256=0.94921875, acc@1[2]=254.0/256=0.9921875 +[2023-10-08 00:12:54,670][root][INFO] - rank=2; last iteration 1557 +[2023-10-08 00:12:54,673][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 00:12:54,673][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-08 00:12:54,678][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 00:12:54,679][root][INFO] - Epoch finished on 2 +[2023-10-08 00:12:54,694][root][INFO] - rank=1; last iteration 1557 +[2023-10-08 00:12:54,694][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 00:12:54,694][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-08 00:12:54,698][root][INFO] - rank=3; last iteration 1557 +[2023-10-08 00:12:54,698][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 00:12:54,698][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-08 00:12:54,701][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 00:12:54,702][root][INFO] - Epoch finished on 1 +[2023-10-08 00:12:54,705][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 00:12:54,706][root][INFO] - Epoch finished on 3 +[2023-10-08 00:12:54,725][root][INFO] - rank=0; last iteration 1557 +[2023-10-08 00:12:54,726][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 00:12:54,726][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-08 00:12:54,733][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 00:12:54,733][root][INFO] - Epoch finished on 0 +[2023-10-08 00:13:09,631][root][INFO] - Saved checkpoint at ./vdr_14 +[2023-10-08 00:13:09,632][root][INFO] - Av Loss per epoch=0.061729 +[2023-10-08 00:13:09,632][root][INFO] - epoch total (1) correct predictions=379533 +[2023-10-08 00:13:09,632][root][INFO] - epoch total (2) correct predictions=391546 +[2023-10-08 00:13:09,637][root][INFO] - ***** Epoch 15 ***** +[2023-10-08 00:13:09,641][root][INFO] - rank=2; Iteration start +[2023-10-08 00:13:09,641][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 00:13:09,642][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 00:13:09,643][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-08 00:13:32,156][root][INFO] - Saved checkpoint at ./vdr_14 +[2023-10-08 00:13:32,161][root][INFO] - Av Loss per epoch=0.061729 +[2023-10-08 00:13:32,161][root][INFO] - epoch total (1) correct predictions=379533 +[2023-10-08 00:13:32,161][root][INFO] - epoch total (2) correct predictions=391546 +[2023-10-08 00:13:32,161][root][INFO] - Saved checkpoint at ./vdr_14 +[2023-10-08 00:13:32,161][root][INFO] - Av Loss per epoch=0.061729 +[2023-10-08 00:13:32,161][root][INFO] - epoch total (1) correct predictions=379533 +[2023-10-08 00:13:32,161][root][INFO] - epoch total (2) correct predictions=391546 +[2023-10-08 00:13:32,157][root][INFO] - Saved checkpoint at ./vdr_14 +[2023-10-08 00:13:32,163][root][INFO] - Av Loss per epoch=0.061729 +[2023-10-08 00:13:32,163][root][INFO] - epoch total (1) correct predictions=379533 +[2023-10-08 00:13:32,163][root][INFO] - epoch total (2) correct predictions=391546 +[2023-10-08 00:13:32,165][root][INFO] - ***** Epoch 15 ***** +[2023-10-08 00:13:32,167][root][INFO] - ***** Epoch 15 ***** +[2023-10-08 00:13:32,167][root][INFO] - ***** Epoch 15 ***** +[2023-10-08 00:13:32,171][root][INFO] - rank=0; Iteration start +[2023-10-08 00:13:32,171][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 00:13:32,172][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 00:13:32,172][root][INFO] - rank=1; Iteration start +[2023-10-08 00:13:32,173][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 00:13:32,173][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 00:13:32,173][root][INFO] - rank=3; Iteration start +[2023-10-08 00:13:32,173][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 00:13:32,173][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 00:13:32,173][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-08 00:13:32,175][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-08 00:13:32,175][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-08 00:13:33,135][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29107.9/29522=98.60% | mean: 0.01 | max: 5.33 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.14 | max: 6.20 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] are expenses to become a u. s. citizen tax deductible? [SEP] ### +### [P_TEXT]: [CLS] caregiver irs tax rules. to qualify for caregiver tax deductions and credits, the ### +### person you are caring for must be a spouse, dependent, or qualifying relative, as well as a united ### +### states citizen or resident of the u. s., canada, or mexico. [SEP] ### +### ======================================= h_v_q | Gates: 27937 ======================================= ### +### ('citizen', 0, 13) ('expenses', 1, 11106) ('become', 2, 188) ('de', 3, 32) ('tax', 4, 10) ### +### ('becoming', 5, 11278) ('u', 6, 701) ('citizens', 7, 73) ('familiarity', 8, 26092) ('.', 9, 17552) ### +### ('united', 10, 2144) ('states', 11, 92) ('costs', 12, 2619) ('degree', 13, 12706) ### +### ('##du', 14, 29316) ('america', 15, 6093) ('stylized', 16, 29049) ('resident', 17, 16) ### +### ('american', 18, 402) ('$', 19, 1843) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('irs', 24011, 0) ('credits', 17709, 1) ('ˈ', 84, 2) ('qualifying', 4312, 3) ('care', 3347, 4) ### +### ('crashing', 38, 5) ('qualify', 5539, 6) ('caring', 4549, 7) ('mexico', 89, 8) ('−', 39, 9) ### +### ('tax', 4, 10) ('rules', 4838, 11) ('##ο', 171, 12) ('citizen', 0, 13) ('canada', 716, 14) ### +### ('unwilling', 87, 15) ('resident', 17, 16) ('dependent', 11315, 17) ('person', 52, 18) ### +### ('##gi', 16699, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('citizen', 0, 13) ('tax', 4, 10) ('de', 3, 32) ('become', 2, 188) ('citizens', 7, 73) ### +### ('resident', 17, 16) ('states', 11, 92) ('crashing', 38, 5) ('−', 39, 9) ('us', 25, 44) ### +### ('crashed', 35, 26) ('americans', 24, 79) ('person', 52, 18) ('fbi', 26, 88) ('u', 6, 701) ### +### ('ˈ', 84, 2) ('usa', 28, 84) ('##₂', 63, 23) ('mexico', 89, 8) ('unwilling', 87, 15) ### +############################################################################################################ +[2023-10-08 00:13:33,136][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:13:33,136][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:13:33,535][root][INFO] - Epoch: 15: Step: 1/1557, loss[v]=0.042531, lr=0.000005, acc@1[1]=244.0/256=0.953125, acc@1[2]=254.5/256=0.994140625 +[2023-10-08 00:14:49,941][root][INFO] - Train batch 100 +[2023-10-08 00:14:49,945][root][INFO] - Avg. loss per last 100 batches: 0.059925 +[2023-10-08 00:14:50,669][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29160.1/29522=98.77% | mean: 0.01 | max: 5.56 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.29 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] when is best time to plant cabbage [SEP] ### +### [P_TEXT]: [CLS] the best time to plant seeds for growing cabbage is between the months of march and ### +### june, or in the cool fall months of september and october depending on your climate. plant cabbages ### +### seeds or small plants at 3 to 4 week intervals. [SEP] ### +### ======================================= h_v_q | Gates: 27292 ======================================= ### +### ('cabbage', 0, 0) ('plant', 1, 11) ('time', 2, 7) ('best', 3, 31) ('.', 4, 5511) ('2017', 5, 16941) ### +### ('minutes', 6, 29) ('spring', 7, 169) ('hours', 8, 1374) ('days', 9, 298) ('march', 10, 13) ### +### ('early', 11, 207) ('april', 12, 586) ('to', 13, 270) ('plants', 14, 8) ('familiarity', 15, 22140) ### +### ('timing', 16, 183) ('september', 17, 21) ('2016', 18, 10190) ('moment', 19, 467) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('cabbage', 0, 0) ('seeds', 1509, 1) ('depending', 11595, 2) ('fall', 1728, 3) ('weeks', 49, 4) ### +### ('long', 2575, 5) ('growing', 4760, 6) ('time', 2, 7) ('plants', 14, 8) ('june', 43, 9) ### +### ('seed', 1855, 10) ('plant', 1, 11) ('−', 145, 12) ('march', 10, 13) ('unwilling', 154, 14) ### +### ('month', 386, 15) ('abandon', 479, 16) ('##ο', 244, 17) ('crashing', 308, 18) ### +### ('cyrillic', 917, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('cabbage', 0, 0) ('plant', 1, 11) ('time', 2, 7) ('best', 3, 31) ('minutes', 6, 29) ### +### ('march', 10, 13) ('plants', 14, 8) ('september', 17, 21) ('october', 24, 24) ('when', 23, 27) ### +### ('spring', 7, 169) ('weeks', 49, 4) ('june', 43, 9) ('early', 11, 207) ('months', 53, 25) ### +### ('timing', 16, 183) ('july', 21, 137) ('days', 9, 298) ('to', 13, 270) ('better', 29, 129) ### +############################################################################################################ +[2023-10-08 00:14:50,670][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:14:50,670][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:14:51,095][root][INFO] - Epoch: 15: Step: 101/1557, loss[v]=0.053883, lr=0.000005, acc@1[1]=239.0/256=0.93359375, acc@1[2]=250.5/256=0.978515625 +[2023-10-08 00:16:07,134][root][INFO] - Train batch 200 +[2023-10-08 00:16:07,137][root][INFO] - Avg. loss per last 100 batches: 0.061122 +[2023-10-08 00:16:07,862][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29146.5/29522=98.73% | mean: 0.01 | max: 5.54 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.34 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what does don't drop the soap mean [SEP] ### +### [P_TEXT]: [CLS] don't drop the soap! definition. 1. a remark meant to some one being hauled to ### +### prison, specially some body you dislike. once in jail, you imply that if the person ( a male ) ### +### falls a bar of detergent into the bath, they will be forced to bend over and access it. therefore, ### +### with there buttocks spread and in clear picture, should be subject to anal rape by a fellow inmate. ### +### by danelle cuddy report definition. [SEP] ### +### ======================================= h_v_q | Gates: 27556 ======================================= ### +### ('soap', 0, 0) ('drop', 1, 18) ('dropped', 2, 35) ('don', 3, 368) ('drops', 4, 77) ### +### ('relating', 5, 24365) ('means', 6, 81) ('mean', 7, 106) ('not', 8, 648) ('never', 9, 431) ### +### ('t', 10, 327) ('without', 11, 11697) ('.', 12, 9406) ('cannot', 13, 1043) ('dropping', 14, 15) ### +### ('refers', 15, 14408) ('familiarity', 16, 26825) ('noun', 17, 24699) ('##sam', 18, 27051) ### +### ('stylized', 19, 27947) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('soap', 0, 0) ('anal', 15150, 1) ('jail', 8307, 2) ('bath', 213, 3) ('prison', 2421, 4) ### +### ('dane', 4620, 5) ('inmate', 10660, 6) ('butt', 4068, 7) ('definitions', 2825, 8) ### +### ('##ocks', 26889, 9) ('rape', 16721, 10) ('ˈ', 165, 11) ('definition', 23, 12) ('deter', 21750, 13) ### +### ('##ο', 314, 14) ('dropping', 14, 15) ('male', 5847, 16) ('inmates', 13184, 17) ('drop', 1, 18) ### +### ('hauled', 11718, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('soap', 0, 0) ('drop', 1, 18) ('dropped', 2, 35) ('drops', 4, 77) ('means', 6, 81) ### +### ('dropping', 14, 15) ('mean', 7, 106) ('definition', 23, 12) ('don', 3, 368) ('meaning', 24, 40) ### +### ('t', 10, 327) ('never', 9, 431) ('encompasses', 34, 108) ('unwilling', 66, 27) ('not', 8, 648) ### +### ('person', 77, 39) ('crashing', 87, 29) ('bath', 213, 3) ('meant', 69, 67) ('ˈ', 165, 11) ### +############################################################################################################ +[2023-10-08 00:16:07,862][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:16:07,862][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:16:08,271][root][INFO] - Epoch: 15: Step: 201/1557, loss[v]=0.080157, lr=0.000005, acc@1[1]=242.0/256=0.9453125, acc@1[2]=247.0/256=0.96484375 +[2023-10-08 00:17:26,803][root][INFO] - Train batch 300 +[2023-10-08 00:17:26,806][root][INFO] - Avg. loss per last 100 batches: 0.060725 +[2023-10-08 00:17:27,539][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29093.1/29522=98.55% | mean: 0.01 | max: 5.36 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.15 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how long lasagna in oven to cook [SEP] ### +### [P_TEXT]: [CLS] bake for 30 - 40 minutes at 375a, covered with aluminum foil. put a layer of ### +### aluminum over the lasagna before it goes into the oven. to prevent the sauce from spilling over the ### +### edge as it heats up, you can also place the entire dish on a baking tray to keep the sauce from ### +### getting on your oven. [SEP] ### +### ======================================= h_v_q | Gates: 27119 ======================================= ### +### ('##ag', 0, 5) ('oven', 1, 6) ('las', 2, 16) ('minutes', 3, 3) ('##na', 4, 27) ('days', 5, 1596) ### +### ('cook', 6, 94) ('.', 7, 15124) ('weeks', 8, 45) ('hours', 9, 2927) ('months', 10, 14880) ### +### ('familiarity', 11, 25906) ('years', 12, 3504) ('stylized', 13, 28750) ('##acker', 14, 150) ### +### ('approximately', 15, 7413) ('consisting', 16, 23102) ('miles', 17, 6080) ('relating', 18, 28026) ### +### ('to', 19, 4625) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('sauce', 2653, 0) ('aluminum', 3034, 1) ('foil', 14881, 2) ('minutes', 3, 3) ('30', 34, 4) ### +### ('##ag', 0, 5) ('oven', 1, 6) ('ˈ', 94, 7) ('heats', 7055, 8) ('hating', 93, 9) ('tray', 9876, 10) ### +### ('unwilling', 43, 11) ('##ο', 101, 12) ('baking', 42, 13) ('long', 22, 14) ('−', 50, 15) ### +### ('las', 2, 16) ('crashing', 86, 17) ('hesitated', 84, 18) ('##₂', 100, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##ag', 0, 5) ('oven', 1, 6) ('las', 2, 16) ('minutes', 3, 3) ('##na', 4, 27) ('weeks', 8, 45) ### +### ('cook', 6, 94) ('long', 22, 14) ('30', 34, 4) ('simon', 25, 32) ('days', 5, 1596) ### +### ('##acker', 14, 150) ('##α', 33, 21) ('onto', 26, 54) ('minute', 20, 93) ('baking', 42, 13) ### +### ('unwilling', 43, 11) ('−', 50, 15) ('##大', 44, 34) ('ˈ', 94, 7) ### +############################################################################################################ +[2023-10-08 00:17:27,539][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:17:27,539][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:17:27,962][root][INFO] - Epoch: 15: Step: 301/1557, loss[v]=0.050373, lr=0.000005, acc@1[1]=245.5/256=0.958984375, acc@1[2]=252.5/256=0.986328125 +[2023-10-08 00:18:44,689][root][INFO] - Train batch 400 +[2023-10-08 00:18:44,694][root][INFO] - Avg. loss per last 100 batches: 0.061300 +[2023-10-08 00:18:45,381][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29036.8/29522=98.36% | mean: 0.01 | max: 5.57 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.22 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] meaning of forgery [SEP] ### +### [P_TEXT]: [CLS] forgery meaning, definition, what is forgery : an illegal copy of a document, ### +### painting, etc. or the crime of making such illegal copies :. learn more. [SEP] ### +### ======================================= h_v_q | Gates: 25977 ======================================= ### +### ('forge', 0, 0) ('##ry', 1, 3) ('noun', 2, 18081) ('definition', 3, 8) ('meaning', 4, 5) ### +### ('forged', 5, 12) ('something', 6, 4032) ('relating', 7, 25341) ('plural', 8, 14015) ### +### ('familiarity', 9, 27886) ('stylized', 10, 27007) ('refers', 11, 6718) ('specified', 12, 18166) ### +### ('consisting', 13, 15912) ('latin', 14, 1750) ('##º', 15, 23485) ('or', 16, 20567) ### +### ('manner', 17, 16208) (';', 18, 4727) ('means', 19, 17) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('forge', 0, 0) ('illegal', 3475, 1) ('definitions', 98, 2) ('##ry', 1, 3) ('define', 5358, 4) ### +### ('meaning', 4, 5) ('ˈ', 78, 6) ('painting', 1880, 7) ('definition', 3, 8) ('copies', 13348, 9) ### +### ('documents', 1221, 10) ('##ο', 416, 11) ('forged', 5, 12) ('crashing', 67, 13) ('−', 90, 14) ### +### ('document', 505, 15) ('stumbled', 155, 16) ('means', 19, 17) ('wingspan', 1643, 18) ### +### ('unwilling', 436, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('forge', 0, 0) ('##ry', 1, 3) ('meaning', 4, 5) ('definition', 3, 8) ('forged', 5, 12) ### +### ('means', 19, 17) ('defined', 20, 21) ('ˈ', 78, 6) ('definitions', 98, 2) ('crashing', 67, 13) ### +### ('meanings', 53, 22) ('crashed', 54, 34) ('−', 90, 14) ('mean', 57, 36) ('##α', 72, 27) ### +### ('ছ', 86, 38) ('##₂', 104, 30) ('treason', 105, 33) ('stumbled', 155, 16) ('angrily', 133, 32) ### +############################################################################################################ +[2023-10-08 00:18:45,382][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:18:45,382][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:18:45,786][root][INFO] - Epoch: 15: Step: 401/1557, loss[v]=0.037344, lr=0.000005, acc@1[1]=243.5/256=0.951171875, acc@1[2]=255.5/256=0.998046875 +[2023-10-08 00:20:03,151][root][INFO] - Train batch 500 +[2023-10-08 00:20:03,155][root][INFO] - Avg. loss per last 100 batches: 0.060044 +[2023-10-08 00:20:03,846][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29053.1/29522=98.41% | mean: 0.01 | max: 5.55 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.32 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is an etiolated plant [SEP] ### +### [P_TEXT]: [CLS] answer wiki. an etiolated plant is a plant that grows in low light conditions. this ### +### results in weaker and elongated stems ( usually curving and stretching in search of light ) and ### +### yellowing. as far as a moderately etiolated plant has adapted in its environment it can continue ### +### growing. [SEP] ### +### ======================================= h_v_q | Gates: 27366 ======================================= ### +### ('##ated', 0, 10) ('et', 1, 0) ('##iol', 2, 6) ('plant', 3, 1) ('definition', 4, 54) ### +### ('familiarity', 5, 23862) ('relating', 6, 24504) ('##sam', 7, 26310) ('is', 8, 802) ### +### ('refers', 9, 15638) ('stylized', 10, 28985) ('plants', 11, 3) ('encompasses', 12, 15) ### +### ('plural', 13, 15883) ('consisting', 14, 23591) ('encyclopedia', 15, 15517) ('term', 16, 9616) ### +### ('noun', 17, 22018) ('.', 18, 15494) ('an', 19, 230) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('et', 1, 0) ('plant', 3, 1) ('ˈ', 183, 2) ('plants', 11, 3) ('crashing', 30, 4) ### +### ('stems', 15436, 5) ('##iol', 2, 6) ('hating', 80, 7) ('##ο', 152, 8) ('stretching', 8928, 9) ### +### ('##ated', 0, 10) ('##ང', 204, 11) ('elongated', 3419, 12) ('sharply', 54, 13) ('−', 37, 14) ### +### ('encompasses', 12, 15) ('growing', 9252, 16) ('moderately', 25888, 17) ('##₂', 68, 18) ### +### ('low', 6392, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('et', 1, 0) ('##iol', 2, 6) ('##ated', 0, 10) ('plant', 3, 1) ('plants', 11, 3) ### +### ('definition', 4, 54) ('encompasses', 12, 15) ('crashing', 30, 4) ('−', 37, 14) ('crashed', 35, 20) ### +### ('unwilling', 40, 22) ('sharply', 54, 13) ('##α', 42, 39) ('angrily', 45, 32) ('ছ', 51, 23) ### +### ('stumbled', 50, 26) ('hating', 80, 7) ('##₂', 68, 18) ('an', 19, 230) ('julian', 39, 76) ### +############################################################################################################ +[2023-10-08 00:20:03,847][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:20:03,847][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:20:04,270][root][INFO] - Epoch: 15: Step: 501/1557, loss[v]=0.061468, lr=0.000005, acc@1[1]=243.0/256=0.94921875, acc@1[2]=252.0/256=0.984375 +[2023-10-08 00:21:21,477][root][INFO] - Train batch 600 +[2023-10-08 00:21:21,478][root][INFO] - Avg. loss per last 100 batches: 0.060589 +[2023-10-08 00:21:22,199][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29096.6/29522=98.56% | mean: 0.01 | max: 5.34 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.15 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what does ching ching mean in japanese [SEP] ### +### [P_TEXT]: [CLS] they are so many words in chinese with the ching ( qing in pinyin ) and chong ( ### +### chong in pinyin ) that i cannot possibly tell you what they could mean! that only one word chong ### +### means something or some words that are repeated and unnecessary. and ching means celebrating. [SEP] ### +### ======================================= h_v_q | Gates: 25971 ======================================= ### +### ('ching', 0, 0) ('japanese', 1, 226) ('means', 2, 10) ('.', 3, 9692) ('relating', 4, 23535) ### +### ('noun', 5, 11614) ('japan', 6, 3992) ('meaning', 7, 15) ('definition', 8, 150) ('spanish', 9, 253) ### +### ('mean', 10, 26) ('familiarity', 11, 25953) ('is', 12, 3038) ('refers', 13, 5280) ### +### ('symbol', 14, 148) ('sense', 15, 5501) ('chinese', 16, 1) ('stylized', 17, 27072) ### +### ('something', 18, 147) ('german', 19, 454) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('ching', 0, 0) ('chinese', 16, 1) ('pinyin', 3523, 2) ('chong', 18110, 3) ('words', 94, 4) ### +### ('qing', 1145, 5) ('unnecessary', 8899, 6) ('ˈ', 293, 7) ('word', 58, 8) ('crashing', 162, 9) ### +### ('means', 2, 10) ('##ο', 292, 11) ('celebrating', 15738, 12) ('hesitated', 685, 13) ('−', 65, 14) ### +### ('meaning', 7, 15) ('unwilling', 167, 16) ('wingspan', 468, 17) ('sharply', 214, 18) ### +### ('china', 272, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ching', 0, 0) ('means', 2, 10) ('japanese', 1, 226) ('meaning', 7, 15) ('chinese', 16, 1) ### +### ('mean', 10, 26) ('definition', 8, 150) ('symbol', 14, 148) ('spanish', 9, 253) ('word', 58, 8) ### +### ('something', 18, 147) ('meant', 40, 39) ('words', 94, 4) ('language', 21, 143) ('−', 65, 14) ### +### ('crashed', 68, 24) ('italian', 25, 221) ('crashing', 162, 9) ('german', 19, 454) ('##α', 91, 46) ### +############################################################################################################ +[2023-10-08 00:21:22,199][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:21:22,199][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:21:22,624][root][INFO] - Epoch: 15: Step: 601/1557, loss[v]=0.082554, lr=0.000005, acc@1[1]=243.5/256=0.951171875, acc@1[2]=250.5/256=0.978515625 +[2023-10-08 00:22:40,066][root][INFO] - Train batch 700 +[2023-10-08 00:22:40,071][root][INFO] - Avg. loss per last 100 batches: 0.059932 +[2023-10-08 00:22:40,759][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29068.3/29522=98.46% | mean: 0.01 | max: 5.46 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.37 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what's a domestic business corporation [SEP] ### +### [P_TEXT]: [CLS] a domestic corporation is a corporation that does business within the country where ### +### it was established, headquartered or based. for example, if a company opens in the united states, ### +### then within the united states that particular company is considered a domestic company. [SEP] ### +### ======================================= h_v_q | Gates: 26365 ======================================= ### +### ('domestic', 0, 0) ('corporation', 1, 1) ('business', 2, 18) ('relating', 3, 19200) ### +### ('encompasses', 4, 10) ('plural', 5, 9037) ('is', 6, 207) ('definition', 7, 6) ('noun', 8, 14877) ### +### ('##sam', 9, 24750) ('stylized', 10, 26815) ('encyclopedia', 11, 3710) ('refers', 12, 7406) ### +### ('organization', 13, 94) ('familiarity', 14, 22486) ('businesses', 15, 180) ('company', 16, 2) ### +### ('consisting', 17, 23064) ('corporations', 18, 7) ('designed', 19, 14395) ### +### ======================================= h_v_p | Gates: 29520 ======================================= ### +### ('domestic', 0, 0) ('corporation', 1, 1) ('company', 16, 2) ('define', 6672, 3) ('ˈ', 272, 4) ### +### ('crashing', 73, 5) ('definition', 7, 6) ('corporations', 18, 7) ('headquartered', 135, 8) ### +### ('##ο', 471, 9) ('encompasses', 4, 10) ('unwilling', 263, 11) ('corp', 79, 12) ('−', 124, 13) ### +### ('country', 120, 14) ('hating', 230, 15) ('meaning', 507, 16) ('stumbled', 86, 17) ### +### ('business', 2, 18) ('##₂', 163, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('domestic', 0, 0) ('corporation', 1, 1) ('business', 2, 18) ('encompasses', 4, 10) ### +### ('definition', 7, 6) ('company', 16, 2) ('corporations', 18, 7) ('organization', 13, 94) ### +### ('is', 6, 207) ('crashing', 73, 5) ('businesses', 15, 180) ('crashed', 59, 23) ('corp', 79, 12) ### +### ('stumbled', 86, 17) ('national', 22, 165) ('a', 24, 146) ('−', 124, 13) ('headquartered', 135, 8) ### +### ('country', 120, 14) ('defined', 34, 103) ### +############################################################################################################ +[2023-10-08 00:22:40,759][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:22:40,760][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:22:41,183][root][INFO] - Epoch: 15: Step: 701/1557, loss[v]=0.066294, lr=0.000005, acc@1[1]=239.5/256=0.935546875, acc@1[2]=250.0/256=0.9765625 +[2023-10-08 00:23:58,184][root][INFO] - Train batch 800 +[2023-10-08 00:23:58,187][root][INFO] - Avg. loss per last 100 batches: 0.057092 +[2023-10-08 00:23:58,873][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29130.5/29522=98.67% | mean: 0.01 | max: 5.03 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.6/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.16 | max: 6.04 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] effect of acetylcholine on behavior [SEP] ### +### [P_TEXT]: [CLS] acetylcholine is also used as a neurotransmitter in the autonomic nervous system, ### +### both as an internal transmitter for the sympathetic nervous system and as the final product ### +### released by the parasympathetic nervous system. heir effect on target cells is usually inhibitory. ### +### muscarinic acetylcholine receptors are found in both the central nervous system and the peripheral ### +### nervous system of the heart, lungs, upper gastrointestinal tract, and sweat glands. [SEP] ### +### ======================================= h_v_q | Gates: 27952 ======================================= ### +### ('##lch', 0, 50) ('ace', 1, 6) ('##ty', 2, 151) ('behavior', 3, 6327) ('##olin', 4, 1) ### +### ('##e', 5, 60) ('.', 6, 21297) ('effect', 7, 11) ('familiarity', 8, 27786) ('stylized', 9, 28308) ### +### ('effects', 10, 76) ('behavioral', 11, 7428) ('consisting', 12, 24111) ('relating', 13, 28367) ### +### ('−', 14, 17) ('onto', 15, 436) ('plural', 16, 11628) ('##α', 17, 75) ('affect', 18, 152) ### +### ('ability', 19, 4405) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('heir', 7678, 0) ('##olin', 4, 1) ('ˈ', 41, 2) ('sweat', 5302, 3) ('sympathetic', 9968, 4) ### +### ('receptors', 1213, 5) ('ace', 1, 6) ('lungs', 19311, 7) ('inhibitor', 2843, 8) ('crashing', 22, 9) ### +### ('transmitter', 153, 10) ('effect', 7, 11) ('hating', 57, 12) ('nervous', 3820, 13) ('##ο', 55, 14) ### +### ('unwilling', 31, 15) ('##rini', 21192, 16) ('−', 14, 17) ('receptor', 593, 18) ### +### ('hesitated', 32, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ace', 1, 6) ('##olin', 4, 1) ('##lch', 0, 50) ('effect', 7, 11) ('##ty', 2, 151) ('##e', 5, 60) ### +### ('effects', 10, 76) ('−', 14, 17) ('crashing', 22, 9) ('crashed', 21, 38) ('ˈ', 41, 2) ### +### ('simon', 20, 48) ('unwilling', 31, 15) ('##α', 17, 75) ('hesitated', 32, 19) ('angrily', 27, 39) ### +### ('##₂', 33, 32) ('behavior', 3, 6327) ('hugh', 38, 29) ('ruined', 35, 37) ### +############################################################################################################ +[2023-10-08 00:23:58,873][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:23:58,873][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:23:59,274][root][INFO] - Epoch: 15: Step: 801/1557, loss[v]=0.077011, lr=0.000005, acc@1[1]=241.0/256=0.94140625, acc@1[2]=250.0/256=0.9765625 +[2023-10-08 00:25:15,870][root][INFO] - Train batch 900 +[2023-10-08 00:25:15,873][root][INFO] - Avg. loss per last 100 batches: 0.059818 +[2023-10-08 00:25:16,589][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29103.7/29522=98.58% | mean: 0.01 | max: 5.64 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.16 | max: 6.26 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what industry is washington county eoc ny [SEP] ### +### [P_TEXT]: [CLS] view tracias full profile. program director for head start / early head ### +### startwashington county eoc, inc. washington county economic opportunity council, inc. is the ### +### community action agency for washington county. it is also the grantee agency for the head start ### +### program. [SEP] ### +### ======================================= h_v_q | Gates: 27479 ======================================= ### +### ('washington', 0, 1) ('county', 1, 2) ('industry', 2, 1298) ('e', 3, 12) ('##oc', 4, 11) ### +### ('industries', 5, 1092) ('york', 6, 878) ('business', 7, 4852) ('industrial', 8, 2553) ### +### ('agriculture', 9, 1555) ('ny', 10, 3076) ('businesses', 11, 8801) ('commercial', 12, 1682) ### +### ('technology', 13, 5050) ('market', 14, 1873) ('economy', 15, 203) ('stylized', 16, 28311) ### +### ('activities', 17, 2456) ('is', 18, 580) ('electronic', 19, 2026) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('head', 1899, 0) ('washington', 0, 1) ('county', 1, 2) ('##ington', 15958, 3) ('council', 3130, 4) ### +### ('agency', 766, 5) ('director', 2816, 6) ('heads', 12530, 7) ('grant', 1198, 8) ('ˈ', 50, 9) ### +### ('##wash', 27257, 10) ('##oc', 4, 11) ('e', 3, 12) ('action', 1456, 13) ('##ο', 317, 14) ### +### ('program', 830, 15) ('start', 11207, 16) ('crashing', 44, 17) ('programs', 2730, 18) ### +### ('counties', 77, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('washington', 0, 1) ('county', 1, 2) ('##oc', 4, 11) ('e', 3, 12) ('economic', 20, 22) ### +### ('industry', 2, 1298) ('industries', 5, 1092) ('york', 6, 878) ('ˈ', 50, 9) ('crashing', 44, 17) ### +### ('economy', 15, 203) ('encompasses', 59, 31) ('simon', 49, 47) ('unwilling', 65, 26) ### +### ('counties', 77, 19) ('crashed', 58, 39) ('−', 79, 25) ('##α', 80, 38) ('opportunity', 147, 21) ### +### ('ছ', 116, 34) ### +############################################################################################################ +[2023-10-08 00:25:16,590][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:25:16,590][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:25:17,014][root][INFO] - Epoch: 15: Step: 901/1557, loss[v]=0.069258, lr=0.000005, acc@1[1]=239.0/256=0.93359375, acc@1[2]=252.0/256=0.984375 +[2023-10-08 00:26:34,346][root][INFO] - Train batch 1000 +[2023-10-08 00:26:34,349][root][INFO] - Avg. loss per last 100 batches: 0.059372 +[2023-10-08 00:26:35,045][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29127.5/29522=98.66% | mean: 0.01 | max: 5.28 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.16 | max: 6.31 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what county is haledon nj [SEP] ### +### [P_TEXT]: [CLS] haledon, new jersey. homefacts city report. haledon is located in passaic county, ### +### nj. the population is 8, 375. there are 2 public schools in haledon with an average homefacts ### +### rating of c -. the total crime rate for haledon is very low, and there are 2 registered sex ### +### offenders residing in the city. [SEP] ### +### ======================================= h_v_q | Gates: 27405 ======================================= ### +### ('hale', 0, 0) ('##don', 1, 1) ('county', 2, 14) ('jersey', 3, 3) ('familiarity', 4, 26660) ### +### ('stylized', 5, 28786) ('relating', 6, 27202) ('consisting', 7, 25850) ('plural', 8, 16827) ### +### ('is', 9, 1919) ('nj', 10, 2) ('encompasses', 11, 64) ('mathematics', 12, 25693) ### +### ('counties', 13, 34) ('province', 14, 2588) ('.', 15, 14161) ('hesitated', 16, 15) ### +### ('phillies', 17, 100) ('division', 18, 4039) ('delaware', 19, 116) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('hale', 0, 0) ('##don', 1, 1) ('nj', 10, 2) ('jersey', 3, 3) ('crime', 811, 4) ('ˈ', 32, 5) ### +### ('##cts', 22244, 6) ('population', 459, 7) ('crashing', 24, 8) ('##ο', 105, 9) ### +### ('schools', 5881, 10) ('unwilling', 41, 11) ('stumbled', 115, 12) ('300', 1374, 13) ### +### ('county', 2, 14) ('hesitated', 16, 15) ('crashed', 36, 16) ('pass', 10899, 17) ### +### ('public', 1463, 18) ('−', 30, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('hale', 0, 0) ('##don', 1, 1) ('jersey', 3, 3) ('county', 2, 14) ('nj', 10, 2) ### +### ('encompasses', 11, 64) ('counties', 13, 34) ('ˈ', 32, 5) ('crashing', 24, 8) ('hesitated', 16, 15) ### +### ('unwilling', 41, 11) ('−', 30, 19) ('crashed', 36, 16) ('simon', 23, 41) ('ছ', 46, 20) ### +### ('##₂', 21, 62) ('stark', 54, 24) ('##α', 31, 49) ('##ο', 105, 9) ('angrily', 44, 43) ### +############################################################################################################ +[2023-10-08 00:26:35,045][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:26:35,045][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:26:35,438][root][INFO] - Epoch: 15: Step: 1001/1557, loss[v]=0.048359, lr=0.000005, acc@1[1]=247.0/256=0.96484375, acc@1[2]=252.0/256=0.984375 +[2023-10-08 00:27:52,480][root][INFO] - Train batch 1100 +[2023-10-08 00:27:52,481][root][INFO] - Avg. loss per last 100 batches: 0.059217 +[2023-10-08 00:27:53,189][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29078.7/29522=98.50% | mean: 0.01 | max: 5.31 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.37 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is always true about a isosceles triangle [SEP] ### +### [P_TEXT]: [CLS] an isosceles triangle has two sides of equal length. an isosceles triangle also has ### +### two angles of the same measure ; namely, the angles opposite to the two sides of the same length ; ### +### this fact is the content of the isosceles triangle theorem, which was known by euclid. [SEP] ### +### ======================================= h_v_q | Gates: 27519 ======================================= ### +### ('iso', 0, 1) ('triangle', 1, 0) ('true', 2, 4888) ('##eles', 3, 4) ('always', 4, 7441) ### +### ('.', 5, 13400) ('concerning', 6, 434) ('##sc', 7, 52) ('familiarity', 8, 24127) ### +### ('relating', 9, 23702) ('stylized', 10, 28254) ('truth', 11, 284) ('##ele', 12, 365) ### +### ('consisting', 13, 23696) ('governing', 14, 13565) ('everywhere', 15, 2310) ### +### ('surrounding', 16, 12184) ('never', 17, 6351) ('plural', 18, 11480) ('encyclopedia', 19, 10154) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('triangle', 1, 0) ('iso', 0, 1) ('sides', 14522, 2) ('theorem', 168, 3) ('##eles', 3, 4) ### +### ('angles', 13049, 5) ('ˈ', 59, 6) ('length', 6886, 7) ('crashing', 33, 8) ('wingspan', 275, 9) ### +### ('##ο', 73, 10) ('stumbled', 133, 11) ('hesitated', 75, 12) ('unwilling', 47, 13) ### +### ('measure', 8579, 14) ('hating', 92, 15) ('−', 44, 16) ('sharply', 56, 17) ('cyrillic', 158, 18) ### +### ('equal', 4811, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('iso', 0, 1) ('triangle', 1, 0) ('##eles', 3, 4) ('##sc', 7, 52) ('crashing', 33, 8) ### +### ('concerning', 6, 434) ('true', 2, 4888) ('ˈ', 59, 6) ('##₂', 36, 21) ('angrily', 30, 25) ### +### ('unwilling', 47, 13) ('simon', 22, 50) ('−', 44, 16) ('##α', 29, 37) ('crashed', 41, 28) ### +### ('##ο', 73, 10) ('sharply', 56, 17) ('truth', 11, 284) ('hesitated', 75, 12) ('ruined', 49, 30) ### +############################################################################################################ +[2023-10-08 00:27:53,189][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:27:53,189][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:27:53,613][root][INFO] - Epoch: 15: Step: 1101/1557, loss[v]=0.049407, lr=0.000005, acc@1[1]=247.5/256=0.966796875, acc@1[2]=250.5/256=0.978515625 +[2023-10-08 00:29:11,093][root][INFO] - Train batch 1200 +[2023-10-08 00:29:11,094][root][INFO] - Avg. loss per last 100 batches: 0.059571 +[2023-10-08 00:29:11,798][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29059.5/29522=98.43% | mean: 0.01 | max: 5.52 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.20 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the dimensions of a king size comforter [SEP] ### +### [P_TEXT]: [CLS] eastern or dual king beds have mattresses measuring 78 inches wide and 80 inches ### +### long. a comforter to fit these king beds will generally be 102 inches in width, and range in length ### +### from 86 to 94 inches. [SEP] ### +### ======================================= h_v_q | Gates: 26771 ======================================= ### +### ('king', 0, 3) ('comfort', 1, 13) ('size', 2, 6) ('dimensions', 3, 23) ('##er', 4, 54) ### +### ('inches', 5, 49) ('diameter', 6, 38) ('kings', 7, 24) ('.', 8, 7086) ('familiarity', 9, 20729) ### +### ('stylized', 10, 29242) ('plural', 11, 2488) ('is', 12, 3998) ('prince', 13, 106) ### +### ('volume', 14, 2739) ('sizes', 15, 76) ('800', 16, 1066) ('relating', 17, 27123) ### +### ('encyclopedia', 18, 8986) ('=', 19, 17880) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('mattress', 4851, 0) ('dual', 11103, 1) ('beds', 8692, 2) ('king', 0, 3) ('bed', 199, 4) ### +### ('length', 24, 5) ('size', 2, 6) ('width', 57, 7) ('eastern', 1266, 8) ('wide', 64, 9) ### +### ('ˈ', 146, 10) ('80', 1318, 11) ('fit', 4718, 12) ('comfort', 1, 13) ('hesitated', 121, 14) ### +### ('hating', 115, 15) ('bunk', 8994, 16) ('##ο', 164, 17) ('big', 685, 18) ('long', 2288, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('king', 0, 3) ('comfort', 1, 13) ('size', 2, 6) ('dimensions', 3, 23) ('##er', 4, 54) ### +### ('inches', 5, 49) ('diameter', 6, 38) ('kings', 7, 24) ('length', 24, 5) ('sizes', 15, 76) ### +### ('prince', 13, 106) ('width', 57, 7) ('wide', 64, 9) ('height', 45, 45) ('unwilling', 61, 29) ### +### ('hesitated', 121, 14) ('hating', 115, 15) ('dimension', 31, 103) ('ˈ', 146, 10) ('bed', 199, 4) ### +############################################################################################################ +[2023-10-08 00:29:11,798][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:29:11,798][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:29:12,221][root][INFO] - Epoch: 15: Step: 1201/1557, loss[v]=0.050880, lr=0.000004, acc@1[1]=246.0/256=0.9609375, acc@1[2]=252.0/256=0.984375 +[2023-10-08 00:30:28,873][root][INFO] - Train batch 1300 +[2023-10-08 00:30:28,874][root][INFO] - Avg. loss per last 100 batches: 0.059336 +[2023-10-08 00:30:29,587][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29066.4/29522=98.46% | mean: 0.01 | max: 5.39 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.24 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] maryland was founded as a colony for _ _ _ _ _. [SEP] ### +### [P_TEXT]: [CLS] interesting maryland colony facts : the maryland colony's original name was the ### +### province of maryland. the maryland colony was founded as a refuge for english catholics. although ### +### the charter had been originally issued to george calvert, 1st baron baltimore, he died before it ### +### was formally executed and his son cecil calvert, 2nd baron baltimore was granted the charter. [SEP] ### +### ======================================= h_v_q | Gates: 27012 ======================================= ### +### ('maryland', 0, 0) ('colony', 1, 4) ('founded', 2, 21) ('established', 3, 142) ('was', 4, 129) ### +### ('founder', 5, 253) ('.', 6, 10528) ('familiarity', 7, 27825) ('opened', 8, 9023) ### +### ('stylized', 9, 26955) ('began', 10, 15217) ('knew', 11, 104) ('were', 12, 255) ### +### ('launched', 13, 4939) ('relating', 14, 27759) ('for', 15, 11165) ('md', 16, 11) ### +### ('became', 17, 1295) ('pennsylvania', 18, 731) ('consisting', 19, 25113) ### +### ======================================= h_v_p | Gates: 29520 ======================================= ### +### ('maryland', 0, 0) ('calvert', 303, 1) ('baltimore', 35, 2) ('cecil', 1936, 3) ('colony', 1, 4) ### +### ('charter', 859, 5) ('province', 719, 6) ('refuge', 7295, 7) ('ˈ', 91, 8) ('facts', 6266, 9) ### +### ('##ο', 172, 10) ('md', 16, 11) ('crashing', 94, 12) ('issued', 1086, 13) ('catholics', 17021, 14) ### +### ('−', 43, 15) ('stumbled', 109, 16) ('hesitated', 229, 17) ('##大', 121, 18) ('colonies', 22, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('maryland', 0, 0) ('colony', 1, 4) ('founded', 2, 21) ('established', 3, 142) ('was', 4, 129) ### +### ('md', 16, 11) ('baltimore', 35, 2) ('colonies', 22, 19) ('founder', 5, 253) ('knew', 11, 104) ### +### ('−', 43, 15) ('ˈ', 91, 8) ('calvert', 303, 1) ('ছ', 56, 34) ('crashing', 94, 12) ('were', 12, 255) ### +### ('##α', 57, 38) ('remained', 20, 126) ('##₂', 66, 31) ('stumbled', 109, 16) ### +############################################################################################################ +[2023-10-08 00:30:29,587][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:30:29,587][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:30:30,019][root][INFO] - Epoch: 15: Step: 1301/1557, loss[v]=0.064678, lr=0.000004, acc@1[1]=240.5/256=0.939453125, acc@1[2]=251.0/256=0.98046875 +[2023-10-08 00:31:47,538][root][INFO] - Train batch 1400 +[2023-10-08 00:31:47,538][root][INFO] - Avg. loss per last 100 batches: 0.058799 +[2023-10-08 00:31:48,238][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29146.6/29522=98.73% | mean: 0.01 | max: 5.43 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.25 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is meta mean on reddit [SEP] ### +### [P_TEXT]: [CLS] a term, especially in art, used to characterize something that is ### +### characteristically self - referential. dude, that's so meta. a term, especially in art, used to ### +### characterize something that is characteristically self - referential. dude, that's so meta. when ### +### you create new layers of abstraction between the thing or event, you are becoming more meta. for ### +### example : a footnote that is needed to explain another footnote is meta. [SEP] ### +### ======================================= h_v_q | Gates: 27401 ======================================= ### +### ('meta', 0, 1) ('##dit', 1, 27260) ('red', 2, 5147) ('noun', 3, 8910) ('mean', 4, 27) ### +### ('means', 5, 33) ('definition', 6, 9) ('meaning', 7, 14) ('is', 8, 942) ('familiarity', 9, 26050) ### +### ('stylized', 10, 26227) ('##º', 11, 26705) ('relating', 12, 22640) ('something', 13, 296) ### +### ('refers', 14, 679) ('consisting', 15, 24624) ('plural', 16, 9664) ('encompasses', 17, 37) ### +### ('sense', 18, 11219) ('language', 19, 3148) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('dude', 6346, 0) ('meta', 0, 1) ('abstraction', 7944, 2) ('layers', 17755, 3) ('art', 445, 4) ### +### ('define', 4909, 5) ('character', 148, 6) ('##ential', 24739, 7) ('characteristic', 1991, 8) ### +### ('definition', 6, 9) ('##ally', 12727, 10) ('##note', 21119, 11) ('##ize', 21835, 12) ('ˈ', 68, 13) ### +### ('meaning', 7, 14) ('definitions', 336, 15) ('term', 24, 16) ('crashing', 43, 17) ### +### ('synonym', 462, 18) ('so', 12690, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('meta', 0, 1) ('definition', 6, 9) ('meaning', 7, 14) ('mean', 4, 27) ('means', 5, 33) ### +### ('encompasses', 17, 37) ('term', 24, 16) ('crashing', 43, 17) ('−', 33, 24) ('something', 13, 296) ### +### ('ˈ', 68, 13) ('crashed', 32, 52) ('word', 42, 43) ('is', 8, 942) ('defined', 38, 55) ### +### ('##₂', 58, 28) ('red', 2, 5147) ('hating', 64, 31) ('##α', 51, 54) ('character', 148, 6) ### +############################################################################################################ +[2023-10-08 00:31:48,238][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:31:48,238][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:31:48,663][root][INFO] - Epoch: 15: Step: 1401/1557, loss[v]=0.054776, lr=0.000004, acc@1[1]=246.0/256=0.9609375, acc@1[2]=252.5/256=0.986328125 +[2023-10-08 00:33:05,316][root][INFO] - Train batch 1500 +[2023-10-08 00:33:05,317][root][INFO] - Avg. loss per last 100 batches: 0.058706 +[2023-10-08 00:33:06,048][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29093.5/29522=98.55% | mean: 0.01 | max: 5.41 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.9/29522=100.00% | mean: 0.15 | max: 6.45 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is false doctrine [SEP] ### +### [P_TEXT]: [CLS] what is false doctrine? answer : doctrine is aa set of ideas or beliefs that are ### +### taught or believed to be true. a biblical doctrine refers to teachings that align with the revealed ### +### word of god, the bible. false doctrine is any idea that adds to, takes away from, contradicts, or ### +### nullifies the doctrine given in godas word. [SEP] ### +### ======================================= h_v_q | Gates: 25719 ======================================= ### +### ('doctrine', 0, 0) ('false', 1, 1) ('encompasses', 2, 7) ('definition', 3, 25) ### +### ('relating', 4, 17438) ('refers', 5, 244) ('is', 6, 241) ('noun', 7, 17155) ('##sam', 8, 24163) ### +### ('familiarity', 9, 26692) ('stylized', 10, 27326) ('plural', 11, 12331) ('encyclopedia', 12, 11439) ### +### ('consisting', 13, 24678) ('.', 14, 4914) ('something', 15, 742) ('term', 16, 7149) ### +### ('or', 17, 10590) ('theory', 18, 204) ('technology', 19, 2805) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('doctrine', 0, 0) ('false', 1, 1) ('biblical', 2003, 2) ('true', 28, 3) ('doctrines', 30, 4) ### +### ('bible', 295, 5) ('ˈ', 178, 6) ('encompasses', 2, 7) ('crashing', 129, 8) ('align', 27759, 9) ### +### ('define', 6089, 10) ('teachings', 794, 11) ('beliefs', 576, 12) ('##ο', 419, 13) ('−', 115, 14) ### +### ('revealed', 11639, 15) ('fake', 25, 16) ('stumbled', 153, 17) ('hating', 181, 18) ('##₂', 138, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('doctrine', 0, 0) ('false', 1, 1) ('encompasses', 2, 7) ('definition', 3, 25) ('true', 28, 3) ### +### ('doctrines', 30, 4) ('refers', 5, 244) ('is', 6, 241) ('fake', 25, 16) ('defined', 20, 140) ### +### ('crashed', 67, 27) ('principle', 29, 107) ('crashing', 129, 8) ('fraud', 39, 81) ### +### ('theory', 18, 204) ('−', 115, 14) ('ˈ', 178, 6) ('shoved', 85, 32) ('wrong', 47, 74) ### +### ('concept', 23, 187) ### +############################################################################################################ +[2023-10-08 00:33:06,048][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:33:06,048][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:33:06,470][root][INFO] - Epoch: 15: Step: 1501/1557, loss[v]=0.031976, lr=0.000004, acc@1[1]=249.5/256=0.974609375, acc@1[2]=253.0/256=0.98828125 +[2023-10-08 00:33:49,486][root][INFO] - rank=1; last iteration 1557 +[2023-10-08 00:33:49,486][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 00:33:49,487][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-08 00:33:49,490][root][INFO] - rank=0; last iteration 1557 +[2023-10-08 00:33:49,491][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 00:33:49,491][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-08 00:33:49,494][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 00:33:49,494][root][INFO] - Epoch finished on 1 +[2023-10-08 00:33:49,494][root][INFO] - rank=3; last iteration 1557 +[2023-10-08 00:33:49,495][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 00:33:49,495][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-08 00:33:49,496][root][INFO] - rank=2; last iteration 1557 +[2023-10-08 00:33:49,496][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 00:33:49,496][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-08 00:33:49,498][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 00:33:49,498][root][INFO] - Epoch finished on 0 +[2023-10-08 00:33:49,503][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 00:33:49,503][root][INFO] - Epoch finished on 3 +[2023-10-08 00:33:49,504][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 00:33:49,505][root][INFO] - Epoch finished on 2 +[2023-10-08 00:34:29,694][root][INFO] - Saved checkpoint at ./vdr_15 +[2023-10-08 00:34:29,695][root][INFO] - Av Loss per epoch=0.059603 +[2023-10-08 00:34:29,695][root][INFO] - epoch total (1) correct predictions=379798 +[2023-10-08 00:34:29,695][root][INFO] - epoch total (2) correct predictions=391708 +[2023-10-08 00:34:29,695][root][INFO] - Saved checkpoint at ./vdr_15 +[2023-10-08 00:34:29,696][root][INFO] - Av Loss per epoch=0.059603 +[2023-10-08 00:34:29,696][root][INFO] - epoch total (1) correct predictions=379798 +[2023-10-08 00:34:29,697][root][INFO] - epoch total (2) correct predictions=391708 +[2023-10-08 00:34:29,696][root][INFO] - Saved checkpoint at ./vdr_15 +[2023-10-08 00:34:29,697][root][INFO] - Av Loss per epoch=0.059603 +[2023-10-08 00:34:29,698][root][INFO] - epoch total (1) correct predictions=379798 +[2023-10-08 00:34:29,698][root][INFO] - epoch total (2) correct predictions=391708 +[2023-10-08 00:34:29,698][root][INFO] - Saved checkpoint at ./vdr_15 +[2023-10-08 00:34:29,698][root][INFO] - Av Loss per epoch=0.059603 +[2023-10-08 00:34:29,698][root][INFO] - epoch total (1) correct predictions=379798 +[2023-10-08 00:34:29,699][root][INFO] - epoch total (2) correct predictions=391708 +[2023-10-08 00:34:29,700][root][INFO] - ***** Epoch 16 ***** +[2023-10-08 00:34:29,701][root][INFO] - ***** Epoch 16 ***** +[2023-10-08 00:34:29,702][root][INFO] - ***** Epoch 16 ***** +[2023-10-08 00:34:29,702][root][INFO] - ***** Epoch 16 ***** +[2023-10-08 00:34:29,706][root][INFO] - rank=1; Iteration start +[2023-10-08 00:34:29,707][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 00:34:29,707][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 00:34:29,707][root][INFO] - rank=3; Iteration start +[2023-10-08 00:34:29,707][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 00:34:29,708][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 00:34:29,709][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-08 00:34:29,709][root][INFO] - rank=2; Iteration start +[2023-10-08 00:34:29,709][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 00:34:29,709][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-08 00:34:29,709][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 00:34:29,710][root][INFO] - rank=0; Iteration start +[2023-10-08 00:34:29,710][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 00:34:29,710][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 00:34:29,711][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-08 00:34:29,712][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-08 00:34:30,681][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29056.8/29522=98.42% | mean: 0.01 | max: 5.33 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.19 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is darpa grand challenge? [SEP] ### +### [P_TEXT]: [CLS] darpa grand challenge. the darpa grand challenge is a prize competition for ### +### american autonomous vehicles, funded by the defense advanced research projects agency, the most ### +### prominent research organization of the united states department of defense. [SEP] ### +### ======================================= h_v_q | Gates: 27301 ======================================= ### +### ('dar', 0, 3) ('##pa', 1, 1) ('grand', 2, 5) ('challenge', 3, 0) ('is', 4, 272) ### +### ('encompasses', 5, 8) ('challenges', 6, 23) ('stylized', 7, 28189) ('familiarity', 8, 20630) ### +### ('relating', 9, 21807) ('consisting', 10, 23387) ('plural', 11, 18130) ('##sam', 12, 25769) ### +### ('refers', 13, 19058) ('competition', 14, 6) ('what', 15, 163) ('league', 16, 1719) ### +### ('provides', 17, 7002) ('designed', 18, 8160) ('.', 19, 10402) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('challenge', 3, 0) ('##pa', 1, 1) ('autonomous', 2440, 2) ('dar', 0, 3) ('vehicles', 2798, 4) ### +### ('grand', 2, 5) ('competition', 14, 6) ('funded', 4801, 7) ('encompasses', 5, 8) ('ˈ', 76, 9) ### +### ('prize', 209, 10) ('prizes', 4864, 11) ('research', 1852, 12) ('defense', 1988, 13) ### +### ('crashing', 38, 14) ('vehicle', 3370, 15) ('challenged', 42, 16) ('sharply', 73, 17) ### +### ('stumbled', 78, 18) ('##ο', 101, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##pa', 1, 1) ('dar', 0, 3) ('grand', 2, 5) ('challenge', 3, 0) ('encompasses', 5, 8) ### +### ('challenges', 6, 23) ('competition', 14, 6) ('is', 4, 272) ('−', 24, 21) ('contest', 26, 29) ### +### ('crashing', 38, 14) ('challenged', 42, 16) ('crashed', 32, 36) ('ˈ', 76, 9) ('sharply', 73, 17) ### +### ('##₂', 60, 26) ('##α', 47, 39) ('stumbled', 78, 18) ('unwilling', 61, 31) ('hating', 80, 24) ### +############################################################################################################ +[2023-10-08 00:34:30,681][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:34:30,681][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:34:31,077][root][INFO] - Epoch: 16: Step: 1/1557, loss[v]=0.036909, lr=0.000004, acc@1[1]=248.0/256=0.96875, acc@1[2]=255.0/256=0.99609375 +[2023-10-08 00:35:48,011][root][INFO] - Train batch 100 +[2023-10-08 00:35:48,012][root][INFO] - Avg. loss per last 100 batches: 0.058351 +[2023-10-08 00:35:48,703][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29097.8/29522=98.56% | mean: 0.01 | max: 5.71 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.15 | max: 6.27 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is vci [SEP] ### +### [P_TEXT]: [CLS] vci is an acronym for volatile corrosion inhibitor. vci chemicals are a class of ### +### corrosion inhibiting compounds, which have sufficient vapor pressure to release molecules from the ### +### compound into the air. armor protective packaginga® vci products are packaging products, which ### +### contain armoras proprietary vci nanotechnologya¢ chemicals directly in the packaging. [SEP] ### +### ======================================= h_v_q | Gates: 27050 ======================================= ### +### ('vc', 0, 0) ('##i', 1, 5) ('is', 2, 576) ('encompasses', 3, 12) ('stylized', 4, 26793) ### +### ('##sam', 5, 28036) ('definition', 6, 41) ('plural', 7, 20600) ('familiarity', 8, 28052) ### +### ('consisting', 9, 23870) ('relating', 10, 26029) ('refers', 11, 19759) ('designed', 12, 7587) ### +### ('noun', 13, 20450) ('genus', 14, 5914) ('unit', 15, 1440) ('encyclopedia', 16, 13146) ### +### ('provides', 17, 8186) ('stands', 18, 7149) ('term', 19, 6069) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('vc', 0, 0) ('packaging', 2901, 1) ('corrosion', 12235, 2) ('armor', 6122, 3) ### +### ('volatile', 7007, 4) ('##i', 1, 5) ('chemicals', 3857, 6) ('inhibitor', 16250, 7) ### +### ('protective', 9127, 8) ('ˈ', 119, 9) ('proprietary', 389, 10) ('crashing', 78, 11) ### +### ('encompasses', 3, 12) ('nano', 5721, 13) ('inhibit', 5183, 14) ('##®', 24758, 15) ('##α', 230, 16) ### +### ('##ο', 591, 17) ('hating', 141, 18) ('−', 37, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('vc', 0, 0) ('##i', 1, 5) ('encompasses', 3, 12) ('definition', 6, 41) ('is', 2, 576) ### +### ('−', 37, 19) ('crashing', 78, 11) ('v', 24, 86) ('ˈ', 119, 9) ('stumbled', 80, 25) ### +### ('crashed', 64, 45) ('ছ', 75, 34) ('hating', 141, 18) ('angrily', 86, 48) ('julian', 69, 55) ### +### ('sharply', 134, 31) ('annoyance', 91, 53) ('ruined', 103, 51) ('##ང', 159, 29) ### +### ('unwilling', 177, 23) ### +############################################################################################################ +[2023-10-08 00:35:48,704][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:35:48,704][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:35:49,125][root][INFO] - Epoch: 16: Step: 101/1557, loss[v]=0.037592, lr=0.000004, acc@1[1]=243.5/256=0.951171875, acc@1[2]=254.0/256=0.9921875 +[2023-10-08 00:37:05,194][root][INFO] - Train batch 200 +[2023-10-08 00:37:05,195][root][INFO] - Avg. loss per last 100 batches: 0.059812 +[2023-10-08 00:37:05,903][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29107.2/29522=98.60% | mean: 0.01 | max: 5.30 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.5/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.09 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] penalty for cultivation of cannabis [SEP] ### +### [P_TEXT]: [CLS] in florida, cultivation of marijuana is the manufacturing or growing of cannabis ### +### for any purpose. cultivation is generally classified as a third degree felony, with penalties that ### +### may include prison and drug offender probation for even first - time offenders. nder section 893. ### +### 1351, florida statutes, cultivating or manufacturing of marijuana may be classified as a second ### +### degree felony, punishable by up to fifteen years in prison, where the accused is in actual or ### +### constructive possession of the grow house and is found in possession of more than 25 cannabis ### +### plants. [SEP] ### +### ======================================= h_v_q | Gates: 26443 ======================================= ### +### ('cannabis', 0, 3) ('marijuana', 1, 0) ('penalty', 2, 244) ('cultivation', 3, 1) ('.', 4, 10479) ### +### ('punishment', 5, 240) ('familiarity', 6, 27722) ('cultivated', 7, 96) ('of', 8, 6022) ### +### ('penalties', 9, 61) ('stylized', 10, 25521) ('for', 11, 20160) ('consisting', 12, 20888) ### +### ('=', 13, 11781) ('relating', 14, 26087) ('$', 15, 3347) ('specified', 16, 5159) ### +### ('plural', 17, 19837) ('prize', 18, 12207) ('agriculture', 19, 213) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('marijuana', 1, 0) ('cultivation', 3, 1) ('felony', 1279, 2) ('cannabis', 0, 3) ### +### ('florida', 1913, 4) ('classified', 15502, 5) ('ˈ', 297, 6) ('cult', 2779, 7) ('prison', 45, 8) ### +### ('encompasses', 1225, 9) ('crashing', 44, 10) ('weed', 512, 11) ('−', 34, 12) ('##ο', 123, 13) ### +### ('wingspan', 628, 14) ('growing', 1105, 15) ('hesitated', 79, 16) ('hating', 214, 17) ### +### ('unwilling', 46, 18) ('second', 5527, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('cannabis', 0, 3) ('marijuana', 1, 0) ('cultivation', 3, 1) ('penalty', 2, 244) ### +### ('cultivated', 7, 96) ('punishment', 5, 240) ('penalties', 9, 61) ('−', 34, 12) ('prison', 45, 8) ### +### ('crashing', 44, 10) ('gabe', 22, 50) ('##α', 33, 23) ('punish', 32, 34) ('unwilling', 46, 18) ### +### ('simon', 23, 69) ('opium', 29, 60) ('hesitated', 79, 16) ('degree', 55, 28) ('##₂', 61, 31) ### +### ('sharply', 71, 24) ### +############################################################################################################ +[2023-10-08 00:37:05,903][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:37:05,903][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:37:06,307][root][INFO] - Epoch: 16: Step: 201/1557, loss[v]=0.057797, lr=0.000004, acc@1[1]=244.0/256=0.953125, acc@1[2]=251.0/256=0.98046875 +[2023-10-08 00:38:22,881][root][INFO] - Train batch 300 +[2023-10-08 00:38:22,882][root][INFO] - Avg. loss per last 100 batches: 0.059240 +[2023-10-08 00:38:23,612][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29035.8/29522=98.35% | mean: 0.01 | max: 5.52 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.44 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] who are davita medical group's competitors [SEP] ### +### [P_TEXT]: [CLS] fresenius medical care : fresenius is by far davitaas largest competitor, with ### +### about 1 / 3 market share and about 150, 000 patients. this company may have a cost advantage over ### +### davita because they manufacture dialysis supplies and equipment. [SEP] ### +### ======================================= h_v_q | Gates: 27421 ======================================= ### +### ('##vita', 0, 0) ('da', 1, 6) ('competitors', 2, 11) ('group', 3, 2684) ('medical', 4, 10) ### +### ('competitor', 5, 9) ('competition', 6, 42) ('competitive', 7, 208) ('competing', 8, 240) ### +### ('.', 9, 17444) ('are', 10, 5529) ('rivals', 11, 7) ('stylized', 12, 28219) ### +### ('familiarity', 13, 26036) ('who', 14, 313) ('compete', 15, 548) ('competed', 16, 1131) ### +### ('were', 17, 6539) ('medicine', 18, 39) ('whose', 19, 1034) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##vita', 0, 0) ('##nius', 24371, 1) ('fr', 13874, 2) ('largest', 1475, 3) ('ˈ', 156, 4) ### +### ('advantage', 8828, 5) ('da', 1, 6) ('rivals', 11, 7) ('##ese', 8437, 8) ('competitor', 5, 9) ### +### ('medical', 4, 10) ('competitors', 2, 11) ('##ο', 109, 12) ('care', 1519, 13) ### +### ('patients', 3015, 14) ('hesitated', 155, 15) ('hating', 59, 16) ('wingspan', 543, 17) ### +### ('−', 54, 18) ('stumbled', 188, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##vita', 0, 0) ('da', 1, 6) ('competitors', 2, 11) ('medical', 4, 10) ('competitor', 5, 9) ### +### ('competition', 6, 42) ('rivals', 11, 7) ('competitive', 7, 208) ('group', 3, 2684) ### +### ('medicine', 18, 39) ('rival', 22, 28) ('competing', 8, 240) ('hospital', 25, 84) ### +### ('hating', 59, 16) ('−', 54, 18) ('health', 23, 120) ('who', 14, 313) ('crashing', 66, 23) ### +### ('crashed', 37, 57) ('ˈ', 156, 4) ### +############################################################################################################ +[2023-10-08 00:38:23,613][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:38:23,613][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:38:24,041][root][INFO] - Epoch: 16: Step: 301/1557, loss[v]=0.033574, lr=0.000004, acc@1[1]=246.5/256=0.962890625, acc@1[2]=255.0/256=0.99609375 +[2023-10-08 00:39:40,265][root][INFO] - Train batch 400 +[2023-10-08 00:39:40,266][root][INFO] - Avg. loss per last 100 batches: 0.057522 +[2023-10-08 00:39:40,950][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29058.9/29522=98.43% | mean: 0.01 | max: 5.66 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.35 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] age group most affected by car accidents [SEP] ### +### [P_TEXT]: [CLS] young drivers are essentially the group most at risk of being involved in a car ### +### accident with two drivers under the age of 25 dying as the result of car accidents every day. young ### +### male drivers are also at more risk of having a car accident than young female drivers. [SEP] ### +### ======================================= h_v_q | Gates: 27144 ======================================= ### +### ('car', 0, 8) ('age', 1, 6) ('accidents', 2, 4) ('affected', 3, 5075) ('group', 4, 38) ### +### ('.', 5, 15575) ('accident', 6, 1) ('most', 7, 114) ('ages', 8, 1164) ('affect', 9, 11153) ### +### ('cars', 10, 25) ('familiarity', 11, 24013) ('older', 12, 53) ('old', 13, 47) ('years', 14, 3276) ### +### ('highest', 15, 325) ('stylized', 16, 28774) ('automobile', 17, 87) ('consisting', 18, 26250) ### +### ('damaged', 19, 992) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('drivers', 2239, 0) ('accident', 6, 1) ('driver', 2215, 2) ('young', 61, 3) ('accidents', 2, 4) ### +### ('risk', 1647, 5) ('age', 1, 6) ('ˈ', 375, 7) ('car', 0, 8) ('wingspan', 330, 9) ('die', 10151, 10) ### +### ('crashed', 67, 11) ('##ο', 75, 12) ('hesitated', 145, 13) ('##₂', 126, 14) ('−', 56, 15) ### +### ('hating', 315, 16) ('driving', 1398, 17) ('risks', 4880, 18) ('crashing', 63, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('accidents', 2, 4) ('car', 0, 8) ('age', 1, 6) ('accident', 6, 1) ('group', 4, 38) ### +### ('most', 7, 114) ('cars', 10, 25) ('older', 12, 53) ('old', 13, 47) ('automobile', 17, 87) ### +### ('affected', 3, 5075) ('young', 61, 3) ('vehicle', 20, 76) ('crashed', 67, 11) ('−', 56, 15) ### +### ('##ο', 75, 12) ('crashing', 63, 19) ('groups', 22, 127) ('ages', 8, 1164) ('highest', 15, 325) ### +############################################################################################################ +[2023-10-08 00:39:40,950][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:39:40,950][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:39:41,371][root][INFO] - Epoch: 16: Step: 401/1557, loss[v]=0.042489, lr=0.000004, acc@1[1]=243.5/256=0.951171875, acc@1[2]=252.0/256=0.984375 +[2023-10-08 00:40:58,066][root][INFO] - Train batch 500 +[2023-10-08 00:40:58,067][root][INFO] - Avg. loss per last 100 batches: 0.057610 +[2023-10-08 00:40:58,770][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29063.6/29522=98.45% | mean: 0.01 | max: 5.60 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.14 | max: 6.25 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] rice lake is in what county [SEP] ### +### [P_TEXT]: [CLS] rice lake is a city in saint louis county, minnesota, united states. the population ### +### was 4, 095 at the 2010 census. main routes include rice lake road and martin road. rice lake road ### +### runs northasouth, and martin road runs eastawest. other routes include howard gnesen road, arnold ### +### road, calvary road, west tischer road, and west beyer road. [SEP] ### +### ======================================= h_v_q | Gates: 25740 ======================================= ### +### ('rice', 0, 0) ('lake', 1, 1) ('county', 2, 13) ('familiarity', 3, 24110) ('is', 4, 298) ### +### ('stylized', 5, 28393) ('relating', 6, 25692) ('.', 7, 5323) ('plural', 8, 15535) ### +### ('consisting', 9, 24621) ('pond', 10, 43) ('counties', 11, 41) ('lakes', 12, 5) ### +### ('mathematics', 13, 23691) ('mountain', 14, 182) ('province', 15, 2794) ('simon', 16, 53) ### +### ('reservoir', 17, 31) ('designed', 18, 16288) ('cotton', 19, 179) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('rice', 0, 0) ('lake', 1, 1) ('arnold', 714, 2) ('minnesota', 1224, 3) ('martin', 446, 4) ### +### ('lakes', 12, 5) ('ˈ', 73, 6) ('bey', 19607, 7) ('routes', 16975, 8) ('##outh', 26384, 9) ### +### ('louis', 1971, 10) ('crashing', 77, 11) ('road', 778, 12) ('county', 2, 13) ('mn', 27419, 14) ### +### ('saint', 3867, 15) ('unwilling', 21, 16) ('stumbled', 93, 17) ('−', 51, 18) ### +### ('population', 359, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('rice', 0, 0) ('lake', 1, 1) ('county', 2, 13) ('lakes', 12, 5) ('pond', 10, 43) ### +### ('counties', 11, 41) ('unwilling', 21, 16) ('reservoir', 17, 31) ('simon', 16, 53) ('ˈ', 73, 6) ### +### ('−', 51, 18) ('is', 4, 298) ('crashing', 77, 11) ('crashed', 42, 29) ('ছ', 48, 23) ### +### ('census', 37, 42) ('shoved', 49, 35) ('angrily', 44, 46) ('stumbled', 93, 17) ('hating', 58, 40) ### +############################################################################################################ +[2023-10-08 00:40:58,770][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:40:58,770][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:40:59,194][root][INFO] - Epoch: 16: Step: 501/1557, loss[v]=0.065750, lr=0.000004, acc@1[1]=245.0/256=0.95703125, acc@1[2]=250.0/256=0.9765625 +[2023-10-08 00:42:16,507][root][INFO] - Train batch 600 +[2023-10-08 00:42:16,508][root][INFO] - Avg. loss per last 100 batches: 0.060224 +[2023-10-08 00:42:17,256][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29068.2/29522=98.46% | mean: 0.01 | max: 5.41 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.16 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how long it takes to get a master's degree [SEP] ### +### [P_TEXT]: [CLS] in most cases, a master's degree program takes two years to complete, although ### +### there are exceptions to the rule. if you'd like to know how long it would take to earn a master's ### +### degree, you should consider how much time you could devote to school and the specific type of ### +### program you'll be enrolling in. [SEP] ### +### ======================================= h_v_q | Gates: 26549 ======================================= ### +### ('master', 0, 4) ('degree', 1, 1) ('takes', 2, 37) ('minutes', 3, 43) ('weeks', 4, 9) ### +### ('days', 5, 394) ('years', 6, 51) ('take', 7, 109) ('hours', 8, 1028) ('months', 9, 847) ### +### ('took', 10, 435) ('.', 11, 8590) ('familiarity', 12, 20274) ('long', 13, 0) ('masters', 14, 12) ### +### ('stylized', 15, 28208) ('length', 16, 87) ('get', 17, 1197) ('time', 18, 50) ('miles', 19, 5596) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('long', 13, 0) ('degree', 1, 1) ('ˈ', 81, 2) ('unwilling', 48, 3) ('master', 0, 4) ### +### ('hating', 123, 5) ('crashing', 60, 6) ('##ο', 363, 7) ('cyrillic', 509, 8) ('weeks', 4, 9) ### +### ('##₂', 71, 10) ('hesitated', 173, 11) ('masters', 14, 12) ('sharply', 63, 13) ('##ང', 170, 14) ### +### ('wingspan', 1007, 15) ('hated', 179, 16) ('−', 66, 17) ('stumbled', 155, 18) ('program', 153, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('master', 0, 4) ('degree', 1, 1) ('takes', 2, 37) ('weeks', 4, 9) ('minutes', 3, 43) ### +### ('years', 6, 51) ('long', 13, 0) ('take', 7, 109) ('masters', 14, 12) ('days', 5, 394) ### +### ('time', 18, 50) ('length', 16, 87) ('longer', 24, 25) ('took', 10, 435) ('unwilling', 48, 3) ### +### ('degrees', 28, 29) ('taking', 21, 85) ('months', 9, 847) ('ˈ', 81, 2) ('crashing', 60, 6) ### +############################################################################################################ +[2023-10-08 00:42:17,256][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:42:17,256][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:42:17,678][root][INFO] - Epoch: 16: Step: 601/1557, loss[v]=0.048904, lr=0.000004, acc@1[1]=247.5/256=0.966796875, acc@1[2]=255.0/256=0.99609375 +[2023-10-08 00:43:35,607][root][INFO] - Train batch 700 +[2023-10-08 00:43:35,608][root][INFO] - Avg. loss per last 100 batches: 0.057394 +[2023-10-08 00:43:36,323][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29059.1/29522=98.43% | mean: 0.01 | max: 5.33 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.8/29522=100.00% | mean: 0.16 | max: 6.13 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] where is c + + used today [SEP] ### +### [P_TEXT]: [CLS] c + + is used today in any field that requires performance, like gaming for ### +### instance. c + + is also very reliable and is the prime choice for embedded systems, which must be ### +### reliable to work and not cause exceptions. c + + can be used to create virtual... [SEP] ### +### ======================================= h_v_q | Gates: 26996 ======================================= ### +### ('+', 0, 2) ('c', 1, 7) ('today', 2, 15) ('used', 3, 18) ('downtown', 4, 1620) ('plus', 5, 147) ### +### ('stylized', 6, 28656) ('yesterday', 7, 139) ('located', 8, 9457) ('tonight', 9, 221) ### +### ('florida', 10, 17677) ('africa', 11, 9967) ('southern', 12, 12954) ('.', 13, 14468) ### +### ('where', 14, 5463) ('familiarity', 15, 27030) ('america', 16, 23660) ('situated', 17, 5700) ### +### ('consisting', 18, 25493) ('colorado', 19, 14485) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('embedded', 7522, 0) ('reliable', 7945, 1) ('+', 0, 2) ('exceptions', 15674, 3) ('ˈ', 151, 4) ### +### ('gaming', 6212, 5) ('exception', 11759, 6) ('c', 1, 7) ('virtual', 3968, 8) ('hating', 594, 9) ### +### ('##ο', 119, 10) ('unwilling', 138, 11) ('cyrillic', 528, 12) ('crashing', 64, 13) ### +### ('hesitated', 106, 14) ('today', 2, 15) ('sharply', 204, 16) ('−', 41, 17) ('used', 3, 18) ### +### ('choice', 5517, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('+', 0, 2) ('c', 1, 7) ('today', 2, 15) ('used', 3, 18) ('plus', 5, 147) ('−', 41, 17) ### +### ('yesterday', 7, 139) ('crashing', 64, 13) ('uses', 66, 22) ('ˈ', 151, 4) ('hesitated', 106, 14) ### +### ('##ο', 119, 10) ('downtown', 4, 1620) ('tonight', 9, 221) ('crashed', 87, 35) ('·', 30, 144) ### +### ('unwilling', 138, 11) ('use', 57, 70) ('##α', 102, 37) ('##₂', 123, 24) ### +############################################################################################################ +[2023-10-08 00:43:36,324][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:43:36,324][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:43:36,747][root][INFO] - Epoch: 16: Step: 701/1557, loss[v]=0.058058, lr=0.000004, acc@1[1]=245.5/256=0.958984375, acc@1[2]=250.0/256=0.9765625 +[2023-10-08 00:44:54,680][root][INFO] - Train batch 800 +[2023-10-08 00:44:54,681][root][INFO] - Avg. loss per last 100 batches: 0.057872 +[2023-10-08 00:44:55,381][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29103.8/29522=98.58% | mean: 0.01 | max: 5.23 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.12 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is acidic and basic? [SEP] ### +### [P_TEXT]: [CLS] acidic and basic are two extremes that describe a chemical property chemicals. ### +### mixing acids and bases can cancel out or neutralize their extreme effects. a substance that is ### +### neither acidic nor basic is neutral. the ph scale measures how acidic or basic a substance is. the ### +### ph scale ranges from 0 to 14. a ph of 7 is neutral. a ph less than 7 is acidic. a ph greater than 7 ### +### is basic. the ph scale is logarithmic and as a result, each whole ph value below 7 is ten times ### +### more acidic than the next higher value. for example, ph 4 is ten times more acidic than ph 5 and ### +### 100 times ( 10 times 10 ) more acidic than ph 6. the same holds true for ph values above 7, each of ### +### which is ten times more alkaline ( another way to say basic ) than the next lower whole value. ### +### [SEP] ### +### ======================================= h_v_q | Gates: 26371 ======================================= ### +### ('acidic', 0, 0) ('basic', 1, 2) ('is', 2, 553) ('acid', 3, 18) ('definition', 4, 120) ### +### ('encompasses', 5, 19) ('familiarity', 6, 24848) ('refers', 7, 11629) ('stylized', 8, 28521) ### +### ('relating', 9, 24084) ('stands', 10, 5337) ('.', 11, 13339) ('designed', 12, 15300) ### +### ('consisting', 13, 21557) ('encyclopedia', 14, 11129) ('noun', 15, 26087) ('##sam', 16, 28034) ### +### ('provides', 17, 6154) ('mathematics', 18, 19547) ('plural', 19, 18334) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('acidic', 0, 0) ('ph', 2685, 1) ('basic', 1, 2) ('neutral', 394, 3) ('neither', 10614, 4) ### +### ('chemicals', 2148, 5) ('extremes', 15286, 6) ('ˈ', 302, 7) ('wingspan', 748, 8) ('scale', 6406, 9) ### +### ('##ο', 207, 10) ('mixing', 4670, 11) ('hating', 134, 12) ('##₂', 108, 13) ('crashing', 122, 14) ### +### ('acids', 30, 15) ('unwilling', 182, 16) ('hesitated', 319, 17) ('acid', 3, 18) ### +### ('encompasses', 5, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('acidic', 0, 0) ('basic', 1, 2) ('acid', 3, 18) ('encompasses', 5, 19) ('definition', 4, 120) ### +### ('base', 20, 27) ('angrily', 22, 35) ('is', 2, 553) ('acids', 30, 15) ('chemical', 31, 32) ### +### ('ruined', 35, 44) ('−', 77, 20) ('##₂', 108, 13) ('crashing', 122, 14) ('hating', 134, 12) ### +### ('sharply', 109, 21) ('crashed', 92, 34) ('hated', 99, 36) ('##α', 93, 41) ('unwilling', 182, 16) ### +############################################################################################################ +[2023-10-08 00:44:55,381][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:44:55,381][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:44:55,804][root][INFO] - Epoch: 16: Step: 801/1557, loss[v]=0.080280, lr=0.000004, acc@1[1]=243.0/256=0.94921875, acc@1[2]=250.5/256=0.978515625 +[2023-10-08 00:46:13,542][root][INFO] - Train batch 900 +[2023-10-08 00:46:13,543][root][INFO] - Avg. loss per last 100 batches: 0.060593 +[2023-10-08 00:46:14,243][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29091.8/29522=98.54% | mean: 0.01 | max: 5.47 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.18 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what shape has all equal angles [SEP] ### +### [P_TEXT]: [CLS] the shapes of elementary geometry are invariably convex. starting with the most ### +### regular quadrilateral, namely, the square, we shall define other shapes by relaxing its properties. ### +### a square is a quadrilateral with all sides equal and all angles also equal. angles in any ### +### quadrilateral add up to 360a°. it follows that, in a square, all angles measure 90a°. an ### +### equiangular quadrilateral, i. e. the one with all angles equal is a rectangle. all angles of a ### +### rectangle equal 90a°. an equilateral quadrilateral, i. e. the one with all sides equal, is a ### +### rhombus. in a square, rectangle, or rhombus, the opposite side lines are parallel. [SEP] ### +### ======================================= h_v_q | Gates: 26906 ======================================= ### +### ('angles', 0, 5) ('equal', 1, 30) ('shape', 2, 8) ('all', 3, 33) ('has', 4, 12354) ('form', 5, 228) ### +### ('.', 6, 16190) ('familiarity', 7, 26302) ('having', 8, 3605) ('stylized', 9, 28151) ### +### ('consisting', 10, 24181) ('have', 11, 15204) ('shapes', 12, 2) ('had', 13, 10750) ### +### ('relating', 14, 23877) ('plural', 15, 14023) ('identical', 16, 542) ('shaped', 17, 108) ### +### ('##α', 18, 52) ('julian', 19, 69) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('convex', 11467, 0) ('quad', 5515, 1) ('shapes', 12, 2) ('square', 266, 3) ('geometry', 339, 4) ### +### ('angles', 0, 5) ('##tangle', 9015, 6) ('ˈ', 59, 7) ('shape', 2, 8) ('elementary', 5115, 9) ### +### ('90', 5917, 10) ('opposite', 1637, 11) ('hating', 47, 12) ('##ο', 53, 13) ('unwilling', 33, 14) ### +### ('crashing', 31, 15) ('hesitated', 36, 16) ('angle', 22, 17) ('wingspan', 159, 18) ### +### ('##hom', 27951, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('angles', 0, 5) ('shape', 2, 8) ('equal', 1, 30) ('all', 3, 33) ('shapes', 12, 2) ('form', 5, 228) ### +### ('angle', 22, 17) ('##α', 18, 52) ('crashing', 31, 15) ('unwilling', 33, 14) ('##₂', 30, 22) ### +### ('ruined', 24, 34) ('hesitated', 36, 16) ('ˈ', 59, 7) ('hating', 47, 12) ('julian', 19, 69) ### +### ('angrily', 23, 48) ('##ο', 53, 13) ('sharply', 43, 26) ('crashed', 37, 38) ### +############################################################################################################ +[2023-10-08 00:46:14,243][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:46:14,243][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:46:14,645][root][INFO] - Epoch: 16: Step: 901/1557, loss[v]=0.113528, lr=0.000004, acc@1[1]=240.5/256=0.939453125, acc@1[2]=246.0/256=0.9609375 +[2023-10-08 00:47:32,365][root][INFO] - Train batch 1000 +[2023-10-08 00:47:32,366][root][INFO] - Avg. loss per last 100 batches: 0.062211 +[2023-10-08 00:47:33,110][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29078.6/29522=98.50% | mean: 0.01 | max: 5.56 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.14 | max: 6.36 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] definition of dehydration [SEP] ### +### [P_TEXT]: [CLS] dehydration is a condition where the water levels in the body are too low. this can ### +### result in numerous minor and major effects within the body and can be deadly. this article ### +### addresses dehydration, its causes, its symptoms and effects. [SEP] ### +### ======================================= h_v_q | Gates: 27755 ======================================= ### +### ('de', 0, 1) ('##tion', 1, 6) ('##hy', 2, 69) ('##dra', 3, 123) ('definition', 4, 32) ### +### ('relating', 5, 22935) ('stylized', 6, 28300) ('noun', 7, 23143) ('familiarity', 8, 25802) ### +### ('plural', 9, 10652) ('consisting', 10, 24628) ('refers', 11, 6531) ('encyclopedia', 12, 11206) ### +### ('defined', 13, 352) ('something', 14, 6138) ('##º', 15, 27562) ('term', 16, 9005) ### +### ('definitions', 17, 20) ('mathematics', 18, 22154) ('von', 19, 188) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('deadly', 9763, 0) ('de', 0, 1) ('hating', 94, 2) ('ˈ', 36, 3) ('hesitated', 63, 4) ### +### ('water', 3647, 5) ('##tion', 1, 6) ('crashing', 20, 7) ('sharply', 35, 8) ('−', 25, 9) ### +### ('wingspan', 124, 10) ('##ο', 117, 11) ('unwilling', 21, 12) ('##ང', 57, 13) ('cyrillic', 129, 14) ### +### ('##₂', 29, 15) ('stumbled', 55, 16) ('encompasses', 28, 17) ('crashed', 22, 18) ('gideon', 61, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('de', 0, 1) ('##tion', 1, 6) ('definition', 4, 32) ('##hy', 2, 69) ('##dra', 3, 123) ### +### ('crashing', 20, 7) ('definitions', 17, 20) ('unwilling', 21, 12) ('−', 25, 9) ('crashed', 22, 18) ### +### ('ˈ', 36, 3) ('angrily', 23, 26) ('sharply', 35, 8) ('##₂', 29, 15) ('encompasses', 28, 17) ### +### ('ruined', 26, 30) ('##α', 24, 34) ('hesitated', 63, 4) ('hugh', 39, 27) ('##ང', 57, 13) ### +############################################################################################################ +[2023-10-08 00:47:33,111][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:47:33,111][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:47:33,537][root][INFO] - Epoch: 16: Step: 1001/1557, loss[v]=0.048406, lr=0.000004, acc@1[1]=244.5/256=0.955078125, acc@1[2]=252.5/256=0.986328125 +[2023-10-08 00:48:50,970][root][INFO] - Train batch 1100 +[2023-10-08 00:48:50,971][root][INFO] - Avg. loss per last 100 batches: 0.058804 +[2023-10-08 00:48:51,732][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29076.2/29522=98.49% | mean: 0.01 | max: 5.46 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.26 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what creditors can report if there is a credit freeze? [SEP] ### +### [P_TEXT]: [CLS] a credit report freeze blocks your credit reports from being shared with any new ### +### potential creditors, such as banks or credit card issuers, or any company that requests to see it, ### +### such as cell phone or utility services. [SEP] ### +### ======================================= h_v_q | Gates: 26855 ======================================= ### +### ('freeze', 0, 0) ('credit', 1, 1) ('report', 2, 4) ('creditors', 3, 14) ('.', 4, 17997) ### +### ('if', 5, 8537) ('reported', 6, 71) ('familiarity', 7, 27624) ('there', 8, 9464) ('reports', 9, 2) ### +### ('whenever', 10, 948) ('can', 11, 12541) ('stylized', 12, 28711) ('couldn', 13, 96) ### +### ('consisting', 14, 26041) ('reporting', 15, 43) ('stands', 16, 6417) ('freezing', 17, 36) ### +### ('relating', 18, 24884) ('hoped', 19, 1028) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('freeze', 0, 0) ('credit', 1, 1) ('reports', 9, 2) ('blocks', 5660, 3) ('report', 2, 4) ### +### ('ˈ', 78, 5) ('utility', 13645, 6) ('shared', 8699, 7) ('crashing', 42, 8) ('−', 52, 9) ### +### ('hating', 68, 10) ('hesitated', 95, 11) ('##ο', 91, 12) ('unwilling', 28, 13) ('creditors', 3, 14) ### +### ('cyrillic', 217, 15) ('wingspan', 315, 16) ('##ང', 142, 17) ('sharply', 62, 18) ### +### ('stumbled', 145, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('freeze', 0, 0) ('credit', 1, 1) ('report', 2, 4) ('creditors', 3, 14) ('reports', 9, 2) ### +### ('reported', 6, 71) ('reporting', 15, 43) ('freezing', 17, 36) ('unwilling', 28, 13) ### +### ('couldn', 13, 96) ('crashing', 42, 8) ('##₂', 30, 30) ('−', 52, 9) ('ˈ', 78, 5) ('ruined', 29, 35) ### +### ('simon', 24, 58) ('hating', 68, 10) ('angrily', 35, 32) ('sharply', 62, 18) ('##α', 34, 39) ### +############################################################################################################ +[2023-10-08 00:48:51,732][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:48:51,732][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:48:52,135][root][INFO] - Epoch: 16: Step: 1101/1557, loss[v]=0.057741, lr=0.000003, acc@1[1]=238.0/256=0.9296875, acc@1[2]=252.5/256=0.986328125 +[2023-10-08 00:50:09,732][root][INFO] - Train batch 1200 +[2023-10-08 00:50:09,733][root][INFO] - Avg. loss per last 100 batches: 0.058108 +[2023-10-08 00:50:10,464][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29109.8/29522=98.60% | mean: 0.01 | max: 5.33 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.19 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] how much does botox cost for frown lines [SEP] ### +### [P_TEXT]: [CLS] the cost per unit varies from 10 dollars to 20 dollars. therefore when you multiply ### +### 25 units by 20 dollars you will be able to get how much is botox for the frown lines, which is 500 ### +### dollars. men needs more units than women to treatment and prevent appearance of lines between the ### +### eyebrows. ccording to some of the dermatologist, botox is charged depending with the area where it ### +### is injected. forehead has two parts and the charges is 300 dollars per part, therefore you will ### +### walk away with a total bill of 600 dollars for the whole forehead. your geographical area ### +### determines. and other parts of the body. [SEP] ### +### ======================================= h_v_q | Gates: 26912 ======================================= ### +### ('frown', 0, 3) ('bot', 1, 5) ('$', 2, 33) ('##ox', 3, 31) ('lines', 4, 7) ('##£', 5, 22122) ### +### ('cost', 6, 6) ('familiarity', 7, 25677) ('stylized', 8, 29119) ('.', 9, 13991) ('430', 10, 28871) ### +### ('pounds', 11, 2730) ('answer', 12, 24694) ('considerable', 13, 36) ('much', 14, 56) ### +### ('significantly', 15, 90) ('line', 16, 44) ('frowning', 17, 350) ('cents', 18, 3182) ### +### ('average', 19, 3067) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('forehead', 1461, 0) ('depending', 4230, 1) ('eyebrows', 6397, 2) ('frown', 0, 3) ('ˈ', 42, 4) ### +### ('bot', 1, 5) ('cost', 6, 6) ('lines', 4, 7) ('##ord', 15142, 8) ('##ο', 52, 9) ### +### ('unwilling', 49, 10) ('hating', 41, 11) ('hesitated', 97, 12) ('wingspan', 580, 13) ### +### ('injected', 15691, 14) ('crashing', 65, 15) ('units', 325, 16) ('cc', 12292, 17) ('costs', 23, 18) ### +### ('unit', 181, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('frown', 0, 3) ('bot', 1, 5) ('$', 2, 33) ('lines', 4, 7) ('##ox', 3, 31) ('cost', 6, 6) ### +### ('considerable', 13, 36) ('line', 16, 44) ('costs', 23, 18) ('much', 14, 56) ('ˈ', 42, 4) ### +### ('greatly', 22, 49) ('significantly', 15, 90) ('hating', 41, 11) ('price', 28, 37) ### +### ('unwilling', 49, 10) ('##ο', 52, 9) ('##α', 30, 47) ('300', 48, 20) ('##₂', 39, 43) ### +############################################################################################################ +[2023-10-08 00:50:10,465][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:50:10,465][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:50:10,886][root][INFO] - Epoch: 16: Step: 1201/1557, loss[v]=0.076045, lr=0.000003, acc@1[1]=241.5/256=0.943359375, acc@1[2]=250.0/256=0.9765625 +[2023-10-08 00:51:28,875][root][INFO] - Train batch 1300 +[2023-10-08 00:51:28,876][root][INFO] - Avg. loss per last 100 batches: 0.056848 +[2023-10-08 00:51:29,624][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29072.8/29522=98.48% | mean: 0.01 | max: 5.62 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.20 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what cut of beef is brisket [SEP] ### +### [P_TEXT]: [CLS] beef brisket is a cut from the breast section of the animal. it is usually sold ### +### boneless. because brisket is a tough cut of meat, it's best when braised ( that is, simmered in a ### +### small amount of liquid ), either in the oven, the slow cooker, or on the stove top. two different ### +### cuts of brisket are available. unless the recipe specifies one or the other, either may be used in ### +### recipes calling for boneless beef brisket : beef brisket flat half ( also called thin cut, flat ### +### cut, first cut, or center cut ) : with its minimal fat, this cut is generally the pricier of the ### +### two. [SEP] ### +### ======================================= h_v_q | Gates: 26927 ======================================= ### +### ('brisk', 0, 0) ('cut', 1, 3) ('##et', 2, 22) ('beef', 3, 2) ('cuts', 4, 8) ('is', 5, 253) ### +### ('stylized', 6, 28296) ('familiarity', 7, 27674) ('cutting', 8, 41) ('.', 9, 16122) ### +### ('of', 10, 22073) ('consisting', 11, 20881) ('plural', 12, 13938) ('relating', 13, 25630) ### +### ('encompasses', 14, 6) ('##ets', 15, 287) ('split', 16, 470) ('meat', 17, 5) ('force', 18, 3345) ### +### ('slice', 19, 1353) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('brisk', 0, 0) ('ˈ', 84, 1) ('beef', 3, 2) ('cut', 1, 3) ('##ο', 67, 4) ('meat', 17, 5) ### +### ('encompasses', 14, 6) ('hating', 59, 7) ('cuts', 4, 8) ('unwilling', 26, 9) ('crashing', 36, 10) ### +### ('half', 10164, 11) ('hesitated', 55, 12) ('stumbled', 119, 13) ('##less', 7530, 14) ('−', 22, 15) ### +### ('cyrillic', 168, 16) ('gideon', 96, 17) ('sharply', 54, 18) ('##大', 56, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('brisk', 0, 0) ('cut', 1, 3) ('beef', 3, 2) ('##et', 2, 22) ('cuts', 4, 8) ('cutting', 8, 41) ### +### ('encompasses', 14, 6) ('is', 5, 253) ('meat', 17, 5) ('unwilling', 26, 9) ('−', 22, 15) ### +### ('crashing', 36, 10) ('shoved', 21, 28) ('angrily', 30, 30) ('crashed', 31, 33) ('hating', 59, 7) ### +### ('ˈ', 84, 1) ('hesitated', 55, 12) ('ছ', 47, 23) ('##α', 37, 35) ### +############################################################################################################ +[2023-10-08 00:51:29,624][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:51:29,624][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:51:30,028][root][INFO] - Epoch: 16: Step: 1301/1557, loss[v]=0.037283, lr=0.000003, acc@1[1]=246.0/256=0.9609375, acc@1[2]=254.0/256=0.9921875 +[2023-10-08 00:52:47,547][root][INFO] - Train batch 1400 +[2023-10-08 00:52:47,548][root][INFO] - Avg. loss per last 100 batches: 0.060828 +[2023-10-08 00:52:48,296][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29118.4/29522=98.63% | mean: 0.01 | max: 5.36 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.7/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.16 | max: 6.39 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is redemption value of a bond [SEP] ### +### [P_TEXT]: [CLS] description. the redemption value is the par or premium price of a debt security or ### +### preferred stock issue for which it can be repaid at or before its maturity date. it is the price at ### +### which a bond or preferred stock can be called by the issuing company. efinition redemption value. ### +### the redemption value is the par or premium price of a debt security or preferred stock issue for ### +### which it can be repaid at or before its maturity date. [SEP] ### +### ======================================= h_v_q | Gates: 26468 ======================================= ### +### ('redemption', 0, 0) ('bond', 1, 7) ('value', 2, 1) ('$', 3, 160) ('##£', 4, 21583) ('.', 5, 17027) ### +### ('familiarity', 6, 25347) ('is', 7, 313) ('stylized', 8, 27748) ('relating', 9, 24710) ### +### ('plural', 10, 17628) ('bonds', 11, 42) ('price', 12, 8) ('consisting', 13, 23156) ### +### ('encyclopedia', 14, 11054) ('narrow', 15, 9568) ('definition', 16, 24) ('values', 17, 35) ### +### ('mathematics', 18, 17820) ('encompasses', 19, 13) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('redemption', 0, 0) ('value', 2, 1) ('description', 5679, 2) ('ˈ', 216, 3) ('maturity', 3460, 4) ### +### ('prices', 44, 5) ('debt', 3950, 6) ('bond', 1, 7) ('price', 12, 8) ('preferred', 8163, 9) ### +### ('crashing', 25, 10) ('hating', 46, 11) ('##ο', 191, 12) ('encompasses', 19, 13) ('−', 51, 14) ### +### ('rep', 7908, 15) ('hesitated', 114, 16) ('unwilling', 64, 17) ('stumbled', 165, 18) ### +### ('wingspan', 615, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('redemption', 0, 0) ('bond', 1, 7) ('value', 2, 1) ('$', 3, 160) ('price', 12, 8) ### +### ('bonds', 11, 42) ('encompasses', 19, 13) ('definition', 16, 24) ('crashing', 25, 10) ### +### ('values', 17, 35) ('ruined', 24, 30) ('is', 7, 313) ('prices', 44, 5) ('crashed', 23, 40) ### +### ('##α', 26, 31) ('hating', 46, 11) ('security', 29, 37) ('##₂', 42, 26) ('−', 51, 14) ### +### ('angrily', 39, 33) ### +############################################################################################################ +[2023-10-08 00:52:48,297][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:52:48,297][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:52:48,723][root][INFO] - Epoch: 16: Step: 1401/1557, loss[v]=0.041691, lr=0.000003, acc@1[1]=246.5/256=0.962890625, acc@1[2]=255.0/256=0.99609375 +[2023-10-08 00:54:05,866][root][INFO] - Train batch 1500 +[2023-10-08 00:54:05,867][root][INFO] - Avg. loss per last 100 batches: 0.058833 +[2023-10-08 00:54:06,578][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29085.5/29522=98.52% | mean: 0.01 | max: 5.44 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.18 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] are cold and throat viruses contageous [SEP] ### +### [P_TEXT]: [CLS] a cold is an upper respiratory infection characterized by a combination of ### +### coughing, sneezing, congestion, runny nose, headache, and / or sore throat. it is caused by one of ### +### a multitude of viruses and is very contagious. the period of contagiousness actually starts a day ### +### or two before you even feel sick. cold is an upper respiratory infection characterized by a ### +### combination of coughing, sneezing, congestion, runny nose, headache, and / or sore throat. it is ### +### caused by one of a multitude of viruses and is very contagious. [SEP] ### +### ======================================= h_v_q | Gates: 27497 ======================================= ### +### ('cold', 0, 0) ('throat', 1, 17) ('viruses', 2, 57) ('##tage', 3, 24121) ('.', 4, 12481) ### +### ('##ous', 5, 14245) ('con', 6, 93) ('familiarity', 7, 24246) ('are', 8, 6048) ### +### ('stylized', 9, 26811) ('virus', 10, 164) ('consisting', 11, 24679) ('simon', 12, 60) ### +### ('relating', 13, 22219) ('mouth', 14, 431) ('hugh', 15, 35) ('were', 16, 11801) ### +### ('include', 17, 1010) ('−', 18, 11) ('##α', 19, 54) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('cold', 0, 0) ('respiratory', 516, 1) ('coughing', 15246, 2) ('congestion', 18515, 3) ### +### ('crashing', 26, 4) ('ˈ', 57, 5) ('encompasses', 91, 6) ('hating', 33, 7) ('nose', 752, 8) ### +### ('infection', 312, 9) ('headache', 7011, 10) ('−', 18, 11) ('##ང', 131, 12) ('##ο', 44, 13) ### +### ('hesitated', 52, 14) ('upper', 1893, 15) ('##nee', 16353, 16) ('throat', 1, 17) ### +### ('unwilling', 20, 18) ('stumbled', 86, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('cold', 0, 0) ('throat', 1, 17) ('viruses', 2, 57) ('con', 6, 93) ('−', 18, 11) ### +### ('crashing', 26, 4) ('unwilling', 20, 18) ('simon', 12, 60) ('hugh', 15, 35) ('crashed', 22, 25) ### +### ('virus', 10, 164) ('ruined', 25, 28) ('hating', 33, 7) ('##α', 19, 54) ('angrily', 29, 32) ### +### ('##₂', 34, 26) ('ˈ', 57, 5) ('julian', 21, 72) ('altogether', 32, 33) ('##ο', 44, 13) ### +############################################################################################################ +[2023-10-08 00:54:06,579][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:54:06,579][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:54:07,000][root][INFO] - Epoch: 16: Step: 1501/1557, loss[v]=0.040113, lr=0.000003, acc@1[1]=246.0/256=0.9609375, acc@1[2]=252.0/256=0.984375 +[2023-10-08 00:54:50,826][root][INFO] - rank=1; last iteration 1557 +[2023-10-08 00:54:50,827][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 00:54:50,826][root][INFO] - rank=3; last iteration 1557 +[2023-10-08 00:54:50,827][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-08 00:54:50,827][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 00:54:50,827][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-08 00:54:50,829][root][INFO] - rank=2; last iteration 1557 +[2023-10-08 00:54:50,830][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 00:54:50,830][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-08 00:54:50,833][root][INFO] - rank=0; last iteration 1557 +[2023-10-08 00:54:50,833][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 00:54:50,833][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-08 00:54:50,836][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 00:54:50,836][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 00:54:50,836][root][INFO] - Epoch finished on 1 +[2023-10-08 00:54:50,836][root][INFO] - Epoch finished on 3 +[2023-10-08 00:54:50,838][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 00:54:50,838][root][INFO] - Epoch finished on 2 +[2023-10-08 00:54:50,840][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 00:54:50,840][root][INFO] - Epoch finished on 0 +[2023-10-08 00:55:05,381][root][INFO] - Saved checkpoint at ./vdr_16 +[2023-10-08 00:55:05,381][root][INFO] - Saved checkpoint at ./vdr_16 +[2023-10-08 00:55:05,381][root][INFO] - Saved checkpoint at ./vdr_16 +[2023-10-08 00:55:05,382][root][INFO] - Av Loss per epoch=0.058994 +[2023-10-08 00:55:05,382][root][INFO] - Av Loss per epoch=0.058994 +[2023-10-08 00:55:05,382][root][INFO] - epoch total (1) correct predictions=379932 +[2023-10-08 00:55:05,382][root][INFO] - Av Loss per epoch=0.058994 +[2023-10-08 00:55:05,382][root][INFO] - epoch total (1) correct predictions=379932 +[2023-10-08 00:55:05,382][root][INFO] - epoch total (2) correct predictions=391944 +[2023-10-08 00:55:05,383][root][INFO] - epoch total (2) correct predictions=391944 +[2023-10-08 00:55:05,382][root][INFO] - epoch total (1) correct predictions=379932 +[2023-10-08 00:55:05,383][root][INFO] - epoch total (2) correct predictions=391944 +[2023-10-08 00:55:05,384][root][INFO] - Saved checkpoint at ./vdr_16 +[2023-10-08 00:55:05,385][root][INFO] - Av Loss per epoch=0.058994 +[2023-10-08 00:55:05,385][root][INFO] - epoch total (1) correct predictions=379932 +[2023-10-08 00:55:05,385][root][INFO] - epoch total (2) correct predictions=391944 +[2023-10-08 00:55:05,386][root][INFO] - ***** Epoch 17 ***** +[2023-10-08 00:55:05,387][root][INFO] - ***** Epoch 17 ***** +[2023-10-08 00:55:05,387][root][INFO] - ***** Epoch 17 ***** +[2023-10-08 00:55:05,393][root][INFO] - rank=2; Iteration start +[2023-10-08 00:55:05,393][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 00:55:05,393][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 00:55:05,393][root][INFO] - rank=0; Iteration start +[2023-10-08 00:55:05,393][root][INFO] - rank=3; Iteration start +[2023-10-08 00:55:05,394][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 00:55:05,394][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 00:55:05,394][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 00:55:05,394][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 00:55:05,393][root][INFO] - ***** Epoch 17 ***** +[2023-10-08 00:55:05,395][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-08 00:55:05,396][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-08 00:55:05,396][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-08 00:55:05,399][root][INFO] - rank=1; Iteration start +[2023-10-08 00:55:05,399][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 00:55:05,399][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 00:55:05,401][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-08 00:55:06,377][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29089.5/29522=98.53% | mean: 0.01 | max: 5.41 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.34 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] when r & d tax credit extension 2015 was passed may 2015 [SEP] ### +### [P_TEXT]: [CLS] house passes permanent extension of the r & d tax credit. 05 / 22 / 2015 a · r & d, ### +### tax insight. on may 20th, the us house of representatives passed hr 880, making the research & ### +### development tax credit permanent. ouse passes permanent extension of the r & d tax credit. 05 / 22 ### +### / 2015 a · r & d, tax insight. on may 20th, the us house of representatives passed hr 880, making ### +### the research & development tax credit permanent. [SEP] ### +### ======================================= h_v_q | Gates: 27586 ======================================= ### +### ('2015', 0, 21) ('tax', 1, 2) ('passed', 2, 46) ('extension', 3, 8) ('may', 4, 48) ### +### ('credit', 5, 14) ('&', 6, 42) ('r', 7, 19) ('d', 8, 56) ('february', 9, 818) ('.', 10, 9188) ### +### ('stylized', 11, 29236) ('1939', 12, 11188) ('june', 13, 77) ('when', 14, 186) ('july', 15, 204) ### +### ('march', 16, 1619) ('familiarity', 17, 25263) ('2016', 18, 1455) ('knew', 19, 106) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('·', 409, 0) ('ˈ', 107, 1) ('tax', 1, 2) ('research', 2772, 3) ('permanent', 8953, 4) ### +### ('ou', 27737, 5) ('development', 70, 6) ('cyrillic', 183, 7) ('extension', 3, 8) ('−', 58, 9) ### +### ('crashing', 30, 10) ('##ο', 92, 11) ('insight', 20990, 12) ('taxes', 275, 13) ('credit', 5, 14) ### +### ('hesitated', 80, 15) ('hating', 182, 16) ('unwilling', 77, 17) ('house', 3082, 18) ('r', 7, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('tax', 1, 2) ('2015', 0, 21) ('extension', 3, 8) ('credit', 5, 14) ('passed', 2, 46) ### +### ('may', 4, 48) ('&', 6, 42) ('r', 7, 19) ('d', 8, 56) ('june', 13, 77) ('crashing', 30, 10) ### +### ('pass', 21, 38) ('when', 14, 186) ('july', 15, 204) ('−', 58, 9) ('crashed', 36, 36) ### +### ('##α', 38, 35) ('development', 70, 6) ('knew', 19, 106) ('ˈ', 107, 1) ### +############################################################################################################ +[2023-10-08 00:55:06,378][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:55:06,378][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:55:06,775][root][INFO] - Epoch: 17: Step: 1/1557, loss[v]=0.053785, lr=0.000003, acc@1[1]=245.5/256=0.958984375, acc@1[2]=253.0/256=0.98828125 +[2023-10-08 00:56:23,511][root][INFO] - Train batch 100 +[2023-10-08 00:56:23,512][root][INFO] - Avg. loss per last 100 batches: 0.056610 +[2023-10-08 00:56:24,196][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29046.7/29522=98.39% | mean: 0.01 | max: 5.78 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.4/29522=100.00% | mean: 0.14 | max: 6.60 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] highest bilirubin level allowed [SEP] ### +### [P_TEXT]: [CLS] my best memory was that it was around 48mg / ml.... a ·. the highest level of the ### +### serum bilirubin i have ever seen was slightly more than 500 mcmol / l ( about 30 mg / dl ) in case ### +### of the intrahepatic fistula between a large bile duct and a branch of the vena cava inf.... a ·. ### +### perhaps there is a population difference and other variable e. g. drugs etc. [SEP] ### +### ======================================= h_v_q | Gates: 27867 ======================================= ### +### ('bi', 0, 25) ('allowed', 1, 8980) ('##ru', 2, 177) ('##li', 3, 95) ('highest', 4, 4) ### +### ('##bin', 5, 31) ('level', 6, 13) ('allows', 7, 15120) ('levels', 8, 33) ('high', 9, 220) ### +### ('.', 10, 7369) ('permitted', 11, 11867) ('volume', 12, 1209) ('stylized', 13, 28654) ### +### ('familiarity', 14, 26560) ('allowing', 15, 11990) ('consisting', 16, 26659) ('higher', 17, 211) ### +### ('plural', 18, 11289) ('allow', 19, 9603) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('serum', 14381, 0) ('bile', 20819, 1) ('·', 438, 2) ('ˈ', 23, 3) ('highest', 4, 4) ### +### ('hating', 60, 5) ('hesitated', 31, 6) ('mc', 9224, 7) ('##ο', 79, 8) ('dl', 28350, 9) ### +### ('unwilling', 32, 10) ('wingspan', 107, 11) ('duct', 17293, 12) ('level', 6, 13) ('fist', 8266, 14) ### +### ('ml', 21922, 15) ('intra', 13737, 16) ('slightly', 5974, 17) ('30', 212, 18) ('cyrillic', 227, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('bi', 0, 25) ('highest', 4, 4) ('##bin', 5, 31) ('level', 6, 13) ('##li', 3, 95) ('##ru', 2, 177) ### +### ('levels', 8, 33) ('ˈ', 23, 3) ('high', 9, 220) ('hesitated', 31, 6) ('unwilling', 32, 10) ### +### ('crashing', 30, 27) ('crashed', 25, 57) ('ruined', 28, 42) ('angrily', 34, 28) ('##α', 27, 46) ### +### ('hating', 60, 5) ('allowed', 1, 8980) ('simon', 29, 60) ('julian', 26, 86) ### +############################################################################################################ +[2023-10-08 00:56:24,196][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:56:24,196][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:56:24,621][root][INFO] - Epoch: 17: Step: 101/1557, loss[v]=0.040266, lr=0.000003, acc@1[1]=249.0/256=0.97265625, acc@1[2]=252.0/256=0.984375 +[2023-10-08 00:57:41,691][root][INFO] - Train batch 200 +[2023-10-08 00:57:41,694][root][INFO] - Avg. loss per last 100 batches: 0.057146 +[2023-10-08 00:57:42,401][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29107.4/29522=98.60% | mean: 0.01 | max: 5.53 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.25 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a shift kit [SEP] ### +### [P_TEXT]: [CLS] a shift kit will set your transmission such that you get optimum performance and ### +### you will also get longer service life out of the various transmission system components. improved ### +### shifting, reduced slippage, and improved power output - those are just some benefits you'll get ### +### when you use a shift kit. [SEP] ### +### ======================================= h_v_q | Gates: 26232 ======================================= ### +### ('shift', 0, 0) ('kit', 1, 1) ('shifts', 2, 6) ('encompasses', 3, 93) ('is', 4, 6700) ### +### ('kits', 5, 15) ('definition', 6, 4001) ('refers', 7, 25317) ('stylized', 8, 29195) ### +### ('encyclopedia', 9, 17620) ('change', 10, 124) ('relating', 11, 25230) ('familiarity', 12, 23900) ### +### ('shifting', 13, 26) ('consisting', 14, 21871) ('plural', 15, 22420) ('noun', 16, 28250) ### +### ('genus', 17, 18879) ('uniform', 18, 171) ('especially', 19, 6799) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('shift', 0, 0) ('kit', 1, 1) ('transmission', 3091, 2) ('ˈ', 592, 3) ('crashing', 195, 4) ### +### ('hating', 377, 5) ('shifts', 2, 6) ('benefits', 8050, 7) ('improved', 19508, 8) ('##ο', 378, 9) ### +### ('unwilling', 123, 10) ('slip', 638, 11) ('−', 158, 12) ('cyrillic', 376, 13) ### +### ('hesitated', 341, 14) ('kits', 5, 15) ('longer', 12794, 16) ('set', 464, 17) ('life', 1277, 18) ### +### ('enhanced', 5226, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('shift', 0, 0) ('kit', 1, 1) ('shifts', 2, 6) ('kits', 5, 15) ('encompasses', 3, 93) ### +### ('shifting', 13, 26) ('change', 10, 124) ('uniform', 18, 171) ('shifted', 27, 134) ('move', 32, 83) ### +### ('turn', 33, 102) ('unwilling', 123, 10) ('components', 98, 30) ('crashing', 195, 4) ### +### ('ruined', 88, 37) ('shoved', 121, 27) ('−', 158, 12) ('##α', 101, 40) ('ছ', 130, 29) ### +### ('angrily', 126, 41) ### +############################################################################################################ +[2023-10-08 00:57:42,401][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:57:42,401][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:57:42,822][root][INFO] - Epoch: 17: Step: 201/1557, loss[v]=0.069229, lr=0.000003, acc@1[1]=245.0/256=0.95703125, acc@1[2]=249.0/256=0.97265625 +[2023-10-08 00:59:00,731][root][INFO] - Train batch 300 +[2023-10-08 00:59:00,734][root][INFO] - Avg. loss per last 100 batches: 0.063649 +[2023-10-08 00:59:01,437][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29063.8/29522=98.45% | mean: 0.01 | max: 5.67 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.14 | max: 6.43 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] effect of sodium on kidneys [SEP] ### +### [P_TEXT]: [CLS] eating salt raises the amount of sodium in your bloodstream and wrecks the delicate ### +### balance, reducing the ability of your kidneys to remove the water. the result is a higher blood ### +### pressure due to the extra fluid and extra strain on the delicate blood vessels leading to the ### +### kidneys. [SEP] ### +### ======================================= h_v_q | Gates: 26756 ======================================= ### +### ('sodium', 0, 2) ('kidney', 1, 3) ('.', 2, 14268) ('effect', 3, 774) ('##s', 4, 196) ### +### ('effects', 5, 275) ('familiarity', 6, 27116) ('consisting', 7, 26569) ('stylized', 8, 28596) ### +### ('affect', 9, 443) ('relating', 10, 25503) ('ability', 11, 23) ('of', 12, 14904) ('onto', 13, 615) ### +### ('simon', 14, 40) ('impact', 15, 1500) ('##स', 16, 180) ('##₂', 17, 20) ('damage', 18, 2302) ### +### ('##α', 19, 47) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('delicate', 249, 0) ('salt', 90, 1) ('sodium', 0, 2) ('kidney', 1, 3) ('eating', 8320, 4) ### +### ('hating', 33, 5) ('ˈ', 64, 6) ('wreck', 20909, 7) ('crashing', 68, 8) ('ruins', 156, 9) ### +### ('balance', 190, 10) ('##ο', 79, 11) ('sharply', 42, 12) ('water', 48, 13) ('hesitated', 61, 14) ### +### ('cyrillic', 320, 15) ('−', 26, 16) ('unwilling', 25, 17) ('wingspan', 285, 18) ('crashed', 54, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('sodium', 0, 2) ('kidney', 1, 3) ('##s', 4, 196) ('effects', 5, 275) ('effect', 3, 774) ### +### ('ability', 11, 23) ('simon', 14, 40) ('##₂', 17, 20) ('hating', 33, 5) ('salt', 90, 1) ### +### ('unwilling', 25, 17) ('−', 26, 16) ('ˈ', 64, 6) ('sharply', 42, 12) ('delicate', 249, 0) ### +### ('ruined', 29, 30) ('water', 48, 13) ('angrily', 24, 36) ('##α', 19, 47) ('crashing', 68, 8) ### +############################################################################################################ +[2023-10-08 00:59:01,438][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 00:59:01,438][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 00:59:01,860][root][INFO] - Epoch: 17: Step: 301/1557, loss[v]=0.046188, lr=0.000003, acc@1[1]=247.0/256=0.96484375, acc@1[2]=252.0/256=0.984375 +[2023-10-08 01:00:19,614][root][INFO] - Train batch 400 +[2023-10-08 01:00:19,617][root][INFO] - Avg. loss per last 100 batches: 0.058866 +[2023-10-08 01:00:20,351][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29098.9/29522=98.57% | mean: 0.01 | max: 5.44 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.4/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.4/29522=100.00% | mean: 0.14 | max: 6.05 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] rasmus definition [SEP] ### +### [P_TEXT]: [CLS] a rasmus is a small furry nocturnal creature that's a cross between a koala, sloath ### +### and has large eyes like a bush baby. rasmus or'the rasmus'as their known in plural, live on a diet ### +### of plants and leaves. these creatures are mostly found in woodland areas. [SEP] ### +### ======================================= h_v_q | Gates: 26086 ======================================= ### +### ('##mus', 0, 2) ('ras', 1, 1) ('definition', 2, 38) ('defined', 3, 475) ('familiarity', 4, 27356) ### +### ('##º', 5, 28469) ('relating', 6, 23522) ('noun', 7, 16557) ('stylized', 8, 26511) ### +### ('plural', 9, 79) ('consisting', 10, 21968) ('encyclopedia', 11, 11379) ('.', 12, 11536) ### +### ('refers', 13, 9419) ('something', 14, 10583) ('or', 15, 18118) ('term', 16, 8100) ### +### ('##imus', 17, 724) ('definitions', 18, 43) ('specified', 19, 16394) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('nocturnal', 15587, 0) ('ras', 1, 1) ('##mus', 0, 2) ('furry', 25815, 3) ('creature', 4732, 4) ### +### ('creatures', 3992, 5) ('ˈ', 294, 6) ('eyes', 3725, 7) ('sl', 10438, 8) ('##ο', 437, 9) ### +### ('##ala', 17579, 10) ('crashing', 114, 11) ('hating', 945, 12) ('encompasses', 47, 13) ### +### ('ko', 662, 14) ('##ང', 171, 15) ('hesitated', 167, 16) ('ছ', 191, 17) ('stumbled', 360, 18) ### +### ('−', 68, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##mus', 0, 2) ('ras', 1, 1) ('definition', 2, 38) ('plural', 9, 79) ('defined', 3, 475) ### +### ('definitions', 18, 43) ('encompasses', 47, 13) ('crashed', 45, 21) ('meaning', 23, 95) ### +### ('−', 68, 19) ('crashing', 114, 11) ('ruined', 71, 30) ('##α', 65, 35) ('sharply', 83, 28) ### +### ('shoved', 85, 33) ('angrily', 106, 26) ('gideon', 124, 24) ('ˈ', 294, 6) ('##ང', 171, 15) ### +### ('hesitated', 167, 16) ### +############################################################################################################ +[2023-10-08 01:00:20,352][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:00:20,352][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:00:20,781][root][INFO] - Epoch: 17: Step: 401/1557, loss[v]=0.088588, lr=0.000003, acc@1[1]=240.0/256=0.9375, acc@1[2]=249.0/256=0.97265625 +[2023-10-08 01:01:38,773][root][INFO] - Train batch 500 +[2023-10-08 01:01:38,776][root][INFO] - Avg. loss per last 100 batches: 0.059644 +[2023-10-08 01:01:39,515][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29109.2/29522=98.60% | mean: 0.01 | max: 5.73 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.14 | max: 6.32 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] where did the first crusade begin [SEP] ### +### [P_TEXT]: [CLS] best answer : immediate cause. the immediate cause of the first crusade was alexius ### +### i's appeal to pope urban ii for mercenaries to help him resist muslim advances into territory of ### +### the byzantine empire. [SEP] ### +### ======================================= h_v_q | Gates: 26836 ======================================= ### +### ('crusade', 0, 0) ('began', 1, 18992) ('first', 2, 10) ('begins', 3, 15375) ('.', 4, 12027) ### +### ('begin', 5, 1978) ('founded', 6, 1297) ('america', 7, 23575) ('downtown', 8, 2805) ### +### ('washington', 9, 12184) ('china', 10, 8952) ('where', 11, 422) ('australia', 12, 6203) ### +### ('campaign', 13, 1487) ('started', 14, 218) ('beginning', 15, 94) ('familiarity', 16, 28084) ### +### ('did', 17, 194) ('africa', 18, 2946) ('europe', 19, 11218) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('crusade', 0, 0) ('pope', 2994, 1) ('immediate', 9862, 2) ('alex', 993, 3) ### +### ('mercenaries', 10894, 4) ('byzantine', 830, 5) ('cause', 3078, 6) ('ˈ', 654, 7) ('##ο', 608, 8) ### +### ('wingspan', 550, 9) ('first', 2, 10) ('##ང', 704, 11) ('stumbled', 606, 12) ('unwilling', 248, 13) ### +### ('causes', 4571, 14) ('−', 387, 15) ('crashing', 425, 16) ('hating', 652, 17) ('cyrillic', 490, 18) ### +### ('urban', 1229, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('crusade', 0, 0) ('first', 2, 10) ('beginning', 15, 94) ('1st', 48, 22) ('start', 24, 82) ### +### ('quest', 39, 60) ('started', 14, 218) ('did', 17, 194) ('where', 11, 422) ('church', 63, 84) ### +### ('founded', 6, 1297) ('simon', 91, 62) ('##₂', 175, 21) ('somewhere', 37, 169) ('early', 60, 112) ### +### ('unwilling', 248, 13) ('knew', 35, 230) ('begin', 5, 1978) ('##α', 187, 42) ('jerusalem', 21, 616) ### +############################################################################################################ +[2023-10-08 01:01:39,515][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:01:39,515][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:01:39,920][root][INFO] - Epoch: 17: Step: 501/1557, loss[v]=0.046575, lr=0.000003, acc@1[1]=243.0/256=0.94921875, acc@1[2]=250.5/256=0.978515625 +[2023-10-08 01:02:57,816][root][INFO] - Train batch 600 +[2023-10-08 01:02:57,819][root][INFO] - Avg. loss per last 100 batches: 0.056126 +[2023-10-08 01:02:58,528][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29075.0/29522=98.49% | mean: 0.01 | max: 5.42 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.14 | max: 6.24 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] price for a gucci belt [SEP] ### +### [P_TEXT]: [CLS] $ 219. 99 for a gucci men's web belt in brown ( $ 295 list price ). 1 available ### +### sizes : 85 cm ( 34 in ), 90 cm ( 36 in ), 95 cm ( 38 in ), 100 cm ( 40 in ), 105 cm ( 42 in ), 110 ### +### cm ( 44 in ). 2 ruthenium hardware. [SEP] ### +### ======================================= h_v_q | Gates: 26893 ======================================= ### +### ('##cci', 0, 3) ('belt', 1, 0) ('$', 2, 15) ('gu', 3, 2) ('##£', 4, 23553) ('price', 5, 1) ### +### ('stylized', 6, 28590) ('familiarity', 7, 26821) ('prices', 8, 11) ('.', 9, 7597) ### +### ('relating', 10, 28062) ('cents', 11, 2537) ('plural', 12, 20892) ('cost', 13, 9) ### +### ('commonwealth', 14, 9071) ('consisting', 15, 26069) ('currency', 16, 591) ('unwilling', 17, 19) ### +### ('physics', 18, 5271) ('encyclopedia', 19, 14649) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('belt', 1, 0) ('price', 5, 1) ('gu', 3, 2) ('##cci', 0, 3) ('cm', 17350, 4) ('web', 4744, 5) ### +### ('hardware', 1090, 6) ('ˈ', 49, 7) ('brown', 2284, 8) ('cost', 13, 9) ('ruth', 2525, 10) ### +### ('prices', 8, 11) ('size', 484, 12) ('sizes', 6073, 13) ('men', 676, 14) ('$', 2, 15) ### +### ('hating', 73, 16) ('##ο', 159, 17) ('hesitated', 23, 18) ('unwilling', 17, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##cci', 0, 3) ('belt', 1, 0) ('gu', 3, 2) ('$', 2, 15) ('price', 5, 1) ('prices', 8, 11) ### +### ('cost', 13, 9) ('unwilling', 17, 19) ('hesitated', 23, 18) ('ˈ', 49, 7) ('belts', 24, 38) ### +### ('angrily', 27, 32) ('##α', 29, 42) ('##大', 36, 29) ('simon', 31, 39) ('hating', 73, 16) ### +### ('crashing', 47, 23) ('crashed', 25, 64) ('−', 33, 46) ('afraid', 57, 26) ### +############################################################################################################ +[2023-10-08 01:02:58,528][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:02:58,528][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:02:58,933][root][INFO] - Epoch: 17: Step: 601/1557, loss[v]=0.043765, lr=0.000003, acc@1[1]=248.0/256=0.96875, acc@1[2]=253.0/256=0.98828125 +[2023-10-08 01:04:16,351][root][INFO] - Train batch 700 +[2023-10-08 01:04:16,354][root][INFO] - Avg. loss per last 100 batches: 0.056708 +[2023-10-08 01:04:17,065][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29081.8/29522=98.51% | mean: 0.01 | max: 5.72 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.14 | max: 6.28 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a mala necklace [SEP] ### +### [P_TEXT]: [CLS] a mala is a string of 108 beads with one bead as the summit bead called a'sumeru '. ### +### it is a tool used to keep your mind on the meditation practice. malas are generally made from ### +### different materials such as tulsi ( basil ) wood, sandal wood, rudraksh seeds or crystal. [SEP] ### +### ======================================= h_v_q | Gates: 26210 ======================================= ### +### ('mala', 0, 0) ('necklace', 1, 85) ('is', 2, 207) ('relating', 3, 23621) ('encompasses', 4, 9) ### +### ('definition', 5, 35) ('pendant', 6, 230) ('.', 7, 14802) ('jewelry', 8, 104) ### +### ('encyclopedia', 9, 2395) ('familiarity', 10, 23975) ('refers', 11, 8707) ('##sam', 12, 26626) ### +### ('noun', 13, 14707) ('stands', 14, 6218) ('stylized', 15, 28679) ('consisting', 16, 17956) ### +### ('a', 17, 565) ('plural', 18, 16647) ('narrow', 19, 11173) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('mala', 0, 0) ('beads', 298, 1) ('bea', 3391, 2) ('string', 283, 3) ('meditation', 5067, 4) ### +### ('summit', 9950, 5) ('sum', 1575, 6) ('ˈ', 781, 7) ('##ο', 363, 8) ('encompasses', 4, 9) ### +### ('crashing', 334, 10) ('basil', 8363, 11) ('unwilling', 369, 12) ('hesitated', 320, 13) ### +### ('hating', 325, 14) ('−', 137, 15) ('made', 235, 16) ('##ང', 537, 17) ('##₂', 142, 18) ### +### ('practice', 1085, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('mala', 0, 0) ('necklace', 1, 85) ('encompasses', 4, 9) ('definition', 5, 35) ('is', 2, 207) ### +### ('jewelry', 8, 104) ('pendant', 6, 230) ('ana', 25, 110) ('beads', 298, 1) ('string', 283, 3) ### +### ('called', 68, 47) ('−', 137, 15) ('##₂', 142, 18) ('ruined', 91, 40) ('##α', 110, 33) ### +### ('simon', 78, 59) ('ছ', 134, 29) ('a', 17, 565) ('made', 235, 16) ('crashing', 334, 10) ### +############################################################################################################ +[2023-10-08 01:04:17,066][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:04:17,066][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:04:17,489][root][INFO] - Epoch: 17: Step: 701/1557, loss[v]=0.041539, lr=0.000003, acc@1[1]=247.0/256=0.96484375, acc@1[2]=253.5/256=0.990234375 +[2023-10-08 01:05:35,200][root][INFO] - Train batch 800 +[2023-10-08 01:05:35,202][root][INFO] - Avg. loss per last 100 batches: 0.062253 +[2023-10-08 01:05:35,929][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29079.7/29522=98.50% | mean: 0.01 | max: 5.54 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.14 | max: 6.17 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] who invented the sport of rowing [SEP] ### +### [P_TEXT]: [CLS] the development of rowing equipment by bill miller this page contains information ### +### about the development of rowing equipment beginning about the end of the 18th century. [SEP] ### +### ======================================= h_v_q | Gates: 26443 ======================================= ### +### ('rowing', 0, 0) ('invented', 1, 838) ('.', 2, 7720) ('sport', 3, 156) ('whose', 4, 230) ### +### ('designed', 5, 13762) ('developed', 6, 138) ('created', 7, 1368) ('founded', 8, 189) ### +### ('who', 9, 54) ('introduced', 10, 3211) ('innovation', 11, 574) ('boat', 12, 10) ### +### ('invention', 13, 805) ('of', 14, 17915) ('opened', 15, 20215) ('sailing', 16, 32) ### +### ('sports', 17, 154) ('discovered', 18, 508) ('patent', 19, 17930) ### +### ======================================= h_v_p | Gates: 29520 ======================================= ### +### ('rowing', 0, 0) ('miller', 1368, 1) ('equipment', 517, 2) ('development', 104, 3) ('##ο', 759, 4) ### +### ('ˈ', 1754, 5) ('crashing', 300, 6) ('stumbled', 1428, 7) ('##ང', 735, 8) ('sharply', 1040, 9) ### +### ('boat', 12, 10) ('hating', 815, 11) ('wingspan', 1196, 12) ('afraid', 643, 13) ('−', 407, 14) ### +### ('bill', 3094, 15) ('unwilling', 480, 16) ('ছ', 419, 17) ('century', 24, 18) ('hesitated', 987, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('rowing', 0, 0) ('sport', 3, 156) ('boat', 12, 10) ('invented', 1, 838) ('who', 9, 54) ### +### ('whose', 4, 230) ('developed', 6, 138) ('founded', 8, 189) ('sailing', 16, 32) ('century', 24, 18) ### +### ('rower', 34, 23) ('swimming', 39, 47) ('development', 104, 3) ('innovation', 11, 574) ### +### ('sports', 17, 154) ('hull', 40, 69) ('.', 2, 7720) ('created', 7, 1368) ('invention', 13, 805) ### +### ('cox', 29, 158) ### +############################################################################################################ +[2023-10-08 01:05:35,929][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:05:35,929][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:05:36,358][root][INFO] - Epoch: 17: Step: 801/1557, loss[v]=0.076060, lr=0.000003, acc@1[1]=239.5/256=0.935546875, acc@1[2]=249.0/256=0.97265625 +[2023-10-08 01:06:54,104][root][INFO] - Train batch 900 +[2023-10-08 01:06:54,105][root][INFO] - Avg. loss per last 100 batches: 0.058670 +[2023-10-08 01:06:54,836][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29084.5/29522=98.52% | mean: 0.01 | max: 5.67 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.4/29522=100.00% | mean: 0.14 | max: 6.25 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is barabbas first name? [SEP] ### +### [P_TEXT]: [CLS] some thought he was the ason of god, a the true son of the true father, and the ### +### long - awaited messiah. yet, for all the good jesus was known for, the mob wanted him to die. this ### +### great, good man would die in the place of a common criminal. some unreliable, written sources even ### +### claim that barabbasa first name was jesus. as god ordered it, the only son of the heavenly father, ### +### the promised messiah, would die in the place of another ason of fathera, barabbas. jesus would die ### +### so the sinful son could go free. what barabbas had heard was true. [SEP] ### +### ======================================= h_v_q | Gates: 27718 ======================================= ### +### ('bar', 0, 5) ('##ab', 1, 13) ('##bas', 2, 17) ('first', 3, 46) ('name', 4, 68) ('is', 5, 2503) ### +### ('stylized', 6, 27803) ('surname', 7, 480) ('familiarity', 8, 25747) ('consisting', 9, 23735) ### +### ('bars', 10, 97) ('relating', 11, 25340) ('encompasses', 12, 169) ('plural', 13, 9207) ### +### ('1st', 14, 55) ('genus', 15, 14600) ('mathematics', 16, 23736) ('noun', 17, 26244) ### +### ('##sam', 18, 28067) ('refers', 19, 18237) ### +### ======================================= h_v_p | Gates: 29520 ======================================= ### +### ('jesus', 1910, 0) ('messiah', 21923, 1) ('##on', 10736, 2) ('mob', 3370, 3) ('ˈ', 47, 4) ### +### ('bar', 0, 5) ('##ο', 152, 6) ('promised', 12834, 7) ('heavenly', 26330, 8) ('die', 18504, 9) ### +### ('son', 1687, 10) ('as', 9119, 11) ('cyrillic', 220, 12) ('##ab', 1, 13) ('##ང', 75, 14) ### +### ('stumbled', 58, 15) ('wingspan', 99, 16) ('##bas', 2, 17) ('crashing', 28, 18) ('hating', 90, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('bar', 0, 5) ('##ab', 1, 13) ('##bas', 2, 17) ('first', 3, 46) ('name', 4, 68) ('bars', 10, 97) ### +### ('1st', 14, 55) ('crashing', 28, 18) ('surname', 7, 480) ('encompasses', 12, 169) ('ˈ', 47, 4) ### +### ('−', 31, 25) ('##α', 40, 21) ('nickname', 21, 75) ('crashed', 39, 28) ('unwilling', 50, 20) ### +### ('stumbled', 58, 15) ('gideon', 45, 30) ('ছ', 42, 38) ('##ང', 75, 14) ### +############################################################################################################ +[2023-10-08 01:06:54,837][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:06:54,837][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:06:55,259][root][INFO] - Epoch: 17: Step: 901/1557, loss[v]=0.062153, lr=0.000003, acc@1[1]=245.5/256=0.958984375, acc@1[2]=251.5/256=0.982421875 +[2023-10-08 01:08:13,752][root][INFO] - Train batch 1000 +[2023-10-08 01:08:13,754][root][INFO] - Avg. loss per last 100 batches: 0.056926 +[2023-10-08 01:08:14,462][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29084.4/29522=98.52% | mean: 0.01 | max: 5.57 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.14 | max: 6.28 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what station is the sinner series on [SEP] ### +### [P_TEXT]: [CLS] usa picks up jessica biel series athe sinnera. usa network has picked up the ### +### anthology series the sinner from sibling universal cable productions ( ucp ). the crime thriller is ### +### executive produced by jessica biel, who stars in it as well. bill pullman is in the cast as well as ### +### a rogue investigator. [SEP] ### +### ======================================= h_v_q | Gates: 27732 ======================================= ### +### ('sinn', 0, 0) ('##er', 1, 12) ('series', 2, 31) ('station', 3, 3111) ('stations', 4, 7546) ### +### ('aired', 5, 6236) ('broadcasting', 6, 5800) ('radio', 7, 1579) ('television', 8, 256) ### +### ('broadcast', 9, 684) ('encompasses', 10, 410) ('transmitter', 11, 1204) ('familiarity', 12, 20959) ### +### ('on', 13, 18444) ('.', 14, 11785) ('stylized', 15, 27997) ('consisting', 16, 22924) ### +### ('network', 17, 18) ('sin', 18, 357) ('channel', 19, 40) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('sinn', 0, 0) ('sibling', 11078, 1) ('investigator', 12090, 2) ('thriller', 656, 3) ### +### ('rogue', 8356, 4) ('jessica', 7331, 5) ('usa', 5788, 6) ('uc', 10476, 7) ('##era', 2878, 8) ### +### ('ˈ', 321, 9) ('crime', 3021, 10) ('cast', 12460, 11) ('##er', 1, 12) ('##ο', 151, 13) ### +### ('##el', 2629, 14) ('bi', 9954, 15) ('crashing', 79, 16) ('stumbled', 405, 17) ('network', 17, 18) ### +### ('cable', 514, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('sinn', 0, 0) ('##er', 1, 12) ('series', 2, 31) ('network', 17, 18) ('channel', 19, 40) ### +### ('television', 8, 256) ('crashing', 79, 16) ('station', 3, 3111) ('crashed', 67, 41) ### +### ('encompasses', 10, 410) ('−', 116, 26) ('##ο', 151, 13) ('##ང', 145, 27) ('broadcast', 9, 684) ### +### ('unwilling', 166, 20) ('ruined', 86, 63) ('##大', 146, 34) ('##α', 103, 49) ('sharply', 174, 30) ### +### ('hesitated', 198, 24) ### +############################################################################################################ +[2023-10-08 01:08:14,462][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:08:14,462][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:08:14,885][root][INFO] - Epoch: 17: Step: 1001/1557, loss[v]=0.043599, lr=0.000002, acc@1[1]=250.5/256=0.978515625, acc@1[2]=253.5/256=0.990234375 +[2023-10-08 01:09:32,646][root][INFO] - Train batch 1100 +[2023-10-08 01:09:32,649][root][INFO] - Avg. loss per last 100 batches: 0.057318 +[2023-10-08 01:09:33,352][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29151.5/29522=98.75% | mean: 0.01 | max: 5.28 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.00 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] bitcoin today price [SEP] ### +### [P_TEXT]: [CLS] bitcoin prices today. chinese investors drive bitcoin value to record high. the ### +### total value of all bitcoins in circulation hit a record high above $ 14 billion on thursday, as the ### +### web - based digital currency jumped 5 percent on the day to its highest levels in three years after ### +### more than doubling in price this year. [SEP] ### +### ======================================= h_v_q | Gates: 27492 ======================================= ### +### ('##co', 0, 38) ('today', 1, 18) ('bit', 2, 1) ('$', 3, 20) ('##in', 4, 54) ('price', 5, 6) ### +### ('##£', 6, 20125) ('tonight', 7, 256) ('day', 8, 72) ('familiarity', 9, 25480) ### +### ('stylized', 10, 29091) ('prices', 11, 0) ('yesterday', 12, 564) ('relating', 13, 25339) ### +### ('currently', 14, 306) ('simon', 15, 64) ('tomorrow', 16, 279) ('.', 17, 12081) ### +### ('plural', 18, 17703) ('consisting', 19, 25154) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('prices', 11, 0) ('bit', 2, 1) ('ˈ', 47, 2) ('currency', 32, 3) ('value', 72, 4) ### +### ('chinese', 3456, 5) ('price', 5, 6) ('##ο', 71, 7) ('wingspan', 85, 8) ('hesitated', 37, 9) ### +### ('digital', 77, 10) ('crashing', 38, 11) ('cyrillic', 308, 12) ('hating', 75, 13) ### +### ('stumbled', 96, 14) ('hit', 904, 15) ('##ང', 70, 16) ('thursday', 4820, 17) ('today', 1, 18) ### +### ('−', 20, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('bit', 2, 1) ('today', 1, 18) ('##co', 0, 38) ('price', 5, 6) ('$', 3, 20) ('##in', 4, 54) ### +### ('prices', 11, 0) ('day', 8, 72) ('tonight', 7, 256) ('−', 20, 19) ('currency', 32, 3) ### +### ('crashed', 24, 28) ('ˈ', 47, 2) ('simon', 15, 64) ('unwilling', 27, 24) ('##ins', 30, 21) ### +### ('hesitated', 37, 9) ('crashing', 38, 11) ('value', 72, 4) ('##ο', 71, 7) ### +############################################################################################################ +[2023-10-08 01:09:33,352][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:09:33,352][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:09:33,778][root][INFO] - Epoch: 17: Step: 1101/1557, loss[v]=0.055039, lr=0.000002, acc@1[1]=245.0/256=0.95703125, acc@1[2]=251.5/256=0.982421875 +[2023-10-08 01:10:52,151][root][INFO] - Train batch 1200 +[2023-10-08 01:10:52,154][root][INFO] - Avg. loss per last 100 batches: 0.057538 +[2023-10-08 01:10:52,885][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29071.0/29522=98.47% | mean: 0.01 | max: 5.48 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.19 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a car spindle [SEP] ### +### [P_TEXT]: [CLS] 1 spindle ( textiles ), a device to spin fibers into thread. 2 spindle ( tool ), is ### +### the main rotating part of a machine tool, woodworking machine, etc. 3 segmented spindle, a spindle ### +### consisting of discrete elements that can be assembled and disassembled on the fly. [SEP] ### +### ======================================= h_v_q | Gates: 26690 ======================================= ### +### ('spin', 0, 1) ('##dle', 1, 0) ('car', 2, 9469) ('encompasses', 3, 45) ('definition', 4, 63) ### +### ('is', 5, 1413) ('stylized', 6, 28389) ('refers', 7, 16470) ('familiarity', 8, 25915) ### +### ('consisting', 9, 1489) ('plural', 10, 17088) ('relating', 11, 17997) ('noun', 12, 21826) ### +### ('stands', 13, 3427) ('##sam', 14, 26591) ('encyclopedia', 15, 9262) ('genus', 16, 16983) ### +### ('cars', 17, 5283) ('a', 18, 14776) ('term', 19, 8160) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##dle', 1, 0) ('spin', 0, 1) ('thread', 1342, 2) ('fly', 3771, 3) ('##mbled', 18385, 4) ### +### ('ˈ', 130, 5) ('rotating', 13338, 6) ('crashing', 49, 7) ('fibers', 6774, 8) ('tool', 143, 9) ### +### ('segment', 1152, 10) ('##ο', 140, 11) ('textiles', 1630, 12) ('device', 823, 13) ### +### ('hating', 47, 14) ('textile', 323, 15) ('unwilling', 40, 16) ('stumbled', 141, 17) ('−', 55, 18) ### +### ('machine', 277, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('spin', 0, 1) ('##dle', 1, 0) ('encompasses', 3, 45) ('definition', 4, 63) ('spinning', 20, 50) ### +### ('unwilling', 40, 16) ('crashing', 49, 7) ('hating', 47, 14) ('ruined', 41, 48) ('−', 55, 18) ### +### ('is', 5, 1413) ('car', 2, 9469) ('##₂', 70, 25) ('ˈ', 130, 5) ('crashed', 60, 42) ('hugh', 52, 53) ### +### ('simon', 44, 68) ('gideon', 80, 28) ('##α', 66, 41) ('shoved', 75, 32) ### +############################################################################################################ +[2023-10-08 01:10:52,885][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:10:52,886][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:10:53,310][root][INFO] - Epoch: 17: Step: 1201/1557, loss[v]=0.039547, lr=0.000002, acc@1[1]=243.5/256=0.951171875, acc@1[2]=254.0/256=0.9921875 +[2023-10-08 01:12:10,056][root][INFO] - Train batch 1300 +[2023-10-08 01:12:10,058][root][INFO] - Avg. loss per last 100 batches: 0.055645 +[2023-10-08 01:12:10,743][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29113.9/29522=98.62% | mean: 0.01 | max: 5.30 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.3/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.7/29522=100.00% | mean: 0.15 | max: 6.32 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what vehicles are included in airbag recall [SEP] ### +### [P_TEXT]: [CLS] ford expands its recall to 816, 000 ford, lincoln, and mercury vehicle made in ### +### north america, including 654, 695 sold in the u. s. most vehicles were included included in prior ### +### recall actions, but this move adds the passenger - side airbag inflators. [SEP] ### +### ======================================= h_v_q | Gates: 27189 ======================================= ### +### ('recall', 0, 0) ('##bag', 1, 135) ('air', 2, 35) ('vehicles', 3, 4) ('vehicle', 4, 3) ### +### ('included', 5, 59) ('recalled', 6, 6) ('.', 7, 15452) ('includes', 8, 406) ### +### ('familiarity', 9, 25266) ('include', 10, 182) ('aircraft', 11, 1816) ('stylized', 12, 28981) ### +### ('bag', 13, 91) ('cars', 14, 24) ('consisting', 15, 25295) ('are', 16, 8264) ### +### ('relating', 17, 24022) ('simon', 18, 54) ('inclusion', 19, 137) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('recall', 0, 0) ('ford', 108, 1) ('mercury', 4933, 2) ('vehicle', 4, 3) ('vehicles', 3, 4) ### +### ('ˈ', 153, 5) ('recalled', 6, 6) ('lincoln', 193, 7) ('recalls', 5265, 8) ('##ο', 91, 9) ### +### ('made', 5306, 10) ('−', 37, 11) ('passenger', 166, 12) ('crashing', 24, 13) ('stumbled', 127, 14) ### +### ('wingspan', 146, 15) ('cyrillic', 237, 16) ('hesitated', 56, 17) ('number', 621, 18) ### +### ('hating', 97, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('recall', 0, 0) ('vehicles', 3, 4) ('air', 2, 35) ('##bag', 1, 135) ('vehicle', 4, 3) ### +### ('recalled', 6, 6) ('included', 5, 59) ('cars', 14, 24) ('bag', 13, 91) ('include', 10, 182) ### +### ('simon', 18, 54) ('crashing', 24, 13) ('includes', 8, 406) ('unwilling', 26, 21) ('car', 28, 22) ### +### ('−', 37, 11) ('ford', 108, 1) ('crashed', 25, 40) ('inclusion', 19, 137) ('hugh', 31, 56) ### +############################################################################################################ +[2023-10-08 01:12:10,743][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:12:10,743][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:12:11,166][root][INFO] - Epoch: 17: Step: 1301/1557, loss[v]=0.042234, lr=0.000002, acc@1[1]=246.0/256=0.9609375, acc@1[2]=255.0/256=0.99609375 +[2023-10-08 01:13:28,657][root][INFO] - Train batch 1400 +[2023-10-08 01:13:28,659][root][INFO] - Avg. loss per last 100 batches: 0.054233 +[2023-10-08 01:13:29,379][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29107.5/29522=98.60% | mean: 0.01 | max: 5.45 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.46 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] healthy bakeware [SEP] ### +### [P_TEXT]: [CLS] healthy bakeware that can withstand high temperatures or temperature changes ### +### without breaking or exploding. xtrema healthy bakeware can go from oven to counter top to freezer ### +### without shattering, as some glassware is prone to do. there is no need to worry about high ### +### temperatures. our cookware can withstand 2700a°f. [SEP] ### +### ======================================= h_v_q | Gates: 27967 ======================================= ### +### ('##ware', 0, 0) ('healthy', 1, 2) ('##ke', 2, 39) ('ba', 3, 44) ('health', 4, 2847) ### +### ('familiarity', 5, 27339) ('stylized', 6, 28716) ('.', 7, 12962) ('software', 8, 469) ### +### ('normal', 9, 66) ('consisting', 10, 24489) ('relating', 11, 24909) ('plural', 12, 17003) ### +### ('sustainable', 13, 410) ('simon', 14, 57) ('healthcare', 15, 3389) ('−', 16, 23) ### +### ('mathematics', 17, 21994) ('crashed', 18, 28) ('future', 19, 6696) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##ware', 0, 0) ('withstand', 8080, 1) ('healthy', 1, 2) ('ˈ', 57, 3) ('temperature', 2181, 4) ### +### ('hating', 21, 5) ('glass', 1212, 6) ('temperatures', 14009, 7) ('crashing', 24, 8) ### +### ('explode', 4712, 9) ('exploding', 17935, 10) ('freeze', 9060, 11) ('##ο', 53, 12) ### +### ('exploded', 1393, 13) ('prone', 10408, 14) ('worried', 130, 15) ('hesitated', 49, 16) ### +### ('unwilling', 26, 17) ('cook', 62, 18) ('##₂', 25, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##ware', 0, 0) ('healthy', 1, 2) ('##ke', 2, 39) ('ba', 3, 44) ('normal', 9, 66) ('−', 16, 23) ### +### ('hating', 21, 5) ('crashing', 24, 8) ('simon', 14, 57) ('crashed', 18, 28) ('unwilling', 26, 17) ### +### ('##₂', 25, 19) ('health', 4, 2847) ('ˈ', 57, 3) ('software', 8, 469) ('##ο', 53, 12) ### +### ('##α', 33, 38) ('hesitated', 49, 16) ('##大', 41, 31) ('gideon', 38, 45) ### +############################################################################################################ +[2023-10-08 01:13:29,380][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:13:29,380][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:13:29,784][root][INFO] - Epoch: 17: Step: 1401/1557, loss[v]=0.067778, lr=0.000002, acc@1[1]=246.0/256=0.9609375, acc@1[2]=251.0/256=0.98046875 +[2023-10-08 01:14:46,545][root][INFO] - Train batch 1500 +[2023-10-08 01:14:46,547][root][INFO] - Avg. loss per last 100 batches: 0.058058 +[2023-10-08 01:14:47,275][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29086.0/29522=98.52% | mean: 0.01 | max: 5.38 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.24 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the genre of gothika [SEP] ### +### [P_TEXT]: [CLS] but those are all bothersome details of plausibility and logic, and those are the ### +### last two qualities you should seek in gothika.. this is a psychothriller with the plausibility of a ### +### nightmare - - which is to say, it doesn't make sense, but it keeps your attention. [SEP] ### +### ======================================= h_v_q | Gates: 26798 ======================================= ### +### ('goth', 0, 0) ('##ika', 1, 6) ('genre', 2, 1137) ('encompasses', 3, 93) ('genres', 4, 3852) ### +### ('familiarity', 5, 24433) ('is', 6, 3527) ('.', 7, 13130) ('stylized', 8, 27752) ### +### ('##sam', 9, 27597) ('plural', 10, 12940) ('relating', 11, 25988) ('stands', 12, 7336) ### +### ('consisting', 13, 26063) ('format', 14, 6187) ('refers', 15, 23397) ('##ica', 16, 3576) ### +### ('encyclopedia', 17, 12636) ('album', 18, 18281) ('category', 19, 716) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('goth', 0, 0) ('bother', 1457, 1) ('##ibility', 20810, 2) ('nightmare', 9326, 3) ### +### ('psycho', 3706, 4) ('ˈ', 432, 5) ('##ika', 1, 6) ('seek', 1763, 7) ('logic', 1045, 8) ### +### ('attention', 3150, 9) ('qualities', 5936, 10) ('crashing', 66, 11) ('##ο', 185, 12) ### +### ('hating', 184, 13) ('pl', 9005, 14) ('crashed', 40, 15) ('cyrillic', 498, 16) ### +### ('hesitated', 132, 17) ('unwilling', 102, 18) ('wingspan', 296, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('goth', 0, 0) ('##ika', 1, 6) ('encompasses', 3, 93) ('genre', 2, 1137) ('crashed', 40, 15) ### +### ('−', 34, 25) ('crashing', 66, 11) ('##α', 44, 38) ('##₂', 88, 22) ('unwilling', 102, 18) ### +### ('##ང', 114, 21) ('stumbled', 111, 23) ('angrily', 112, 24) ('hesitated', 132, 17) ### +### ('julian', 39, 84) ('##ο', 185, 12) ('ruined', 98, 35) ('hating', 184, 13) ('gothic', 35, 128) ### +### ('sharply', 188, 20) ### +############################################################################################################ +[2023-10-08 01:14:47,275][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:14:47,275][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:14:47,700][root][INFO] - Epoch: 17: Step: 1501/1557, loss[v]=0.075074, lr=0.000002, acc@1[1]=244.0/256=0.953125, acc@1[2]=253.0/256=0.98828125 +[2023-10-08 01:15:31,543][root][INFO] - rank=1; last iteration 1557 +[2023-10-08 01:15:31,544][root][INFO] - rank=2; last iteration 1557 +[2023-10-08 01:15:31,544][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 01:15:31,544][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 01:15:31,544][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-08 01:15:31,544][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-08 01:15:31,546][root][INFO] - rank=3; last iteration 1557 +[2023-10-08 01:15:31,547][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 01:15:31,547][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-08 01:15:31,547][root][INFO] - rank=0; last iteration 1557 +[2023-10-08 01:15:31,547][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 01:15:31,547][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-08 01:15:31,552][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 01:15:31,552][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 01:15:31,553][root][INFO] - Epoch finished on 1 +[2023-10-08 01:15:31,553][root][INFO] - Epoch finished on 2 +[2023-10-08 01:15:31,555][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 01:15:31,555][root][INFO] - Epoch finished on 0 +[2023-10-08 01:15:31,556][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 01:15:31,556][root][INFO] - Epoch finished on 3 +[2023-10-08 01:16:09,528][root][INFO] - Saved checkpoint at ./vdr_17 +[2023-10-08 01:16:09,529][root][INFO] - Av Loss per epoch=0.057914 +[2023-10-08 01:16:09,529][root][INFO] - epoch total (1) correct predictions=380160 +[2023-10-08 01:16:09,529][root][INFO] - epoch total (2) correct predictions=392155 +[2023-10-08 01:16:09,528][root][INFO] - Saved checkpoint at ./vdr_17 +[2023-10-08 01:16:09,530][root][INFO] - Av Loss per epoch=0.057914 +[2023-10-08 01:16:09,530][root][INFO] - Saved checkpoint at ./vdr_17 +[2023-10-08 01:16:09,530][root][INFO] - epoch total (1) correct predictions=380160 +[2023-10-08 01:16:09,530][root][INFO] - Av Loss per epoch=0.057914 +[2023-10-08 01:16:09,531][root][INFO] - epoch total (2) correct predictions=392155 +[2023-10-08 01:16:09,531][root][INFO] - epoch total (1) correct predictions=380160 +[2023-10-08 01:16:09,531][root][INFO] - epoch total (2) correct predictions=392155 +[2023-10-08 01:16:09,531][root][INFO] - Saved checkpoint at ./vdr_17 +[2023-10-08 01:16:09,532][root][INFO] - Av Loss per epoch=0.057914 +[2023-10-08 01:16:09,532][root][INFO] - epoch total (1) correct predictions=380160 +[2023-10-08 01:16:09,532][root][INFO] - epoch total (2) correct predictions=392155 +[2023-10-08 01:16:09,533][root][INFO] - ***** Epoch 18 ***** +[2023-10-08 01:16:09,535][root][INFO] - ***** Epoch 18 ***** +[2023-10-08 01:16:09,535][root][INFO] - ***** Epoch 18 ***** +[2023-10-08 01:16:09,536][root][INFO] - ***** Epoch 18 ***** +[2023-10-08 01:16:09,540][root][INFO] - rank=3; Iteration start +[2023-10-08 01:16:09,540][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 01:16:09,540][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 01:16:09,541][root][INFO] - rank=0; Iteration start +[2023-10-08 01:16:09,541][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 01:16:09,541][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 01:16:09,542][root][INFO] - rank=1; Iteration start +[2023-10-08 01:16:09,542][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 01:16:09,542][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-08 01:16:09,542][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 01:16:09,543][root][INFO] - rank=2; Iteration start +[2023-10-08 01:16:09,543][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 01:16:09,543][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 01:16:09,543][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-08 01:16:09,544][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-08 01:16:09,545][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-08 01:16:10,482][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29038.5/29522=98.36% | mean: 0.01 | max: 5.61 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.27 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] cost of spray foam insulation [SEP] ### +### [P_TEXT]: [CLS] 1 closed - cell spray foam costs $. 75 to $ 1 per board foot. expressed in terms of ### +### square feet, spray foam installation cost is $ 1. 50 to $ 3. 00 per square foot ( for an average - ### +### sized home between 2, 000 to 2, 500 square feet, thatas a total cost of $ 3, 000 to $ 7, 500. [SEP] ### +### ======================================= h_v_q | Gates: 25948 ======================================= ### +### ('foam', 0, 0) ('$', 1, 7) ('spray', 2, 2) ('##£', 3, 25213) ('insulation', 4, 223) ('cost', 5, 1) ### +### ('.', 6, 11235) ('familiarity', 7, 25922) ('stylized', 8, 28970) ('430', 9, 29016) ### +### ('plural', 10, 20129) ('costs', 11, 3) ('relating', 12, 28170) ('simon', 13, 47) ('=', 14, 18294) ### +### ('consisting', 15, 24638) ('price', 16, 6) ('currency', 17, 940) ('unwilling', 18, 14) ### +### ('money', 19, 480) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('foam', 0, 0) ('cost', 5, 1) ('spray', 2, 2) ('costs', 11, 3) ('ˈ', 78, 4) ('board', 1108, 5) ### +### ('price', 16, 6) ('$', 1, 7) ('hesitated', 45, 8) ('prices', 39, 9) ('installation', 5118, 10) ### +### ('hating', 52, 11) ('##ο', 85, 12) ('cell', 10160, 13) ('unwilling', 18, 14) ('wingspan', 1145, 15) ### +### ('crashing', 28, 16) ('feet', 5324, 17) ('sharply', 106, 18) ('##ང', 140, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('foam', 0, 0) ('spray', 2, 2) ('$', 1, 7) ('cost', 5, 1) ('insulation', 4, 223) ('costs', 11, 3) ### +### ('price', 16, 6) ('unwilling', 18, 14) ('simon', 13, 47) ('prices', 39, 9) ('hesitated', 45, 8) ### +### ('crashing', 28, 16) ('ˈ', 78, 4) ('hating', 52, 11) ('stark', 31, 29) ('##α', 46, 23) ### +### ('##ο', 85, 12) ('−', 41, 31) ('screenwriter', 27, 51) ('angrily', 64, 24) ### +############################################################################################################ +[2023-10-08 01:16:10,482][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:16:10,482][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:16:10,880][root][INFO] - Epoch: 18: Step: 1/1557, loss[v]=0.055187, lr=0.000002, acc@1[1]=244.5/256=0.955078125, acc@1[2]=254.0/256=0.9921875 +[2023-10-08 01:17:27,973][root][INFO] - Train batch 100 +[2023-10-08 01:17:27,974][root][INFO] - Avg. loss per last 100 batches: 0.058261 +[2023-10-08 01:17:38,327][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29004.7/29522=98.25% | mean: 0.01 | max: 5.77 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.4/29522=100.00% | mean: 0.14 | max: 6.47 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is wall sit workout? [SEP] ### +### [P_TEXT]: [CLS] wall sit exercise, side view. a wall sit is an exercise done to strengthen the ### +### quadriceps muscles. it is characterized by the two right angles formed by the body, one at the hips ### +### and one at the knees. the person wall sitting places their back against a wall with their feet ### +### shoulder width apart and a little ways out from the wall. [SEP] ### +### ======================================= h_v_q | Gates: 26196 ======================================= ### +### ('sit', 0, 1) ('wall', 1, 0) ('workout', 2, 147) ('sat', 3, 43) ('encompasses', 4, 15) ### +### ('is', 5, 892) ('familiarity', 6, 25351) ('.', 7, 12966) ('relating', 8, 25774) ('training', 9, 65) ### +### ('refers', 10, 13562) ('stylized', 11, 27579) ('sits', 12, 45) ('encyclopedia', 13, 14535) ### +### ('sitting', 14, 2) ('##sam', 15, 28063) ('stands', 16, 8038) ('consisting', 17, 24201) ### +### ('abbreviated', 18, 13385) ('designed', 19, 5238) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('wall', 1, 0) ('sit', 0, 1) ('sitting', 14, 2) ('exercise', 68, 3) ('quad', 15774, 4) ### +### ('hips', 10509, 5) ('strengthen', 2892, 6) ('ˈ', 253, 7) ('muscles', 138, 8) ('knees', 11204, 9) ### +### ('view', 3857, 10) ('##rice', 20305, 11) ('crashing', 91, 12) ('hating', 47, 13) ('##ο', 183, 14) ### +### ('encompasses', 4, 15) ('−', 69, 16) ('##₂', 82, 17) ('walls', 103, 18) ('glimpse', 2265, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('sit', 0, 1) ('wall', 1, 0) ('workout', 2, 147) ('encompasses', 4, 15) ('sat', 3, 43) ### +### ('sitting', 14, 2) ('training', 9, 65) ('sits', 12, 45) ('exercise', 68, 3) ('definition', 22, 50) ### +### ('hating', 47, 13) ('unwilling', 49, 22) ('crashing', 91, 12) ('−', 69, 16) ('stand', 20, 140) ### +### ('simon', 29, 64) ('##α', 59, 30) ('is', 5, 892) ('muscles', 138, 8) ('##₂', 82, 17) ### +############################################################################################################ +[2023-10-08 01:17:38,327][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:17:38,327][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:17:38,725][root][INFO] - Epoch: 18: Step: 101/1557, loss[v]=0.063125, lr=0.000002, acc@1[1]=242.5/256=0.947265625, acc@1[2]=251.0/256=0.98046875 +[2023-10-08 01:18:55,735][root][INFO] - Train batch 200 +[2023-10-08 01:18:55,736][root][INFO] - Avg. loss per last 100 batches: 0.058081 +[2023-10-08 01:18:56,444][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29032.7/29522=98.34% | mean: 0.01 | max: 5.56 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.14 | max: 6.42 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] when do children get shots for epsdt [SEP] ### +### [P_TEXT]: [CLS] a¢ children 18 years old and younger can get epsdt with no co - pay for any covered ### +### service. a¢ adults 19 and 20 years old can get epsdt, but may have a small copay for some services. ### +### -. a¢ children in department of social and human services custody can get espdt with no co - pay if ### +### they. are 18 or younger. they may have some co - pays if they are 19 or 20. [SEP] ### +### ======================================= h_v_q | Gates: 27648 ======================================= ### +### ('eps', 0, 6) ('##dt', 1, 0) ('children', 2, 4) ('shots', 3, 15309) ('child', 4, 51) ### +### ('.', 5, 13503) ('2017', 6, 20185) ('shot', 7, 7859) ('april', 8, 7523) ('spring', 9, 11422) ### +### ('when', 10, 751) ('september', 11, 2652) ('for', 12, 5324) ('weeks', 13, 120) ('2016', 14, 21407) ### +### ('kids', 15, 39) ('get', 16, 114) ('familiarity', 17, 20914) ('age', 18, 22) ('months', 19, 3228) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('##dt', 1, 0) ('ˈ', 334, 1) ('custody', 16088, 2) ('department', 3697, 3) ('children', 2, 4) ### +### ('##ο', 282, 5) ('eps', 0, 6) ('hesitated', 148, 7) ('copa', 20920, 8) ('hating', 179, 9) ### +### ('##pd', 26239, 10) ('old', 515, 11) ('crashing', 85, 12) ('unwilling', 127, 13) ### +### ('services', 2439, 14) ('−', 68, 15) ('service', 2031, 16) ('co', 8328, 17) ('cyrillic', 194, 18) ### +### ('pay', 7577, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##dt', 1, 0) ('eps', 0, 6) ('children', 2, 4) ('child', 4, 51) ('age', 18, 22) ('kids', 15, 39) ### +### ('weeks', 13, 120) ('get', 16, 114) ('adults', 47, 25) ('−', 68, 15) ('crashing', 85, 12) ### +### ('ep', 23, 254) ('unwilling', 127, 13) ('##α', 91, 29) ('hesitated', 148, 7) ('shoved', 96, 28) ### +### ('when', 10, 751) ('##₂', 107, 30) ('hating', 179, 9) ('simon', 70, 75) ### +############################################################################################################ +[2023-10-08 01:18:56,444][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:18:56,444][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:18:56,869][root][INFO] - Epoch: 18: Step: 201/1557, loss[v]=0.042113, lr=0.000002, acc@1[1]=244.5/256=0.955078125, acc@1[2]=253.0/256=0.98828125 +[2023-10-08 01:20:13,392][root][INFO] - Train batch 300 +[2023-10-08 01:20:13,393][root][INFO] - Avg. loss per last 100 batches: 0.055950 +[2023-10-08 01:20:14,075][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29052.6/29522=98.41% | mean: 0.01 | max: 5.64 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.14 | max: 6.30 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what side to wear name tag [SEP] ### +### [P_TEXT]: [CLS] where to place a nametag or b adge : networking a whether at professional functions ### +### or at social events, always wear it on your upper right shoulder. hereas why : 1 place the tag or ### +### badge as high up on your right shoulder as possible to give other people the best and easiest view ### +### of both the tag and your face. [SEP] ### +### ======================================= h_v_q | Gates: 27280 ======================================= ### +### ('tag', 0, 0) ('name', 1, 21) ('side', 2, 6744) ('wear', 3, 44) ('sides', 4, 6780) ### +### ('front', 5, 7922) ('wearing', 6, 295) ('.', 7, 4865) ('familiarity', 8, 25066) ('names', 9, 13) ### +### ('stylized', 10, 28868) ('wore', 11, 3652) ('cheek', 12, 546) ('consisting', 13, 26423) ### +### ('onto', 14, 27) ('relating', 15, 26839) ('term', 16, 4118) ('title', 17, 1651) ### +### ('towards', 18, 1549) ('wears', 19, 1800) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('tag', 0, 0) ('networking', 7936, 1) ('shoulder', 170, 2) ('##ge', 12962, 3) ('badge', 2553, 4) ### +### ('ˈ', 33, 5) ('badges', 8498, 6) ('glimpse', 2912, 7) ('##ο', 98, 8) ('crashing', 72, 9) ### +### ('ad', 736, 10) ('professional', 722, 11) ('shoulders', 10851, 12) ('names', 9, 13) ### +### ('hating', 134, 14) ('face', 319, 15) ('hesitated', 194, 16) ('unwilling', 75, 17) ### +### ('cyrillic', 127, 18) ('view', 723, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('tag', 0, 0) ('name', 1, 21) ('wear', 3, 44) ('names', 9, 13) ('onto', 14, 27) ('wearing', 6, 295) ### +### ('side', 2, 6744) ('ˈ', 33, 5) ('crashing', 72, 9) ('tags', 21, 71) ('−', 63, 20) ('##α', 35, 45) ### +### ('unwilling', 75, 17) ('##ο', 98, 8) ('shoulder', 170, 2) ('surname', 25, 92) ('named', 31, 80) ### +### ('angrily', 66, 42) ('hating', 134, 14) ('cyrillic', 127, 18) ### +############################################################################################################ +[2023-10-08 01:20:14,076][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:20:14,076][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:20:14,498][root][INFO] - Epoch: 18: Step: 301/1557, loss[v]=0.048793, lr=0.000002, acc@1[1]=245.5/256=0.958984375, acc@1[2]=253.0/256=0.98828125 +[2023-10-08 01:21:31,333][root][INFO] - Train batch 400 +[2023-10-08 01:21:31,333][root][INFO] - Avg. loss per last 100 batches: 0.057224 +[2023-10-08 01:21:32,035][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29077.2/29522=98.49% | mean: 0.01 | max: 5.29 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.4/29522=100.00% | mean: 0.15 | max: 6.15 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] calories in coffee cake [SEP] ### +### [P_TEXT]: [CLS] this nutritional information applies to a square coffee cake measuring 8 in. on a ### +### side. the serving size is 1 / 12 of the entire coffee cake. a serving of coffee cake contains 137 ### +### calories, which is about 7 percent of the daily value for calories. [SEP] ### +### ======================================= h_v_q | Gates: 26655 ======================================= ### +### ('cake', 0, 0) ('coffee', 1, 1) ('cal', 2, 14) ('##ories', 3, 15) ('.', 4, 15662) ### +### ('familiarity', 5, 26775) ('stylized', 6, 28699) ('cakes', 7, 8) ('california', 8, 8026) ### +### ('simon', 9, 51) ('relating', 10, 27926) ('consisting', 11, 19441) ('mathematics', 12, 21994) ### +### ('−', 13, 18) ('in', 14, 7243) ('##α', 15, 24) ('headquartered', 16, 6284) ('crashing', 17, 10) ### +### ('julian', 18, 69) ('##₂', 19, 23) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('cake', 0, 0) ('coffee', 1, 1) ('nutritional', 14359, 2) ('ˈ', 27, 3) ('size', 5533, 4) ### +### ('hesitated', 28, 5) ('hating', 43, 6) ('##ο', 48, 7) ('cakes', 7, 8) ('unwilling', 32, 9) ### +### ('crashing', 17, 10) ('wingspan', 81, 11) ('##大', 30, 12) ('cyrillic', 140, 13) ('cal', 2, 14) ### +### ('##ories', 3, 15) ('##ང', 59, 16) ('sharply', 35, 17) ('−', 13, 18) ('afraid', 100, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('cake', 0, 0) ('coffee', 1, 1) ('cal', 2, 14) ('##ories', 3, 15) ('cakes', 7, 8) ### +### ('crashing', 17, 10) ('simon', 9, 51) ('ˈ', 27, 3) ('−', 13, 18) ('hesitated', 28, 5) ### +### ('##₂', 19, 23) ('##α', 15, 24) ('angrily', 21, 20) ('unwilling', 32, 9) ('##大', 30, 12) ### +### ('hating', 43, 6) ('##ο', 48, 7) ('crashed', 24, 29) ('sharply', 35, 17) ('ruined', 23, 41) ### +############################################################################################################ +[2023-10-08 01:21:32,035][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:21:32,035][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:21:32,459][root][INFO] - Epoch: 18: Step: 401/1557, loss[v]=0.051725, lr=0.000002, acc@1[1]=244.0/256=0.953125, acc@1[2]=254.0/256=0.9921875 +[2023-10-08 01:22:49,050][root][INFO] - Train batch 500 +[2023-10-08 01:22:49,051][root][INFO] - Avg. loss per last 100 batches: 0.054500 +[2023-10-08 01:22:49,746][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29092.1/29522=98.54% | mean: 0.01 | max: 5.30 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.6/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.4/29522=100.00% | mean: 0.15 | max: 6.07 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] does preserving sugar contain pectin [SEP] ### +### [P_TEXT]: [CLS] gelling sugar or ( british ) jam sugar or ( us ) jelly sugar or sugar with pectin ### +### is a kind of sugar, which is used to produce preserves and which contains pectin as a gelling ### +### agent. it also usually contains citric acid as a preservative, sometimes along with other ### +### substances, such as sorbic acid or sodium benzoate. 1 : 1 a use for jellies and jams with equal ### +### weights of fruit and gelling sugar. 2 2 : 1 a use for preserves to produce less sweetness. 3 use ### +### twice as much fruit in weight as you do gelling sugar. 4 3 : 1 a use for preserves to produce ### +### maximum fruit taste. 5 use three times as much fruit in weight as you do gelling sugar. [SEP] ### +### ======================================= h_v_q | Gates: 27580 ======================================= ### +### ('preserving', 0, 34) ('pe', 1, 6) ('sugar', 2, 2) ('##in', 3, 24) ('##ct', 4, 23) ('.', 5, 15792) ### +### ('preservation', 6, 1823) ('does', 7, 17743) ('preserve', 8, 31) ('doesn', 9, 1105) ### +### ('preserved', 10, 102) ('contain', 11, 288) ('familiarity', 12, 27193) ('stylized', 13, 28405) ### +### ('contains', 14, 81) ('answer', 15, 21081) ('consisting', 16, 18299) ('simon', 17, 59) ### +### ('relating', 18, 24256) ('ruined', 19, 26) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('jam', 17137, 0) ('jelly', 3719, 1) ('sugar', 2, 2) ('gel', 8733, 3) ('##ling', 1498, 4) ### +### ('ˈ', 108, 5) ('pe', 1, 6) ('crashing', 32, 7) ('hating', 37, 8) ('preserves', 34, 9) ### +### ('##ο', 39, 10) ('encompasses', 732, 11) ('−', 33, 12) ('fruit', 9409, 13) ('unwilling', 23, 14) ### +### ('stumbled', 185, 15) ('hesitated', 60, 16) ('##₂', 41, 17) ('sharply', 70, 18) ### +### ('cyrillic', 121, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('sugar', 2, 2) ('pe', 1, 6) ('preserving', 0, 34) ('##in', 3, 24) ('##ct', 4, 23) ### +### ('preserve', 8, 31) ('preserved', 10, 102) ('contains', 14, 81) ('simon', 17, 59) ### +### ('ruined', 19, 26) ('unwilling', 23, 14) ('crashing', 32, 7) ('contain', 11, 288) ### +### ('preserves', 34, 9) ('hating', 37, 8) ('−', 33, 12) ('angrily', 22, 32) ('##ο', 39, 10) ### +### ('##α', 26, 36) ('crashed', 30, 28) ### +############################################################################################################ +[2023-10-08 01:22:49,747][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:22:49,747][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:22:50,171][root][INFO] - Epoch: 18: Step: 501/1557, loss[v]=0.051070, lr=0.000002, acc@1[1]=246.0/256=0.9609375, acc@1[2]=253.5/256=0.990234375 +[2023-10-08 01:24:06,462][root][INFO] - Train batch 600 +[2023-10-08 01:24:06,474][root][INFO] - Avg. loss per last 100 batches: 0.059495 +[2023-10-08 01:24:07,171][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29018.9/29522=98.30% | mean: 0.01 | max: 5.36 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.14 | max: 6.27 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] foods highest in polyphenols and fiber [SEP] ### +### [P_TEXT]: [CLS] in fact, most of the foods that are highest in polyphenols are things that most ### +### people wouldnat expect. dark chocolate, coffee, and herbs and spices are a good example of this. ### +### eating some dark chocolate, a few cups of coffee or tea and cooking with spices are all ### +### straightforward and enjoyable things to do. [SEP] ### +### ======================================= h_v_q | Gates: 27615 ======================================= ### +### ('##eno', 0, 34) ('fiber', 1, 15616) ('poly', 2, 4) ('highest', 3, 8) ('foods', 4, 1) ### +### ('##ph', 5, 74) ('##ls', 6, 45) ('.', 7, 12547) ('fish', 8, 11515) ('familiarity', 9, 26874) ### +### ('stylized', 10, 28551) ('high', 11, 148) ('vegetables', 12, 302) ('grapes', 13, 4887) ### +### ('meat', 14, 180) ('simon', 15, 49) ('rice', 16, 5597) ('consisting', 17, 24779) ('##₂', 18, 23) ### +### ('relating', 19, 20044) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('chocolate', 455, 0) ('foods', 4, 1) ('spices', 2295, 2) ('herbs', 6399, 3) ('poly', 2, 4) ### +### ('ˈ', 46, 5) ('coffee', 23, 6) ('hating', 57, 7) ('highest', 3, 8) ('hesitated', 44, 9) ### +### ('examples', 5189, 10) ('crashing', 47, 11) ('enjoyable', 23329, 12) ('##ο', 52, 13) ### +### ('things', 1192, 14) ('wingspan', 118, 15) ('cyrillic', 98, 16) ('−', 27, 17) ('cooking', 869, 18) ### +### ('unwilling', 32, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('foods', 4, 1) ('poly', 2, 4) ('##eno', 0, 34) ('highest', 3, 8) ('##ph', 5, 74) ('##ls', 6, 45) ### +### ('coffee', 23, 6) ('##₂', 18, 23) ('simon', 15, 49) ('−', 27, 17) ('unwilling', 32, 19) ### +### ('ˈ', 46, 5) ('hesitated', 44, 9) ('high', 11, 148) ('∈', 35, 21) ('ruined', 25, 33) ### +### ('crashing', 47, 11) ('food', 39, 22) ('hating', 57, 7) ('##α', 31, 41) ### +############################################################################################################ +[2023-10-08 01:24:07,171][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:24:07,172][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:24:07,580][root][INFO] - Epoch: 18: Step: 601/1557, loss[v]=0.035653, lr=0.000002, acc@1[1]=245.5/256=0.958984375, acc@1[2]=254.0/256=0.9921875 +[2023-10-08 01:25:24,093][root][INFO] - Train batch 700 +[2023-10-08 01:25:24,098][root][INFO] - Avg. loss per last 100 batches: 0.058860 +[2023-10-08 01:25:24,800][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29069.0/29522=98.47% | mean: 0.01 | max: 5.38 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.4/29522=100.00% | mean: 0.15 | max: 6.27 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] where is serial number on tracfone [SEP] ### +### [P_TEXT]: [CLS] enter airtime pin. the airtime pin is a group of numbers found on the back of your ### +### tracfone prepaid wireless airtime card or the airtime pin code ( s ) found on your retail cash ### +### register receipt. to find your phone number. your phone number should be displayed on your screen. ### +### nter airtime pin. the airtime pin is a group of numbers found on the back of your tracfone prepaid ### +### wireless airtime card or the airtime pin code ( s ) found on your retail cash register receipt. to ### +### find your phone number. your phone number should be displayed on your screen. [SEP] ### +### ======================================= h_v_q | Gates: 28014 ======================================= ### +### ('tr', 0, 8) ('serial', 1, 3099) ('##fo', 2, 61) ('number', 3, 28) ('##ac', 4, 64) ### +### ('downtown', 5, 2357) ('stylized', 6, 28980) ('located', 7, 881) ('somewhere', 8, 141) ### +### ('##ne', 9, 98) ('familiarity', 10, 23082) ('800', 11, 15435) ('situated', 12, 7343) ### +### ('=', 13, 16255) ('.', 14, 19322) ('relating', 15, 26067) ('africa', 16, 13277) ('430', 17, 29124) ### +### ('##º', 18, 28790) ('hampshire', 19, 19265) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('pin', 2931, 0) ('##time', 12637, 1) ('nt', 13107, 2) ('ˈ', 125, 3) ('##ο', 56, 4) ### +### ('unwilling', 39, 5) ('hating', 145, 6) ('pins', 14436, 7) ('tr', 0, 8) ('retail', 2451, 9) ### +### ('register', 1664, 10) ('crashing', 95, 11) ('hesitated', 71, 12) ('find', 5451, 13) ### +### ('air', 2338, 14) ('cyrillic', 205, 15) ('##ང', 66, 16) ('−', 48, 17) ('stumbled', 286, 18) ### +### ('dodgers', 214, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('tr', 0, 8) ('number', 3, 28) ('##fo', 2, 61) ('##ac', 4, 64) ('##ne', 9, 98) ('serial', 1, 3099) ### +### ('somewhere', 8, 141) ('unwilling', 39, 5) ('##ο', 56, 4) ('−', 48, 17) ('##α', 27, 43) ### +### ('hesitated', 71, 12) ('##ང', 66, 16) ('crashed', 47, 34) ('found', 29, 69) ('crashing', 95, 11) ### +### ('ˈ', 125, 3) ('angrily', 53, 35) ('numbers', 80, 22) ('located', 7, 881) ### +############################################################################################################ +[2023-10-08 01:25:24,801][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:25:24,801][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:25:25,222][root][INFO] - Epoch: 18: Step: 701/1557, loss[v]=0.087334, lr=0.000002, acc@1[1]=242.0/256=0.9453125, acc@1[2]=251.0/256=0.98046875 +[2023-10-08 01:26:41,985][root][INFO] - Train batch 800 +[2023-10-08 01:26:41,986][root][INFO] - Avg. loss per last 100 batches: 0.060729 +[2023-10-08 01:26:42,670][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29076.2/29522=98.49% | mean: 0.01 | max: 5.43 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.4/29522=100.00% | mean: 0.15 | max: 6.04 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] who is the founder of cartoon network [SEP] ### +### [P_TEXT]: [CLS] for cartoon network channels in other countries, see list of international cartoon ### +### network channels. for other uses, see cartoon network ( disambiguation ). cartoon network is an ### +### american basic cable and satellite television channel that is owned by time warner through the ### +### turner broadcasting system subsidiary. it was founded by betty cohen and launched on october 1, ### +### 1992. [SEP] ### +### ======================================= h_v_q | Gates: 25954 ======================================= ### +### ('cartoon', 0, 0) ('founder', 1, 123) ('network', 2, 5) ('founded', 3, 49) ('founders', 4, 451) ### +### ('cartoons', 5, 9) ('founding', 6, 653) ('.', 7, 8812) ('who', 8, 58) ('whose', 9, 125) ### +### ('networks', 10, 26) ('familiarity', 11, 25841) ('##sam', 12, 118) ('stylized', 13, 27269) ### +### ('american', 14, 314) ('president', 15, 6203) ('established', 16, 543) ('organization', 17, 2037) ### +### ('animation', 18, 47) ('spirit', 19, 1807) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('cartoon', 0, 0) ('channels', 1095, 1) ('cohen', 7906, 2) ('channel', 20, 3) ('turner', 3922, 4) ### +### ('network', 2, 5) ('encompasses', 182, 6) ('ˈ', 959, 7) ('warner', 682, 8) ('cartoons', 5, 9) ### +### ('betty', 837, 10) ('##ο', 239, 11) ('crashing', 131, 12) ('stumbled', 594, 13) ('cable', 516, 14) ### +### ('time', 3250, 15) ('satellite', 90, 16) ('unwilling', 297, 17) ('wingspan', 800, 18) ### +### ('hesitated', 432, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('cartoon', 0, 0) ('network', 2, 5) ('founder', 1, 123) ('cartoons', 5, 9) ('founded', 3, 49) ### +### ('channel', 20, 3) ('networks', 10, 26) ('who', 8, 58) ('founders', 4, 451) ('animation', 18, 47) ### +### ('whose', 9, 125) ('##sam', 12, 118) ('founding', 6, 653) ('animated', 25, 59) ### +### ('international', 68, 29) ('satellite', 90, 16) ('encompasses', 182, 6) ('crashing', 131, 12) ### +### ('angrily', 72, 43) ('knew', 41, 90) ### +############################################################################################################ +[2023-10-08 01:26:42,671][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:26:42,671][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:26:43,075][root][INFO] - Epoch: 18: Step: 801/1557, loss[v]=0.052183, lr=0.000002, acc@1[1]=247.5/256=0.966796875, acc@1[2]=253.0/256=0.98828125 +[2023-10-08 01:27:59,749][root][INFO] - Train batch 900 +[2023-10-08 01:27:59,750][root][INFO] - Avg. loss per last 100 batches: 0.057776 +[2023-10-08 01:28:00,469][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29061.3/29522=98.44% | mean: 0.01 | max: 5.42 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.4/29522=100.00% | mean: 0.15 | max: 6.20 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] toshiba tv price [SEP] ### +### [P_TEXT]: [CLS] price range of toshiba televisions the price of toshiba televisions vary when we ### +### talk about all the products being offered in the market. the most expensive television is toshiba ### +### l9300 213. 36 cm ( 84 ) 4k ( ultra hd ) led television priced at rs. 9, 99, 351. [SEP] ### +### ======================================= h_v_q | Gates: 26771 ======================================= ### +### ('##shi', 0, 6) ('##ba', 1, 19) ('$', 2, 42) ('##£', 3, 17226) ('tv', 4, 8) ('price', 5, 1) ### +### ('to', 6, 240) ('television', 7, 2) ('familiarity', 8, 25210) ('stylized', 9, 29088) ### +### ('towards', 10, 1665) ('prices', 11, 5) ('radio', 12, 68) ('plural', 13, 19912) ('toward', 14, 674) ### +### ('##と', 15, 50) ('network', 16, 1247) ('430', 17, 29132) ('.', 18, 8844) ('consisting', 19, 21696) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('expensive', 323, 0) ('price', 5, 1) ('television', 7, 2) ('priced', 670, 3) ('ˈ', 50, 4) ### +### ('prices', 11, 5) ('##shi', 0, 6) ('##ο', 127, 7) ('tv', 4, 8) ('cost', 26, 9) ('crashing', 49, 10) ### +### ('wingspan', 221, 11) ('cm', 10856, 12) ('hesitated', 186, 13) ('##k', 17012, 14) ### +### ('hating', 73, 15) ('unwilling', 30, 16) ('hd', 12686, 17) ('##ང', 222, 18) ('##ba', 1, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##shi', 0, 6) ('##ba', 1, 19) ('tv', 4, 8) ('$', 2, 42) ('price', 5, 1) ('television', 7, 2) ### +### ('prices', 11, 5) ('to', 6, 240) ('radio', 12, 68) ('cost', 26, 9) ('##と', 15, 50) ### +### ('unwilling', 30, 16) ('ˈ', 50, 4) ('crashing', 49, 10) ('##α', 39, 26) ('hating', 73, 15) ### +### ('channel', 21, 134) ('##ο', 127, 7) ('−', 72, 22) ('crashed', 55, 33) ### +############################################################################################################ +[2023-10-08 01:28:00,469][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:28:00,469][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:28:00,896][root][INFO] - Epoch: 18: Step: 901/1557, loss[v]=0.038310, lr=0.000001, acc@1[1]=245.0/256=0.95703125, acc@1[2]=254.0/256=0.9921875 +[2023-10-08 01:29:17,644][root][INFO] - Train batch 1000 +[2023-10-08 01:29:17,644][root][INFO] - Avg. loss per last 100 batches: 0.054842 +[2023-10-08 01:29:18,370][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29088.3/29522=98.53% | mean: 0.01 | max: 5.61 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.30 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the longest road in the united states [SEP] ### +### [P_TEXT]: [CLS] created by roadtrippers - june 15th 2016. route 66 might be known as america's ### +### mother road, but u. s. route 20 is, without question, the big daddy. it's america's longest road, ### +### stretching from boston, massachusetts to newport, oregon. that's 12 states and 3, 365 miles of ### +### transcontinental american awesomeness to explore. go big and then go home with a mega road trip ### +### along this epic route! [SEP] ### +### ======================================= h_v_q | Gates: 26505 ======================================= ### +### ('longest', 0, 0) ('road', 1, 6) ('states', 2, 42) ('united', 3, 5705) ('roads', 4, 12) ### +### ('america', 5, 65) ('familiarity', 6, 24158) ('americans', 7, 109) ('american', 8, 75) ### +### ('street', 9, 136) ('stylized', 10, 26096) ('alaska', 11, 4295) ('highway', 12, 37) ### +### ('federal', 13, 3882) ('path', 14, 73) ('washington', 15, 1586) ('largest', 16, 8) ### +### ('california', 17, 4199) ('usa', 18, 138) ('.', 19, 9843) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('longest', 0, 0) ('route', 20, 1) ('routes', 572, 2) ('daddy', 10040, 3) ('newport', 4441, 4) ### +### ('20', 2955, 5) ('road', 1, 6) ('ˈ', 149, 7) ('largest', 16, 8) ('big', 2334, 9) ### +### ('boston', 2240, 10) ('66', 14416, 11) ('roads', 4, 12) ('##ο', 142, 13) ('##ntal', 26859, 14) ### +### ('massachusetts', 23, 15) ('mother', 3085, 16) ('##trip', 27934, 17) ('oregon', 113, 18) ### +### ('biggest', 396, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('longest', 0, 0) ('road', 1, 6) ('roads', 4, 12) ('states', 2, 42) ('america', 5, 65) ### +### ('route', 20, 1) ('largest', 16, 8) ('americans', 7, 109) ('highway', 12, 37) ('american', 8, 75) ### +### ('massachusetts', 23, 15) ('path', 14, 73) ('street', 9, 136) ('usa', 18, 138) ('us', 43, 57) ### +### ('−', 73, 23) ('unwilling', 74, 28) ('##α', 54, 54) ('angrily', 51, 63) ('ˈ', 149, 7) ### +############################################################################################################ +[2023-10-08 01:29:18,371][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:29:18,371][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:29:18,793][root][INFO] - Epoch: 18: Step: 1001/1557, loss[v]=0.071040, lr=0.000001, acc@1[1]=244.0/256=0.953125, acc@1[2]=251.0/256=0.98046875 +[2023-10-08 01:30:35,619][root][INFO] - Train batch 1100 +[2023-10-08 01:30:35,620][root][INFO] - Avg. loss per last 100 batches: 0.058136 +[2023-10-08 01:30:36,324][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29098.1/29522=98.56% | mean: 0.01 | max: 5.47 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.08 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what airlines fly into savannah georgia [SEP] ### +### [P_TEXT]: [CLS] cheap flights to savannah recently found by travelers 1 *. $ 78 : baltimore to ### +### savannah on allegiant air - nonstop ( found 08 / 15 / 2016 ) 2 $ 127 : new york city to savannah on ### +### jetblue airways - nonstop ( found 08 / 15 / 2016 ) $ 365 : chicago to savannah on united airlines - ### +### nonstop ( found 08 / 15 / 2016 ) [SEP] ### +### ======================================= h_v_q | Gates: 26065 ======================================= ### +### ('savannah', 0, 0) ('airlines', 1, 5) ('georgia', 2, 189) ('into', 3, 790) ('.', 4, 15407) ### +### ('flight', 5, 11) ('fly', 6, 34) ('flying', 7, 29) ('familiarity', 8, 24232) ('flights', 9, 4) ### +### ('aircraft', 10, 211) ('inside', 11, 1361) ('airline', 12, 16) ('onto', 13, 80) ('airways', 14, 7) ### +### ('through', 15, 2056) ('stylized', 16, 28787) ('wings', 17, 1144) ('relating', 18, 24182) ### +### ('boeing', 19, 97) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('savannah', 0, 0) ('ˈ', 936, 1) ('baltimore', 1992, 2) ('jet', 116, 3) ('flights', 9, 4) ### +### ('airlines', 1, 5) ('crashing', 190, 6) ('airways', 14, 7) ('##eg', 21655, 8) ('$', 4100, 9) ### +### ('##ο', 214, 10) ('flight', 5, 11) ('unwilling', 164, 12) ('price', 8878, 13) ### +### ('stumbled', 1437, 14) ('cost', 14717, 15) ('airline', 12, 16) ('cheap', 10307, 17) ### +### ('##ue', 16057, 18) ('costs', 16048, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('savannah', 0, 0) ('airlines', 1, 5) ('georgia', 2, 189) ('flight', 5, 11) ('fly', 6, 34) ### +### ('flights', 9, 4) ('into', 3, 790) ('flying', 7, 29) ('airways', 14, 7) ('airline', 12, 16) ### +### ('onto', 13, 80) ('travel', 22, 46) ('aircraft', 10, 211) ('air', 24, 53) ('boeing', 19, 97) ### +### ('jet', 116, 3) ('flown', 32, 62) ('aviation', 23, 113) ('flies', 21, 142) ('airport', 30, 102) ### +############################################################################################################ +[2023-10-08 01:30:36,324][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:30:36,324][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:30:36,748][root][INFO] - Epoch: 18: Step: 1101/1557, loss[v]=0.056004, lr=0.000001, acc@1[1]=245.0/256=0.95703125, acc@1[2]=253.0/256=0.98828125 +[2023-10-08 01:31:54,121][root][INFO] - Train batch 1200 +[2023-10-08 01:31:54,121][root][INFO] - Avg. loss per last 100 batches: 0.057748 +[2023-10-08 01:31:54,830][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29098.7/29522=98.57% | mean: 0.01 | max: 5.37 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.25 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] when two variables on the coordinate [SEP] ### +### [P_TEXT]: [CLS] when working with equations that have two variables, the coordinate plane is an ### +### important tool. it's a way to draw pictures of equations that makes them easier to understand. to ### +### create a coordinate plane, start with a sheet of graph or grid paper. next, draw a horizontal line. ### +### [SEP] ### +### ======================================= h_v_q | Gates: 26929 ======================================= ### +### ('coordinate', 0, 0) ('variables', 1, 11) ('two', 2, 50) ('.', 3, 11976) ('variable', 4, 78) ### +### ('when', 5, 1027) ('on', 6, 17936) ('familiarity', 7, 25492) ('stylized', 8, 28319) ### +### ('twin', 9, 104) ('2017', 10, 23187) ('factors', 11, 807) ('onto', 12, 422) ('august', 13, 14167) ### +### ('april', 14, 16511) ('united', 15, 17683) ('spring', 16, 5254) ('september', 17, 12161) ### +### ('suddenly', 18, 455) ('twins', 19, 143) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('coordinate', 0, 0) ('plane', 8607, 1) ('horizontal', 11811, 2) ('equations', 1687, 3) ### +### ('planes', 10396, 4) ('hating', 130, 5) ('crashing', 102, 6) ('ˈ', 153, 7) ('##ο', 72, 8) ### +### ('unwilling', 76, 9) ('hesitated', 180, 10) ('variables', 1, 11) ('important', 1517, 12) ### +### ('tool', 3063, 13) ('wingspan', 281, 14) ('vertical', 1941, 15) ('cyrillic', 196, 16) ### +### ('##₂', 37, 17) ('##ང', 178, 18) ('sharply', 122, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('coordinate', 0, 0) ('variables', 1, 11) ('two', 2, 50) ('variable', 4, 78) ('twin', 9, 104) ### +### ('simon', 22, 56) ('##₂', 37, 17) ('ruined', 43, 28) ('##ο', 72, 8) ('unwilling', 76, 9) ### +### ('##α', 40, 40) ('crashing', 102, 6) ('when', 5, 1027) ('−', 62, 22) ('2', 27, 86) ### +### ('twins', 19, 143) ('hating', 130, 5) ('julian', 42, 67) ('ˈ', 153, 7) ('crashed', 79, 29) ### +############################################################################################################ +[2023-10-08 01:31:54,831][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:31:54,831][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:31:55,254][root][INFO] - Epoch: 18: Step: 1201/1557, loss[v]=0.064347, lr=0.000001, acc@1[1]=245.0/256=0.95703125, acc@1[2]=251.5/256=0.982421875 +[2023-10-08 01:33:11,494][root][INFO] - Train batch 1300 +[2023-10-08 01:33:11,495][root][INFO] - Avg. loss per last 100 batches: 0.058136 +[2023-10-08 01:33:12,218][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29084.5/29522=98.52% | mean: 0.01 | max: 5.39 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.20 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] when is anna duggar due for her baby [SEP] ### +### [P_TEXT]: [CLS] anna duggar is currently pregnant with her fourth child, with a due date of july ### +### 10. with her due date just two days away, jessa notes that it is likely that the newest duggar ### +### grandchild will be here soon. [SEP] ### +### ======================================= h_v_q | Gates: 27357 ======================================= ### +### ('##gar', 0, 1) ('anna', 1, 5) ('baby', 2, 35) ('dug', 3, 0) ('due', 4, 24) ('her', 5, 277) ### +### ('2017', 6, 14939) ('april', 7, 1756) ('.', 8, 15869) ('weeks', 9, 71) ('september', 10, 906) ### +### ('august', 11, 1642) ('familiarity', 12, 21555) ('child', 13, 23) ('july', 14, 7) ### +### ('march', 15, 3586) ('february', 16, 2806) ('months', 17, 1594) ('november', 18, 10235) ### +### ('december', 19, 3102) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('dug', 3, 0) ('##gar', 0, 1) ('jess', 12715, 2) ('pregnant', 584, 3) ('ˈ', 163, 4) ('anna', 1, 5) ### +### ('cyrillic', 157, 6) ('july', 14, 7) ('hesitated', 153, 8) ('fourth', 4086, 9) ('##ང', 138, 10) ### +### ('##child', 22284, 11) ('pregnancy', 55, 12) ('date', 31, 13) ('crashing', 110, 14) ### +### ('unwilling', 116, 15) ('ছ', 104, 16) ('##ο', 173, 17) ('−', 57, 18) ('stumbled', 670, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('##gar', 0, 1) ('dug', 3, 0) ('anna', 1, 5) ('baby', 2, 35) ('due', 4, 24) ('july', 14, 7) ### +### ('child', 13, 23) ('her', 5, 277) ('weeks', 9, 71) ('date', 31, 13) ('dig', 20, 48) ### +### ('babies', 30, 38) ('june', 25, 83) ('pregnancy', 55, 12) ('−', 57, 18) ('summer', 27, 143) ### +### ('children', 73, 28) ('ছ', 104, 16) ('ˈ', 163, 4) ('digging', 48, 73) ### +############################################################################################################ +[2023-10-08 01:33:12,218][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:33:12,218][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:33:12,642][root][INFO] - Epoch: 18: Step: 1301/1557, loss[v]=0.048032, lr=0.000001, acc@1[1]=242.5/256=0.947265625, acc@1[2]=253.0/256=0.98828125 +[2023-10-08 01:34:29,891][root][INFO] - Train batch 1400 +[2023-10-08 01:34:29,891][root][INFO] - Avg. loss per last 100 batches: 0.056690 +[2023-10-08 01:34:30,598][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29057.9/29522=98.43% | mean: 0.01 | max: 5.56 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.5/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.4/29522=100.00% | mean: 0.14 | max: 6.11 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what county is plain city in? [SEP] ### +### [P_TEXT]: [CLS] plain city, madison county, oh foreclosures. search for foreclosed homes for sale ### +### in plain city, madison county, ohio. [SEP] ### +### ======================================= h_v_q | Gates: 26653 ======================================= ### +### ('plain', 0, 0) ('city', 1, 3) ('county', 2, 7) ('familiarity', 3, 20339) ('stylized', 4, 28621) ### +### ('is', 5, 19493) ('consisting', 6, 25739) ('town', 7, 62) ('plural', 8, 12319) ### +### ('encompasses', 9, 647) ('mathematics', 10, 25823) ('relating', 11, 25828) ('census', 12, 4532) ### +### ('league', 13, 3759) ('urban', 14, 75) ('##sam', 15, 25971) ('plains', 16, 140) ### +### ('counties', 17, 44) ('cities', 18, 11) ('campaign', 19, 861) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('plain', 0, 0) ('madison', 866, 1) ('ohio', 1240, 2) ('city', 1, 3) ('search', 6711, 4) ### +### ('ˈ', 127, 5) ('homes', 3239, 6) ('county', 2, 7) ('crashing', 46, 8) ('fore', 23002, 9) ### +### ('##大', 132, 10) ('cities', 18, 11) ('unwilling', 27, 12) ('−', 83, 13) ('stumbled', 321, 14) ### +### ('##ང', 76, 15) ('##ο', 345, 16) ('ছ', 98, 17) ('downtown', 42, 18) ('shoved', 117, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('plain', 0, 0) ('city', 1, 3) ('county', 2, 7) ('town', 7, 62) ('cities', 18, 11) ### +### ('counties', 17, 44) ('unwilling', 27, 12) ('urban', 14, 75) ('crashing', 46, 8) ('simon', 23, 43) ### +### ('downtown', 42, 18) ('crashed', 41, 23) ('hugh', 36, 29) ('angrily', 40, 34) ('ˈ', 127, 5) ### +### ('−', 83, 13) ('##ང', 76, 15) ('stark', 44, 33) ('plains', 16, 140) ('∈', 26, 84) ### +############################################################################################################ +[2023-10-08 01:34:30,598][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:34:30,599][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:34:31,020][root][INFO] - Epoch: 18: Step: 1401/1557, loss[v]=0.048950, lr=0.000001, acc@1[1]=245.0/256=0.95703125, acc@1[2]=250.0/256=0.9765625 +[2023-10-08 01:35:47,631][root][INFO] - Train batch 1500 +[2023-10-08 01:35:47,632][root][INFO] - Avg. loss per last 100 batches: 0.058416 +[2023-10-08 01:35:48,300][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29123.4/29522=98.65% | mean: 0.01 | max: 5.44 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.6/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.14 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what do. robins eat [SEP] ### +### [P_TEXT]: [CLS] robins also eat other small insects, berries, fallen fruits, and especially ### +### rosaceae fruit in the fall and winter. robins cannot digest hard fruits or grains, so they stick to ### +### soft fruits and organisms with softer skeletons ( howell, 1942 ). http : / / www. bio. davidson. ### +### edu / people / vecas... [SEP] ### +### ======================================= h_v_q | Gates: 27274 ======================================= ### +### ('robin', 0, 0) ('eat', 1, 7) ('##s', 2, 77) ('eating', 3, 67) ('foods', 4, 52) ('food', 5, 114) ### +### ('.', 6, 17496) ('fish', 7, 2550) ('familiarity', 8, 26003) ('stylized', 9, 28975) ### +### ('what', 10, 568) ('snack', 11, 362) ('do', 12, 1127) ('feed', 13, 90) ('protein', 14, 1157) ### +### ('group', 15, 5662) ('eaten', 16, 116) ('relating', 17, 27096) ('consisting', 18, 23156) ### +### ('destroy', 19, 1013) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('robin', 0, 0) ('howell', 7750, 1) ('##cas', 19044, 2) ('insects', 479, 3) ('berries', 722, 4) ### +### ('winter', 12637, 5) ('fruits', 758, 6) ('eat', 1, 7) ('##ceae', 4359, 8) ('fall', 7111, 9) ### +### ('hating', 178, 10) ('##ο', 169, 11) ('fruit', 56, 12) ('crashing', 109, 13) ('ˈ', 241, 14) ### +### ('unwilling', 175, 15) ('hesitated', 306, 16) ('wingspan', 753, 17) ('skeletons', 22662, 18) ### +### ('soft', 2682, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('robin', 0, 0) ('eat', 1, 7) ('##s', 2, 77) ('eating', 3, 67) ('foods', 4, 52) ('food', 5, 114) ### +### ('feed', 13, 90) ('eaten', 16, 116) ('fruit', 56, 12) ('##ς', 23, 64) ('simon', 43, 45) ### +### ('meal', 21, 144) ('crashing', 109, 13) ('crashed', 73, 36) ('−', 113, 21) ('snack', 11, 362) ### +### ('##ο', 169, 11) ('hating', 178, 10) ('what', 10, 568) ('##α', 60, 69) ### +############################################################################################################ +[2023-10-08 01:35:48,301][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:35:48,301][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:35:48,714][root][INFO] - Epoch: 18: Step: 1501/1557, loss[v]=0.052415, lr=0.000001, acc@1[1]=241.0/256=0.94140625, acc@1[2]=252.5/256=0.986328125 +[2023-10-08 01:36:31,962][root][INFO] - rank=2; last iteration 1557 +[2023-10-08 01:36:31,962][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 01:36:31,962][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-08 01:36:31,968][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 01:36:31,968][root][INFO] - Epoch finished on 2 +[2023-10-08 01:36:31,986][root][INFO] - rank=1; last iteration 1557 +[2023-10-08 01:36:31,986][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 01:36:31,986][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-08 01:36:31,987][root][INFO] - rank=0; last iteration 1557 +[2023-10-08 01:36:31,987][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 01:36:31,988][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-08 01:36:31,989][root][INFO] - rank=3; last iteration 1557 +[2023-10-08 01:36:31,990][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 01:36:31,990][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-08 01:36:31,993][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 01:36:31,993][root][INFO] - Epoch finished on 1 +[2023-10-08 01:36:31,995][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 01:36:31,995][root][INFO] - Epoch finished on 0 +[2023-10-08 01:36:31,997][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 01:36:31,997][root][INFO] - Epoch finished on 3 +[2023-10-08 01:36:42,168][root][INFO] - Saved checkpoint at ./vdr_18 +[2023-10-08 01:36:42,169][root][INFO] - Av Loss per epoch=0.057619 +[2023-10-08 01:36:42,169][root][INFO] - epoch total (1) correct predictions=380169 +[2023-10-08 01:36:42,169][root][INFO] - epoch total (2) correct predictions=392114 +[2023-10-08 01:36:42,172][root][INFO] - ***** Epoch 19 ***** +[2023-10-08 01:36:42,176][root][INFO] - rank=2; Iteration start +[2023-10-08 01:36:42,176][root][INFO] - rank=2; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 01:36:42,176][root][INFO] - rank=2; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 01:36:42,177][root][INFO] - rank=2; data_src_indices len=1557 +[2023-10-08 01:36:47,510][root][INFO] - Saved checkpoint at ./vdr_18 +[2023-10-08 01:36:47,510][root][INFO] - Saved checkpoint at ./vdr_18 +[2023-10-08 01:36:47,511][root][INFO] - Av Loss per epoch=0.057619 +[2023-10-08 01:36:47,510][root][INFO] - Saved checkpoint at ./vdr_18 +[2023-10-08 01:36:47,511][root][INFO] - epoch total (1) correct predictions=380169 +[2023-10-08 01:36:47,511][root][INFO] - Av Loss per epoch=0.057619 +[2023-10-08 01:36:47,511][root][INFO] - epoch total (2) correct predictions=392114 +[2023-10-08 01:36:47,511][root][INFO] - Av Loss per epoch=0.057619 +[2023-10-08 01:36:47,511][root][INFO] - epoch total (1) correct predictions=380169 +[2023-10-08 01:36:47,511][root][INFO] - epoch total (1) correct predictions=380169 +[2023-10-08 01:36:47,511][root][INFO] - epoch total (2) correct predictions=392114 +[2023-10-08 01:36:47,511][root][INFO] - epoch total (2) correct predictions=392114 +[2023-10-08 01:36:47,514][root][INFO] - ***** Epoch 19 ***** +[2023-10-08 01:36:47,514][root][INFO] - ***** Epoch 19 ***** +[2023-10-08 01:36:47,515][root][INFO] - ***** Epoch 19 ***** +[2023-10-08 01:36:47,520][root][INFO] - rank=0; Iteration start +[2023-10-08 01:36:47,520][root][INFO] - rank=1; Iteration start +[2023-10-08 01:36:47,520][root][INFO] - rank=0; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 01:36:47,520][root][INFO] - rank=1; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 01:36:47,520][root][INFO] - rank=0; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 01:36:47,520][root][INFO] - rank=1; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 01:36:47,521][root][INFO] - rank=3; Iteration start +[2023-10-08 01:36:47,522][root][INFO] - rank=3; Multi set iteration: iteration ptr per set: [0] +[2023-10-08 01:36:47,522][root][INFO] - rank=3; Multi set iteration: source 0, batches to be taken: 1557 +[2023-10-08 01:36:47,522][root][INFO] - rank=0; data_src_indices len=1557 +[2023-10-08 01:36:47,522][root][INFO] - rank=1; data_src_indices len=1557 +[2023-10-08 01:36:47,524][root][INFO] - rank=3; data_src_indices len=1557 +[2023-10-08 01:36:48,449][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29094.3/29522=98.55% | mean: 0.01 | max: 5.56 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.40 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] average weather prescott az [SEP] ### +### [P_TEXT]: [CLS] prescott valley, arizona, gets 16 inches of rain per year. the us average is 39. ### +### snowfall is 12 inches. the average us city gets 26 inches of snow per year. the number of days with ### +### any measurable precipitation is 36. on average, there are 273 sunny days per year in prescott ### +### valley, arizona. the july high is around 90 degrees. the january low is 25. sperling's comfort ### +### index for prescott valley is a 77 out of 100, where a higher score indicates a more comfortable ### +### year - around climate. the us average for the comfort index is 54. [SEP] ### +### ======================================= h_v_q | Gates: 26371 ======================================= ### +### ('prescott', 0, 0) ('average', 1, 22) ('weather', 2, 25) ('arizona', 3, 1) ('normal', 4, 158) ### +### ('.', 5, 6155) ('##°', 6, 526) ('winds', 7, 18944) ('az', 8, 60) ('colorado', 9, 554) ### +### ('temperature', 10, 30) ('familiarity', 11, 20552) ('utah', 12, 304) ('georgia', 13, 10590) ### +### ('degrees', 14, 69) ('30', 15, 71) ('summer', 16, 1095) ('alabama', 17, 9055) ('climate', 18, 4) ### +### ('peaked', 19, 7799) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('prescott', 0, 0) ('arizona', 3, 1) ('valley', 319, 2) ('ˈ', 2109, 3) ('climate', 18, 4) ### +### ('##asurable', 26016, 5) ('precipitation', 20, 6) ('snow', 35, 7) ('sunny', 232, 8) ### +### ('crashing', 461, 9) ('comfortable', 1054, 10) ('hating', 682, 11) ('rain', 26, 12) ### +### ('##ο', 683, 13) ('comfort', 2161, 14) ('unwilling', 204, 15) ('cyrillic', 827, 16) ### +### ('wingspan', 1810, 17) ('##ང', 830, 18) ('stumbled', 1127, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('prescott', 0, 0) ('average', 1, 22) ('arizona', 3, 1) ('weather', 2, 25) ('az', 8, 60) ### +### ('normal', 4, 158) ('temperature', 10, 30) ('climate', 18, 4) ('precipitation', 20, 6) ### +### ('rain', 26, 12) ('degrees', 14, 69) ('snow', 35, 7) ('30', 15, 71) ('##°', 6, 526) ### +### ('july', 43, 43) ('january', 32, 57) ('inches', 41, 56) ('utah', 12, 304) ('simon', 63, 41) ### +### ('−', 102, 21) ### +############################################################################################################ +[2023-10-08 01:36:48,449][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:36:48,449][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:36:48,865][root][INFO] - Epoch: 19: Step: 1/1557, loss[v]=0.077553, lr=0.000001, acc@1[1]=249.5/256=0.974609375, acc@1[2]=254.0/256=0.9921875 +[2023-10-08 01:38:05,080][root][INFO] - Train batch 100 +[2023-10-08 01:38:05,081][root][INFO] - Avg. loss per last 100 batches: 0.055470 +[2023-10-08 01:38:05,798][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29103.7/29522=98.58% | mean: 0.01 | max: 5.36 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.4/29522=100.00% | mean: 0.15 | max: 6.12 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what causes vitiligo skin disease [SEP] ### +### [P_TEXT]: [CLS] causes of vitiligo. melanin is the pigment that gives the skin its characteristic ### +### color. vitiligo is caused by a loss of pigment in the skin, due to destruction of pigment - forming ### +### cells known as melanocytes. the exact cause of the destruction of these cells is not known. ymptoms ### +### of vitiligo include an often rapid pigment loss on several areas of the skin. the initial ### +### appearance of the white patches can be followed by a stable period without any progression of the ### +### condition. later on, further cycles of pigment loss and stability may be observed. [SEP] ### +### ======================================= h_v_q | Gates: 27730 ======================================= ### +### ('vi', 0, 1) ('skin', 1, 7) ('##go', 2, 22) ('##ti', 3, 41) ('##li', 4, 79) ('disease', 5, 3611) ### +### ('causes', 6, 2) ('caused', 7, 100) ('cause', 8, 32) ('familiarity', 9, 26844) ### +### ('stylized', 10, 28839) ('.', 11, 19115) ('relating', 12, 24249) ('diseases', 13, 492) ### +### ('consisting', 14, 24400) ('mathematics', 15, 25159) ('arises', 16, 1109) ('outbreak', 17, 2421) ### +### ('unwilling', 18, 13) ('plural', 19, 12634) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('pigment', 1796, 0) ('vi', 0, 1) ('causes', 6, 2) ('ˈ', 90, 3) ('hesitated', 46, 4) ### +### ('hating', 29, 5) ('cyrillic', 38, 6) ('skin', 1, 7) ('−', 26, 8) ('##ο', 40, 9) ### +### ('encompasses', 1237, 10) ('crashing', 27, 11) ('stable', 13108, 12) ('unwilling', 18, 13) ### +### ('sharply', 37, 14) ('##ང', 61, 15) ('characteristic', 4822, 16) ('mel', 10656, 17) ### +### ('color', 3015, 18) ('stability', 8616, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('vi', 0, 1) ('skin', 1, 7) ('##go', 2, 22) ('causes', 6, 2) ('##ti', 3, 41) ('##li', 4, 79) ### +### ('caused', 7, 100) ('cause', 8, 32) ('unwilling', 18, 13) ('−', 26, 8) ('hating', 29, 5) ### +### ('crashing', 27, 11) ('##₂', 30, 25) ('hugh', 25, 33) ('ruined', 31, 27) ('cyrillic', 38, 6) ### +### ('crashed', 28, 34) ('sharply', 37, 14) ('##α', 22, 56) ('angrily', 33, 28) ### +############################################################################################################ +[2023-10-08 01:38:05,799][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:38:05,799][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:38:06,205][root][INFO] - Epoch: 19: Step: 101/1557, loss[v]=0.084248, lr=0.000001, acc@1[1]=241.0/256=0.94140625, acc@1[2]=249.0/256=0.97265625 +[2023-10-08 01:39:23,143][root][INFO] - Train batch 200 +[2023-10-08 01:39:23,144][root][INFO] - Avg. loss per last 100 batches: 0.056240 +[2023-10-08 01:39:23,835][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29090.0/29522=98.54% | mean: 0.01 | max: 5.38 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.42 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what gene causes cystic fibrosis [SEP] ### +### [P_TEXT]: [CLS] cf is caused by a mutation in the gene cystic fibrosis transmembrane conductance ### +### regulator ( cftr ). the most common mutation, i´f508, is a deletion ( i´ signifying deletion ) of ### +### three nucleotides that results in a loss of the amino acid phenylalanine ( f ) at the 508th ### +### position on the protein. [SEP] ### +### ======================================= h_v_q | Gates: 28051 ======================================= ### +### ('cy', 0, 42) ('gene', 1, 5) ('##sis', 2, 66) ('##bro', 3, 43) ('fi', 4, 28) ('causes', 5, 34) ### +### ('cause', 6, 79) ('caused', 7, 112) ('familiarity', 8, 26358) ('.', 9, 14029) ### +### ('stylized', 10, 28430) ('##stic', 11, 86) ('genes', 12, 41) ('consisting', 13, 24716) ### +### ('relating', 14, 27783) ('genetic', 15, 115) ('simon', 16, 75) ('mutation', 17, 1) ### +### ('noun', 18, 24496) ('arises', 19, 782) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('cf', 6038, 0) ('mutation', 17, 1) ('##tr', 26736, 2) ('##rane', 26842, 3) ('ˈ', 92, 4) ### +### ('gene', 1, 5) ('hating', 73, 6) ('trans', 8552, 7) ('mutations', 62, 8) ('crashing', 28, 9) ### +### ('##ο', 68, 10) ('sharply', 50, 11) ('protein', 31, 12) ('cyrillic', 114, 13) ### +### ('conduct', 11353, 14) ('hesitated', 76, 15) ('wingspan', 281, 16) ('unwilling', 36, 17) ### +### ('−', 22, 18) ('stumbled', 100, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('gene', 1, 5) ('cy', 0, 42) ('##bro', 3, 43) ('fi', 4, 28) ('##sis', 2, 66) ('causes', 5, 34) ### +### ('cause', 6, 79) ('caused', 7, 112) ('mutation', 17, 1) ('genes', 12, 41) ('##stic', 11, 86) ### +### ('−', 22, 18) ('crashing', 28, 9) ('angrily', 27, 24) ('protein', 31, 12) ('hugh', 26, 27) ### +### ('##₂', 25, 29) ('crashed', 23, 37) ('unwilling', 36, 17) ('simon', 16, 75) ### +############################################################################################################ +[2023-10-08 01:39:23,836][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:39:23,836][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:39:24,242][root][INFO] - Epoch: 19: Step: 201/1557, loss[v]=0.087996, lr=0.000001, acc@1[1]=238.0/256=0.9296875, acc@1[2]=251.5/256=0.982421875 +[2023-10-08 01:40:40,797][root][INFO] - Train batch 300 +[2023-10-08 01:40:40,798][root][INFO] - Avg. loss per last 100 batches: 0.055637 +[2023-10-08 01:40:41,513][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29085.8/29522=98.52% | mean: 0.01 | max: 5.33 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.4/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.33 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the thorax in humans [SEP] ### +### [P_TEXT]: [CLS] an example of the thorax is where the lungs are located under the ribcage. pl. - a ### +### · raxa · es or - a · raa · cesa ·. 1 in tetrapods, including humans, the part of the body cavity ### +### from the neck or head to the abdomen, containing the heart, lungs, etc. ; chest : in mammals, the ### +### diaphragm separates it from the abdomen. [SEP] ### +### ======================================= h_v_q | Gates: 27408 ======================================= ### +### ('thor', 0, 2) ('##ax', 1, 1) ('humans', 2, 41) ('is', 3, 7292) ('familiarity', 4, 25337) ### +### ('encompasses', 5, 29) ('stylized', 6, 28425) ('.', 7, 16864) ('human', 8, 124) ### +### ('consisting', 9, 20508) ('plural', 10, 6986) ('stands', 11, 10545) ('##sam', 12, 28548) ### +### ('encyclopedia', 13, 11480) ('relating', 14, 25626) ('definition', 15, 24) ('people', 16, 1961) ### +### ('narrow', 17, 5200) ('mathematics', 18, 24476) ('refers', 19, 13131) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('lungs', 10619, 0) ('##ax', 1, 1) ('thor', 0, 2) ('chest', 2109, 3) ('abdomen', 13714, 4) ### +### ('ˈ', 146, 5) ('examples', 5983, 6) ('mammals', 124, 7) ('crashing', 43, 8) ('cavity', 19458, 9) ### +### ('##pods', 18431, 10) ('example', 10443, 11) ('rib', 24178, 12) ('gideon', 53, 13) ### +### ('hating', 85, 14) ('hesitated', 66, 15) ('·', 1170, 16) ('cyrillic', 92, 17) ('unwilling', 28, 18) ### +### ('separates', 12307, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('thor', 0, 2) ('##ax', 1, 1) ('humans', 2, 41) ('encompasses', 5, 29) ('human', 8, 124) ### +### ('definition', 15, 24) ('unwilling', 28, 18) ('crashing', 43, 8) ('ruined', 25, 44) ### +### ('angrily', 37, 26) ('body', 22, 56) ('simon', 20, 66) ('gideon', 53, 13) ('crashed', 40, 38) ### +### ('−', 48, 30) ('shoved', 45, 40) ('hesitated', 66, 15) ('##α', 39, 59) ('hating', 85, 14) ### +### ('ছ', 62, 23) ### +############################################################################################################ +[2023-10-08 01:40:41,513][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:40:41,513][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:40:41,939][root][INFO] - Epoch: 19: Step: 301/1557, loss[v]=0.035875, lr=0.000001, acc@1[1]=245.0/256=0.95703125, acc@1[2]=256.0/256=1.0 +[2023-10-08 01:41:58,485][root][INFO] - Train batch 400 +[2023-10-08 01:41:58,486][root][INFO] - Avg. loss per last 100 batches: 0.056210 +[2023-10-08 01:41:59,198][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29123.6/29522=98.65% | mean: 0.01 | max: 5.47 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.37 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] meaning of the name colette [SEP] ### +### [P_TEXT]: [CLS] the name colette is an american baby name. in american the meaning of the name ### +### colette is : necklace. victorious. a middle ages feminine form of nicholas which was originally a. ### +### famous bearers - 15th century french nun st colette, and 20th century french writer colette. he ### +### name colette is a greek baby name. in greek the meaning of the name colette is : people's victory. ### +### st. nicholas is the patron saint of children, sailors, and pawnbrokers - santa claus is based on ### +### this saint. [SEP] ### +### ======================================= h_v_q | Gates: 26587 ======================================= ### +### ('cole', 0, 0) ('##tte', 1, 1) ('meaning', 2, 7) ('name', 3, 15) ('familiarity', 4, 26070) ### +### ('stylized', 5, 28015) ('noun', 6, 18386) ('definition', 7, 74) ('surname', 8, 48) ### +### ('genus', 9, 4633) ('consisting', 10, 23431) ('plural', 11, 8468) ('expression', 12, 1170) ### +### ('purpose', 13, 1182) ('sense', 14, 723) ('meanings', 15, 8) ('means', 16, 35) ### +### ('relating', 17, 27469) ('mathematics', 18, 24741) ('##º', 19, 28234) ### +### ======================================= h_v_p | Gates: 29522 ======================================= ### +### ('cole', 0, 0) ('##tte', 1, 1) ('necklace', 11166, 2) ('patron', 10291, 3) ('nicholas', 4803, 4) ### +### ('ˈ', 172, 5) ('victorious', 20562, 6) ('meaning', 2, 7) ('meanings', 15, 8) ('##ο', 125, 9) ### +### ('pawn', 13037, 10) ('saint', 3906, 11) ('claus', 20979, 12) ('crashing', 67, 13) ### +### ('##ttes', 32, 14) ('name', 3, 15) ('−', 90, 16) ('cyrillic', 121, 17) ('names', 40, 18) ### +### ('stumbled', 159, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('cole', 0, 0) ('##tte', 1, 1) ('meaning', 2, 7) ('name', 3, 15) ('meanings', 15, 8) ### +### ('surname', 8, 48) ('definition', 7, 74) ('means', 16, 35) ('##ttes', 32, 14) ('names', 40, 18) ### +### ('crashing', 67, 13) ('shoved', 45, 34) ('mean', 62, 22) ('nickname', 51, 39) ('unwilling', 79, 21) ### +### ('−', 90, 16) ('angrily', 59, 33) ('##ο', 125, 9) ('julian', 30, 86) ('##ང', 84, 23) ### +############################################################################################################ +[2023-10-08 01:41:59,198][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:41:59,198][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:41:59,619][root][INFO] - Epoch: 19: Step: 401/1557, loss[v]=0.045891, lr=0.000001, acc@1[1]=242.5/256=0.947265625, acc@1[2]=254.0/256=0.9921875 +[2023-10-08 01:43:16,294][root][INFO] - Train batch 500 +[2023-10-08 01:43:16,295][root][INFO] - Avg. loss per last 100 batches: 0.052945 +[2023-10-08 01:43:16,985][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29101.9/29522=98.58% | mean: 0.01 | max: 5.88 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.47 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what nationality is paul ryan parents [SEP] ### +### [P_TEXT]: [CLS] paul ryan. by ethnic on. birth name : paul davis ryan. place of birth : janesville, ### +### wisconsin. date of birth : january 29, 1970. ethnicity : * irish ( father ) * german ( maternal ### +### grandfather ) * english, german ( maternal grandmother ). paul ryan is an american politician. his ### +### father was of irish descent. his maternal grandfather was of german ancestry, and his maternal ### +### grandmother was of english and german ancestry. he was the republican partyas nominee for vice ### +### president of the united states in 2012. [SEP] ### +### ======================================= h_v_q | Gates: 27441 ======================================= ### +### ('ryan', 0, 0) ('paul', 1, 1) ('parents', 2, 130) ('nationality', 3, 23) ('ethnic', 4, 6) ### +### ('father', 5, 20) ('origin', 6, 115) ('stylized', 7, 28363) ('american', 8, 149) ('mother', 9, 263) ### +### ('.', 10, 11927) ('is', 11, 660) ('familiarity', 12, 25199) ('language', 13, 11218) ### +### ('parent', 14, 253) ('formerly', 15, 8247) ('international', 16, 11360) ('citizenship', 17, 151) ### +### ('family', 18, 3272) ('heritage', 19, 52) ### +### ======================================= h_v_p | Gates: 29520 ======================================= ### +### ('ryan', 0, 0) ('paul', 1, 1) ('davis', 3582, 2) ('jane', 4895, 3) ('ethnicity', 33, 4) ### +### ('wisconsin', 2184, 5) ('ethnic', 4, 6) ('ˈ', 378, 7) ('birth', 20, 8) ('crashing', 335, 9) ### +### ('grandfather', 38, 10) ('##ο', 710, 11) ('politician', 332, 12) ('stumbled', 453, 13) ### +### ('##ང', 310, 14) ('maternal', 2265, 15) ('irish', 671, 16) ('ancestry', 24, 17) ('ছ', 396, 18) ### +### ('cyrillic', 439, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ryan', 0, 0) ('paul', 1, 1) ('nationality', 3, 23) ('ethnic', 4, 6) ('parents', 2, 130) ### +### ('father', 5, 20) ('birth', 20, 8) ('origin', 6, 115) ('ethnicity', 33, 4) ('ancestry', 24, 17) ### +### ('american', 8, 149) ('heritage', 19, 52) ('grandfather', 38, 10) ('mother', 9, 263) ### +### ('birthplace', 39, 43) ('citizenship', 17, 151) ('racial', 40, 53) ('parent', 14, 253) ### +### ('grandparents', 43, 84) ('born', 28, 119) ### +############################################################################################################ +[2023-10-08 01:43:16,986][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:43:16,986][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:43:17,409][root][INFO] - Epoch: 19: Step: 501/1557, loss[v]=0.088436, lr=0.000001, acc@1[1]=241.5/256=0.943359375, acc@1[2]=251.0/256=0.98046875 +[2023-10-08 01:44:34,005][root][INFO] - Train batch 600 +[2023-10-08 01:44:34,006][root][INFO] - Avg. loss per last 100 batches: 0.057336 +[2023-10-08 01:44:34,713][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29133.8/29522=98.69% | mean: 0.01 | max: 5.25 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.29 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] the preferred source of energy for the brain is [SEP] ### +### [P_TEXT]: [CLS] answered by the community. making the world better, one answer at a time. the ### +### preferred energy source for the brain is always glucose. it is also the preferred energy source for ### +### a fetus, as well as the central nervous system. [SEP] ### +### ======================================= h_v_q | Gates: 27205 ======================================= ### +### ('brain', 0, 1) ('energy', 1, 4) ('preferred', 2, 2) ('sources', 3, 72) ('source', 4, 25) ### +### ('.', 5, 11848) ('favorite', 6, 168) ('stylized', 7, 28999) ('familiarity', 8, 25780) ### +### ('preference', 9, 92) ('thinking', 10, 605) ('power', 11, 224) ('consisting', 12, 23007) ### +### ('mind', 13, 94) ('computer', 14, 1223) ('relating', 15, 25901) ('plural', 16, 12156) ### +### ('primary', 17, 499) ('simon', 18, 54) ('favored', 19, 87) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('glucose', 13769, 0) ('brain', 0, 1) ('preferred', 2, 2) ('ˈ', 108, 3) ('energy', 1, 4) ### +### ('hating', 67, 5) ('crashing', 33, 6) ('cyrillic', 252, 7) ('hesitated', 51, 8) ('##ο', 80, 9) ### +### ('unwilling', 25, 10) ('−', 40, 11) ('wingspan', 101, 12) ('answered', 4310, 13) ### +### ('answer', 130, 14) ('stumbled', 88, 15) ('##₂', 46, 16) ('sharply', 91, 17) ('##ང', 56, 18) ### +### ('answers', 18672, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('brain', 0, 1) ('energy', 1, 4) ('preferred', 2, 2) ('source', 4, 25) ('sources', 3, 72) ### +### ('favorite', 6, 168) ('angrily', 22, 22) ('unwilling', 25, 10) ('crashing', 33, 6) ### +### ('preference', 9, 92) ('mind', 13, 94) ('simon', 18, 54) ('−', 40, 11) ('crashed', 32, 26) ### +### ('##₂', 46, 16) ('hesitated', 51, 8) ('ruined', 36, 31) ('hating', 67, 5) ('##α', 27, 42) ### +### ('ˈ', 108, 3) ### +############################################################################################################ +[2023-10-08 01:44:34,714][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:44:34,714][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:44:35,144][root][INFO] - Epoch: 19: Step: 601/1557, loss[v]=0.065393, lr=0.000001, acc@1[1]=243.0/256=0.94921875, acc@1[2]=252.0/256=0.984375 +[2023-10-08 01:45:50,789][root][INFO] - Train batch 700 +[2023-10-08 01:45:50,790][root][INFO] - Avg. loss per last 100 batches: 0.053686 +[2023-10-08 01:45:51,511][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29117.2/29522=98.63% | mean: 0.01 | max: 5.57 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.0/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.41 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a zester [SEP] ### +### [P_TEXT]: [CLS] a kitchen zester is approximately four inches long, with a handle and a curved ### +### metal end, the top of which is perforated with a row of round holes with sharpened rims. to ### +### operate, the zester is pressed with moderate force against the fruit and drawn across its peel. the ### +### rims cut the zest from the pith underneath. [SEP] ### +### ======================================= h_v_q | Gates: 27508 ======================================= ### +### ('##ster', 0, 3) ('ze', 1, 0) ('definition', 2, 378) ('stylized', 3, 28804) ('plural', 4, 15820) ### +### ('encompasses', 5, 37) ('is', 6, 676) ('consisting', 7, 25685) ('familiarity', 8, 26475) ### +### ('relating', 9, 25851) ('refers', 10, 20198) ('##sters', 11, 19) ('noun', 12, 26433) ### +### ('##sam', 13, 27980) ('a', 14, 705) ('genus', 15, 15650) ('encyclopedia', 16, 6079) ### +### ('mathematics', 17, 22761) ('designed', 18, 13464) ('stands', 19, 5987) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('ze', 1, 0) ('handle', 2388, 1) ('kitchen', 1799, 2) ('##ster', 0, 3) ('pressed', 168, 4) ### +### ('curved', 278, 5) ('ˈ', 223, 6) ('hating', 87, 7) ('crashing', 46, 8) ('handles', 17346, 9) ### +### ('metal', 1449, 10) ('##ο', 267, 11) ('unwilling', 103, 12) ('size', 8288, 13) ('peel', 25910, 14) ### +### ('fruit', 5919, 15) ('wingspan', 75, 16) ('##ང', 92, 17) ('hesitated', 125, 18) ('##sters', 11, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ze', 1, 0) ('##ster', 0, 3) ('encompasses', 5, 37) ('##sters', 11, 19) ('definition', 2, 378) ### +### ('crashing', 46, 8) ('is', 6, 676) ('crashed', 36, 44) ('ছ', 53, 28) ('hating', 87, 7) ### +### ('##α', 56, 33) ('wingspan', 75, 16) ('sharply', 62, 24) ('−', 63, 23) ('##ང', 92, 17) ### +### ('unwilling', 103, 12) ('a', 14, 705) ('##₂', 79, 27) ('##大', 76, 30) ('pressed', 168, 4) ### +############################################################################################################ +[2023-10-08 01:45:51,511][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:45:51,511][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:45:51,918][root][INFO] - Epoch: 19: Step: 701/1557, loss[v]=0.070329, lr=0.000001, acc@1[1]=245.0/256=0.95703125, acc@1[2]=250.5/256=0.978515625 +[2023-10-08 01:47:08,243][root][INFO] - Train batch 800 +[2023-10-08 01:47:08,244][root][INFO] - Avg. loss per last 100 batches: 0.056591 +[2023-10-08 01:47:08,931][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29091.1/29522=98.54% | mean: 0.01 | max: 5.61 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.7/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.39 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] define credit limit [SEP] ### +### [P_TEXT]: [CLS] the maximum amount for which a particular borrower is approved is known as the ### +### credit limit. banks and financial services company do not extend loans past the credit limit. for ### +### example, one may have a credit card with a credit limit of $ 2, 000 ; if one attempts to put $ 2, ### +### 100 on the card, the bank will decline payment. [SEP] ### +### ======================================= h_v_q | Gates: 25495 ======================================= ### +### ('credit', 0, 3) ('limit', 1, 0) ('definition', 2, 47) ('defined', 3, 629) ('limits', 4, 14) ### +### ('relating', 5, 25048) ('noun', 6, 27197) ('##º', 7, 28277) ('refers', 8, 16806) ### +### ('plural', 9, 14968) ('familiarity', 10, 25694) ('encyclopedia', 11, 16963) ('stylized', 12, 26774) ### +### ('maximum', 13, 18) ('.', 14, 16874) ('especially', 15, 8859) ('specified', 16, 15019) ### +### ('consisting', 17, 25651) ('term', 18, 1532) ('something', 19, 12235) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('limit', 1, 0) ('banks', 6700, 1) ('ˈ', 121, 2) ('credit', 0, 3) ('extend', 1461, 4) ### +### ('hating', 423, 5) ('crashing', 156, 6) ('##ο', 400, 7) ('−', 95, 8) ('approved', 7183, 9) ### +### ('hesitated', 128, 10) ('wingspan', 614, 11) ('unwilling', 209, 12) ('amount', 497, 13) ### +### ('limits', 4, 14) ('##₂', 67, 15) ('stumbled', 373, 16) ('financial', 27, 17) ('maximum', 13, 18) ### +### ('sharply', 96, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('credit', 0, 3) ('limit', 1, 0) ('definition', 2, 47) ('limits', 4, 14) ('maximum', 13, 18) ### +### ('defined', 3, 629) ('financial', 27, 17) ('definitions', 31, 90) ('julian', 39, 67) ### +### ('##₂', 67, 15) ('ˈ', 121, 2) ('ruined', 52, 42) ('payment', 64, 22) ('encompasses', 29, 132) ### +### ('−', 95, 8) ('minimal', 43, 82) ('sharply', 96, 19) ('credits', 32, 147) ('hesitated', 128, 10) ### +### ('define', 51, 71) ### +############################################################################################################ +[2023-10-08 01:47:08,931][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:47:08,931][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:47:09,338][root][INFO] - Epoch: 19: Step: 801/1557, loss[v]=0.043541, lr=0.000001, acc@1[1]=246.5/256=0.962890625, acc@1[2]=251.0/256=0.98046875 +[2023-10-08 01:48:26,437][root][INFO] - Train batch 900 +[2023-10-08 01:48:26,438][root][INFO] - Avg. loss per last 100 batches: 0.056518 +[2023-10-08 01:48:27,139][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29143.1/29522=98.72% | mean: 0.01 | max: 5.27 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 8.1/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.16 | max: 6.16 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what group is potassium in [SEP] ### +### [P_TEXT]: [CLS] in the periodic table, potassium is one of seven elements in column ( group ) 1 ( ### +### alkali metals ) : they all have a single valence electron in their outer electron shell, which they ### +### readily give up to create an atom with a positive charge - a cation, and combine with anions to ### +### form salts. he english name for the element potassium comes from the word potash, and refers to the ### +### method by which potassium was obtained a placing in a pot the ash of burnt wood or tree leaves, ### +### adding water, heating, and evaporating the solution. [SEP] ### +### ======================================= h_v_q | Gates: 27176 ======================================= ### +### ('potassium', 0, 0) ('group', 1, 16) ('.', 2, 11911) ('groups', 3, 189) ('band', 4, 501) ### +### ('is', 5, 233) ('league', 6, 11471) ('class', 7, 537) ('unit', 8, 1428) ('familiarity', 9, 26870) ### +### ('team', 10, 3759) ('association', 11, 8795) ('collective', 12, 7535) ('encompasses', 13, 14) ### +### ('club', 14, 7350) ('stylized', 15, 28180) ('organization', 16, 8361) ('plural', 17, 15496) ### +### ('united', 18, 12431) ('sodium', 19, 64) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('potassium', 0, 0) ('element', 5506, 1) ('metals', 3918, 2) ('periodic', 17217, 3) ('ˈ', 327, 4) ### +### ('elements', 1164, 5) ('salts', 6597, 6) ('crashing', 505, 7) ('##ο', 263, 8) ('hating', 489, 9) ### +### ('##ash', 15790, 10) ('##ང', 340, 11) ('stumbled', 436, 12) ('hesitated', 237, 13) ### +### ('encompasses', 13, 14) ('unwilling', 93, 15) ('group', 1, 16) ('−', 75, 17) ('cyrillic', 788, 18) ### +### ('gideon', 189, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('potassium', 0, 0) ('group', 1, 16) ('groups', 3, 189) ('encompasses', 13, 14) ('is', 5, 233) ### +### ('band', 4, 501) ('sodium', 19, 64) ('class', 7, 537) ('−', 75, 17) ('angrily', 78, 21) ### +### ('calcium', 23, 107) ('unwilling', 93, 15) ('∈', 22, 152) ('ruined', 108, 26) ('simon', 81, 54) ### +### ('definition', 85, 61) ('##ο', 263, 8) ('##₂', 170, 20) ('sharply', 172, 22) ('unit', 8, 1428) ### +############################################################################################################ +[2023-10-08 01:48:27,139][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:48:27,139][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:48:27,564][root][INFO] - Epoch: 19: Step: 901/1557, loss[v]=0.048146, lr=0.000000, acc@1[1]=243.5/256=0.951171875, acc@1[2]=254.0/256=0.9921875 +[2023-10-08 01:49:43,683][root][INFO] - Train batch 1000 +[2023-10-08 01:49:43,684][root][INFO] - Avg. loss per last 100 batches: 0.058438 +[2023-10-08 01:49:44,394][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29122.7/29522=98.65% | mean: 0.01 | max: 5.67 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.28 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] cause of brief psychotic disorder [SEP] ### +### [P_TEXT]: [CLS] brief psychotic disorder is a sudden, short - term display of psychotic behavior, ### +### such as hallucinations or delusions, which occurs with a stressful event. causes brief psychotic ### +### disorder is triggered by extreme stress, such as a traumatic accident or loss of a loved one. it is ### +### followed by a return to the previous level of function. the person may or may not be aware of the ### +### strange behavior. this condition most often affects people in their 20s, 30s, and 40s. [SEP] ### +### ======================================= h_v_q | Gates: 26590 ======================================= ### +### ('psychotic', 0, 0) ('brief', 1, 1) ('disorder', 2, 2) ('causes', 3, 19) ('.', 4, 10234) ### +### ('cause', 5, 85) ('caused', 6, 1028) ('disorders', 7, 35) ('familiarity', 8, 22814) ### +### ('stylized', 9, 28075) ('consisting', 10, 23420) ('relating', 11, 20824) ('short', 12, 22) ### +### ('bacteria', 13, 15684) ('outbreak', 14, 2131) ('faint', 15, 379) ('of', 16, 26250) ### +### ('simon', 17, 68) ('mental', 18, 428) ('sad', 19, 2333) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('psychotic', 0, 0) ('brief', 1, 1) ('disorder', 2, 2) ('strange', 313, 3) ('stress', 4048, 4) ### +### ('behavior', 204, 5) ('−', 34, 6) ('crashing', 43, 7) ('ˈ', 271, 8) ('##usions', 19544, 9) ### +### ('##ο', 228, 10) ('hating', 73, 11) ('triggered', 405, 12) ('##₂', 35, 13) ('encompasses', 834, 14) ### +### ('weird', 436, 15) ('wingspan', 322, 16) ('sharply', 104, 17) ('define', 20412, 18) ### +### ('causes', 3, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('psychotic', 0, 0) ('brief', 1, 1) ('disorder', 2, 2) ('causes', 3, 19) ('cause', 5, 85) ### +### ('disorders', 7, 35) ('short', 12, 22) ('−', 34, 6) ('simon', 17, 68) ('crashing', 43, 7) ### +### ('##₂', 35, 13) ('unwilling', 31, 20) ('caused', 6, 1028) ('ruined', 30, 41) ('julian', 24, 78) ### +### ('hating', 73, 11) ('sketches', 40, 42) ('hesitated', 68, 21) ('hugh', 53, 44) ('##α', 51, 48) ### +############################################################################################################ +[2023-10-08 01:49:44,394][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:49:44,394][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:49:44,821][root][INFO] - Epoch: 19: Step: 1001/1557, loss[v]=0.068403, lr=0.000000, acc@1[1]=249.5/256=0.974609375, acc@1[2]=252.5/256=0.986328125 +[2023-10-08 01:51:01,606][root][INFO] - Train batch 1100 +[2023-10-08 01:51:01,607][root][INFO] - Avg. loss per last 100 batches: 0.054694 +[2023-10-08 01:51:02,323][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29149.5/29522=98.74% | mean: 0.01 | max: 5.23 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.1/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.6/29522=100.00% | mean: 0.15 | max: 6.16 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is rcs [SEP] ### +### [P_TEXT]: [CLS] rcs group. the rcs group ( rcs ) is a consumer finance business that offers its ### +### customers a range of financial services products under its own brand name and in association with a ### +### number of retailers in south africa, namibia and botswana. the two primary business areas are cards ### +### and loans. he rcs group ( rcs ) is a consumer finance business that offers its customers a range of ### +### financial services products under its own brand name and in association with a number of retailers ### +### in south africa, namibia and botswana. [SEP] ### +### ======================================= h_v_q | Gates: 27241 ======================================= ### +### ('rc', 0, 0) ('##s', 1, 5) ('definition', 2, 116) ('plural', 3, 16952) ('encompasses', 4, 9) ### +### ('is', 5, 209) ('stylized', 6, 27919) ('familiarity', 7, 25886) ('consisting', 8, 23664) ### +### ('genus', 9, 10085) ('##sam', 10, 28469) ('##ς', 11, 27) ('relating', 12, 18981) ### +### ('noun', 13, 18175) ('refers', 14, 8580) ('##स', 15, 78) ('designed', 16, 14759) ### +### ('stands', 17, 5665) ('language', 18, 22976) ('encyclopedia', 19, 18078) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('rc', 0, 0) ('namibia', 13711, 1) ('finance', 11155, 2) ('botswana', 20783, 3) ('group', 397, 4) ### +### ('##s', 1, 5) ('ˈ', 218, 6) ('africa', 1074, 7) ('crashing', 129, 8) ('encompasses', 4, 9) ### +### ('business', 1217, 10) ('stumbled', 125, 11) ('financial', 3098, 12) ('##ο', 444, 13) ### +### ('customers', 11180, 14) ('consumer', 795, 15) ('−', 92, 16) ('##₂', 85, 17) ('cards', 12604, 18) ### +### ('unwilling', 134, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('rc', 0, 0) ('##s', 1, 5) ('encompasses', 4, 9) ('definition', 2, 116) ('##ς', 11, 27) ### +### ('is', 5, 209) ('##स', 15, 78) ('rf', 21, 98) ('##ս', 42, 49) ('##₂', 85, 17) ('crashing', 129, 8) ### +### ('−', 92, 16) ('crashed', 87, 20) ('stumbled', 125, 11) ('angrily', 60, 37) ('ছ', 95, 22) ### +### ('##س', 36, 81) ('shoved', 66, 35) ('##α', 79, 30) ('julian', 41, 77) ### +############################################################################################################ +[2023-10-08 01:51:02,323][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:51:02,323][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:51:02,747][root][INFO] - Epoch: 19: Step: 1101/1557, loss[v]=0.053028, lr=0.000000, acc@1[1]=243.0/256=0.94921875, acc@1[2]=252.0/256=0.984375 +[2023-10-08 01:52:20,368][root][INFO] - Train batch 1200 +[2023-10-08 01:52:20,369][root][INFO] - Avg. loss per last 100 batches: 0.059309 +[2023-10-08 01:52:21,062][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29098.6/29522=98.57% | mean: 0.01 | max: 5.13 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.2/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.10 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is a congenital infection wikipedia [SEP] ### +### [P_TEXT]: [CLS] a diagnosis of congenital cmv infection can be made if the virus is found in an ### +### infant's urine, saliva, blood, or other body tissues during the first week after birth. antibody ### +### tests cannot be used to diagnose congenital cmv ; a diagnosis can only be made if the virus is ### +### detected during the first week of life. ue to the lower seroprevalence of hcmv in industrialized ### +### countries and higher socioeconomic groups, congenital infections are actually less common in poorer ### +### communities, where more women of child - bearing age are already seropositive. [SEP] ### +### ======================================= h_v_q | Gates: 26037 ======================================= ### +### ('congenital', 0, 0) ('infection', 1, 6) ('.', 2, 18718) ('is', 3, 6750) ('encompasses', 4, 156) ### +### ('definition', 5, 290) ('encyclopedia', 6, 13565) ('wikipedia', 7, 23426) ('refers', 8, 21032) ### +### ('##sam', 9, 28187) ('familiarity', 10, 24773) ('relating', 11, 25793) ('birth', 12, 31) ### +### ('infections', 13, 19) ('stylized', 14, 28543) ('stands', 15, 10099) ('plural', 16, 18096) ### +### ('term', 17, 13549) ('born', 18, 1286) ('noun', 19, 26939) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('congenital', 0, 0) ('cm', 14517, 1) ('diagnosis', 1477, 2) ('ˈ', 409, 3) ('hating', 649, 4) ### +### ('crashing', 172, 5) ('infection', 1, 6) ('##ο', 365, 7) ('hesitated', 334, 8) ('virus', 377, 9) ### +### ('stumbled', 589, 10) ('cyrillic', 565, 11) ('unwilling', 176, 12) ('saliva', 17019, 13) ### +### ('weeks', 3417, 14) ('infant', 1853, 15) ('sharply', 194, 16) ('−', 139, 17) ('##ང', 218, 18) ### +### ('infections', 13, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('congenital', 0, 0) ('infection', 1, 6) ('birth', 12, 31) ('infections', 13, 19) ### +### ('encompasses', 4, 156) ('definition', 5, 290) ('baby', 25, 100) ('child', 39, 62) ### +### ('infected', 82, 26) ('ছ', 80, 30) ('crashing', 172, 5) ('а', 46, 86) ('−', 139, 17) ### +### ('unwilling', 176, 12) ('julian', 54, 93) ('##₂', 111, 32) ('##大', 168, 21) ('sharply', 194, 16) ### +### ('disease', 26, 257) ('ruined', 132, 33) ### +############################################################################################################ +[2023-10-08 01:52:21,062][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:52:21,062][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:52:21,482][root][INFO] - Epoch: 19: Step: 1201/1557, loss[v]=0.045105, lr=0.000000, acc@1[1]=245.5/256=0.958984375, acc@1[2]=253.0/256=0.98828125 +[2023-10-08 01:53:37,217][root][INFO] - Train batch 1300 +[2023-10-08 01:53:37,218][root][INFO] - Avg. loss per last 100 batches: 0.054526 +[2023-10-08 01:53:37,939][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29087.2/29522=98.53% | mean: 0.01 | max: 5.50 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.9/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.4/29522=100.00% | mean: 0.15 | max: 6.23 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what is the closest airport to ogden ut [SEP] ### +### [P_TEXT]: [CLS] major airports near ogden, utah : the nearest major airport is ogden - hinckley ### +### airport ( ogd / kogd ). this airport has domestic flights and is 4 miles from the center of ogden, ### +### ut. another major airport is salt lake city international airport ( slc / kslc ), which has ### +### international and domestic flights from salt lake city, utah and is 41 miles from ogden, ut. search ### +### for direct flights from your hometown and find hotels near ogden, ut, or scroll down for more ### +### international airports or domestic airports. [SEP] ### +### ======================================= h_v_q | Gates: 26702 ======================================= ### +### ('ogden', 0, 0) ('airport', 1, 2) ('closest', 2, 51) ('utah', 3, 1) ('colorado', 4, 211) ### +### ('nearest', 5, 18) ('ut', 6, 33) ('arizona', 7, 296) ('nebraska', 8, 908) ('aircraft', 9, 250) ### +### ('familiarity', 10, 21551) ('oklahoma', 11, 414) ('.', 12, 9983) ('texas', 13, 1469) ### +### ('aviation', 14, 130) ('stylized', 15, 27009) ('arkansas', 16, 278) ('is', 17, 2647) ### +### ('nearby', 18, 17) ('idaho', 19, 245) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('ogden', 0, 0) ('utah', 3, 1) ('airport', 1, 2) ('airports', 38, 3) ('og', 22414, 4) ### +### ('salt', 325, 5) ('sl', 2343, 6) ('airfield', 22, 7) ('major', 2356, 8) ('ˈ', 705, 9) ### +### ('unwilling', 124, 10) ('crashing', 298, 11) ('stumbled', 666, 12) ('ks', 19883, 13) ('−', 203, 14) ### +### ('lake', 1250, 15) ('ko', 2928, 16) ('nearby', 18, 17) ('nearest', 5, 18) ('flights', 309, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ogden', 0, 0) ('airport', 1, 2) ('utah', 3, 1) ('closest', 2, 51) ('nearest', 5, 18) ### +### ('ut', 6, 33) ('airfield', 22, 7) ('nearby', 18, 17) ('airports', 38, 3) ('colorado', 4, 211) ### +### ('aviation', 14, 130) ('arizona', 7, 296) ('aircraft', 9, 250) ('oklahoma', 11, 414) ### +### ('airline', 45, 63) ('unwilling', 124, 10) ('arkansas', 16, 278) ('near', 56, 58) ### +### ('idaho', 19, 245) ('nebraska', 8, 908) ### +############################################################################################################ +[2023-10-08 01:53:37,939][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:53:37,939][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:53:38,366][root][INFO] - Epoch: 19: Step: 1301/1557, loss[v]=0.053841, lr=0.000000, acc@1[1]=244.5/256=0.955078125, acc@1[2]=251.0/256=0.98046875 +[2023-10-08 01:54:54,488][root][INFO] - Train batch 1400 +[2023-10-08 01:54:54,488][root][INFO] - Avg. loss per last 100 batches: 0.055085 +[2023-10-08 01:54:55,163][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29128.8/29522=98.67% | mean: 0.01 | max: 5.29 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 7.5/29522=0.03% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.3/29522=100.00% | mean: 0.15 | max: 6.09 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] earliest known common mammal ancestor [SEP] ### +### [P_TEXT]: [CLS] maintained by pioweb @ ucsc. edu. the common ancestor of placental mammals probably ### +### looked like eomaia scansoria, the earliest known placental mammal, shown here in an artist's ### +### reconstruction based on a 125 - million - year - old fossil skeleton found in china in 2002. ### +### aintained by pioweb @ ucsc. edu. the common ancestor of placental mammals probably looked like ### +### eomaia scansoria, the earliest known placental mammal, shown here in an artist's reconstruction ### +### based on a 125 - million - year - old fossil skeleton found in china in 2002. [SEP] ### +### ======================================= h_v_q | Gates: 26713 ======================================= ### +### ('ancestor', 0, 0) ('mammal', 1, 1) ('common', 2, 16) ('mammals', 3, 3) ('earliest', 4, 21) ### +### ('early', 5, 56) ('.', 6, 17878) ('known', 7, 82) ('animal', 8, 61) ('familiarity', 9, 25360) ### +### ('stylized', 10, 28661) ('oldest', 11, 80) ('commonly', 12, 226) ('human', 13, 945) ### +### ('first', 14, 169) ('inaugural', 15, 1476) ('originally', 16, 1191) ('simon', 17, 76) ### +### ('relating', 18, 24969) ('organization', 19, 7259) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('ancestor', 0, 0) ('mammal', 1, 1) ('##ntal', 24891, 2) ('mammals', 3, 3) ('ancestors', 63, 4) ### +### ('skeleton', 10035, 5) ('fossil', 982, 6) ('ˈ', 212, 7) ('place', 5684, 8) ('pi', 2760, 9) ### +### ('reconstruction', 7717, 10) ('##ο', 85, 11) ('looked', 14812, 12) ('##oma', 20631, 13) ### +### ('crashing', 50, 14) ('##ང', 136, 15) ('common', 2, 16) ('stumbled', 147, 17) ('hating', 45, 18) ### +### ('china', 1034, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('ancestor', 0, 0) ('mammal', 1, 1) ('mammals', 3, 3) ('common', 2, 16) ('earliest', 4, 21) ### +### ('early', 5, 56) ('known', 7, 82) ('animal', 8, 61) ('oldest', 11, 80) ('simon', 17, 76) ### +### ('animals', 22, 47) ('##₂', 24, 54) ('ancestors', 63, 4) ('angrily', 32, 36) ('crashing', 50, 14) ### +### ('unwilling', 38, 27) ('hating', 45, 18) ('−', 36, 39) ('cyrillic', 42, 31) ('ছ', 43, 30) ### +############################################################################################################ +[2023-10-08 01:54:55,163][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:54:55,163][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:54:55,581][root][INFO] - Epoch: 19: Step: 1401/1557, loss[v]=0.050700, lr=0.000000, acc@1[1]=246.5/256=0.962890625, acc@1[2]=254.0/256=0.9921875 +[2023-10-08 01:56:11,218][root][INFO] - Train batch 1500 +[2023-10-08 01:56:11,219][root][INFO] - Avg. loss per last 100 batches: 0.054833 +[2023-10-08 01:56:11,894][root][INFO] - + +### #################################################################################################### ### +### ############################################ INFO CARD ############################################# ### +### #################################################################################################### ### +### =========================================== INFO(h_v_q) ============================================ ### +### shape: (64, 29522) | gate: 29089.9/29522=98.54% | mean: 0.01 | max: 5.25 | min: 0.00 ### +### =========================================== INFO(bow_q) ============================================ ### +### shape: (64, 29522) | gate: 6.8/29522=0.02% | mean: 1.00 | max: 1.00 | min: 0.00 ### +### =========================================== INFO(h_v_p) ============================================ ### +### shape: (64, 29522) | gate: 29521.5/29522=100.00% | mean: 0.15 | max: 6.03 | min: 0.00 ### +### =========================================== Q-P EXAMPLES =========================================== ### +### [Q_TEXT]: [CLS] what must occur before mitosis takes place [SEP] ### +### [P_TEXT]: [CLS] when a new cell is created, it must have the same library of genetic information ### +### all other cells in the body have access to. because all the material in the new cell must come from ### +### the first cell, the original cell must make a copy of its dna before completing the process of ### +### mitosis. these two sets of dna only exist for as long as it takes the cell to undergo mitosis, ### +### which can be anywhere from 30 to 90 minutes in certain human cells. [SEP] ### +### ======================================= h_v_q | Gates: 27488 ======================================= ### +### ('mit', 0, 11) ('##osis', 1, 4) ('must', 2, 25) ('before', 3, 180) ('place', 4, 9124) ### +### ('.', 5, 17468) ('takes', 6, 132) ('occur', 7, 1687) ('familiarity', 8, 25375) ### +### ('stylized', 9, 28694) ('occurs', 10, 11002) ('simon', 11, 51) ('take', 12, 3040) ### +### ('took', 13, 8451) ('consisting', 14, 24610) ('relating', 15, 27580) ('##α', 16, 41) ### +### ('requires', 17, 229) ('julian', 18, 67) ('space', 19, 10144) ### +### ======================================= h_v_p | Gates: 29521 ======================================= ### +### ('dna', 5061, 0) ('ˈ', 73, 1) ('hating', 32, 2) ('##ο', 103, 3) ('##osis', 1, 4) ### +### ('unwilling', 35, 5) ('minutes', 1780, 6) ('crashing', 33, 7) ('hesitated', 55, 8) ### +### ('genetic', 6670, 9) ('cells', 7619, 10) ('mit', 0, 11) ('wingspan', 164, 12) ('cell', 10405, 13) ### +### ('copy', 5802, 14) ('##ང', 159, 15) ('sharply', 43, 16) ('30', 5215, 17) ('##大', 46, 18) ### +### ('cyrillic', 76, 19) ### +### ====================================== TopK-( h_v_q * h_v_p ) ====================================== ### +### ('mit', 0, 11) ('##osis', 1, 4) ('must', 2, 25) ('before', 3, 180) ('takes', 6, 132) ### +### ('simon', 11, 51) ('hating', 32, 2) ('##α', 16, 41) ('−', 21, 21) ('##₂', 25, 20) ### +### ('unwilling', 35, 5) ('crashing', 33, 7) ('angrily', 24, 24) ('ˈ', 73, 1) ('sharply', 43, 16) ### +### ('hesitated', 55, 8) ('##大', 46, 18) ('julian', 18, 67) ('ruined', 29, 37) ('occur', 7, 1687) ### +############################################################################################################ +[2023-10-08 01:56:11,894][root][INFO] - h_v_q.shape: torch.Size([256, 29522]) +[2023-10-08 01:56:11,895][root][INFO] - h_v_p.shape: torch.Size([512, 29522]) +[2023-10-08 01:56:12,298][root][INFO] - Epoch: 19: Step: 1501/1557, loss[v]=0.049728, lr=0.000000, acc@1[1]=245.5/256=0.958984375, acc@1[2]=251.0/256=0.98046875 +[2023-10-08 01:56:55,160][root][INFO] - rank=1; last iteration 1557 +[2023-10-08 01:56:55,161][root][INFO] - rank=1; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 01:56:55,161][root][INFO] - Finished iterating, iteration=1557, shard=1 +[2023-10-08 01:56:55,161][root][INFO] - rank=2; last iteration 1557 +[2023-10-08 01:56:55,161][root][INFO] - rank=2; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 01:56:55,161][root][INFO] - Finished iterating, iteration=1557, shard=2 +[2023-10-08 01:56:55,166][root][INFO] - rank=1; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 01:56:55,166][root][INFO] - Epoch finished on 1 +[2023-10-08 01:56:55,167][root][INFO] - rank=2; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 01:56:55,167][root][INFO] - Epoch finished on 2 +[2023-10-08 01:56:55,187][root][INFO] - rank=3; last iteration 1557 +[2023-10-08 01:56:55,187][root][INFO] - rank=0; last iteration 1557 +[2023-10-08 01:56:55,187][root][INFO] - rank=3; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 01:56:55,187][root][INFO] - rank=0; Multi set iteration finished: iteration per set: [1557] +[2023-10-08 01:56:55,187][root][INFO] - Finished iterating, iteration=1557, shard=3 +[2023-10-08 01:56:55,187][root][INFO] - Finished iterating, iteration=1557, shard=0 +[2023-10-08 01:56:55,195][root][INFO] - rank=0; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 01:56:55,195][root][INFO] - Epoch finished on 0 +[2023-10-08 01:56:55,196][root][INFO] - rank=3; Multi set iteration finished after next: iteration per set: [0] +[2023-10-08 01:56:55,196][root][INFO] - Epoch finished on 3 +[2023-10-08 01:57:06,855][root][INFO] - Saved checkpoint at ./vdr_19 +[2023-10-08 01:57:06,855][root][INFO] - Saved checkpoint at ./vdr_19 +[2023-10-08 01:57:06,855][root][INFO] - Saved checkpoint at ./vdr_19 +[2023-10-08 01:57:06,856][root][INFO] - Av Loss per epoch=0.055751 +[2023-10-08 01:57:06,856][root][INFO] - Av Loss per epoch=0.055751 +[2023-10-08 01:57:06,856][root][INFO] - epoch total (1) correct predictions=380280 +[2023-10-08 01:57:06,856][root][INFO] - Av Loss per epoch=0.055751 +[2023-10-08 01:57:06,856][root][INFO] - epoch total (2) correct predictions=392376 +[2023-10-08 01:57:06,856][root][INFO] - epoch total (1) correct predictions=380280 +[2023-10-08 01:57:06,857][root][INFO] - epoch total (1) correct predictions=380280 +[2023-10-08 01:57:06,857][root][INFO] - epoch total (2) correct predictions=392376 +[2023-10-08 01:57:06,857][root][INFO] - epoch total (2) correct predictions=392376 +[2023-10-08 01:57:06,858][root][INFO] - Saved checkpoint at ./vdr_19 +[2023-10-08 01:57:06,859][root][INFO] - Av Loss per epoch=0.055751 +[2023-10-08 01:57:06,860][root][INFO] - epoch total (1) correct predictions=380280 +[2023-10-08 01:57:06,860][root][INFO] - epoch total (2) correct predictions=392376 +[2023-10-08 01:57:06,860][root][INFO] - Training finished.