Upload folder using huggingface_hub

teaching_arithmetic/out/addition_plain/addition_plain/config.yaml

@@ -0,0 +1,80 @@
+add_space: false
+algo_reason: false
+always_save_checkpoint: false
+backend: nccl
+batch_size: 256
+beta1: 0.9
+beta2: 0.99
+bias: false
+binary: false
+block_size: 256
+ckpt_path_name: ckpt_10000.pt
+compile: true
+data_format: plain
+data_ratio: 0.2
+data_shuffle: true
+data_type: text
+dataset: bal
+decay_lr: true
+device: cuda:0
+dropout: 0.2
+dtype: bfloat16
+eps: 0
+eval_addition: true
+eval_addition_ar: false
+eval_addition_train: true
+eval_interval: 250
+eval_iters: 200
+eval_only: false
+eval_other: false
+eval_text: false
+eval_text_data_path: null
+exp_name: addition_plain
+grad_clip: 1.0
+gradient_accumulation_steps: 1
+init_from: scratch
+learning_rate: 0.001
+log_interval: 10
+lr_decay_iters: 5000
+max_iters: 5000
+meta_path_specified: true
+min_lr: null
+multi_digit: false
+n_embd: 384
+n_head: 6
+n_layer: 6
+num_digit: 5
+operator: +
+other_operator: +
+out_dir: out/addition_plain
+print_interval: 2
+random_A: false
+random_C: false
+result_dir: out/addition_plain/addition_plain
+resume_dir: null
+resume_iter: false
+reverse_ab: false
+reverse_c: false
+save_final: true
+simple: false
+start: FILE:data/bal/test_10000.txt
+start_ar: null
+start_other: null
+start_train: null
+test_batch_size: 128
+tokenizer: char
+train_both: false
+train_data_path: train_3digit_10000.txt
+train_data_path2: train_addition.bin
+use_flash: true
+use_lora: false
+val_data_path: val.bin
+val_data_path2: val_addition.bin
+vocabulary: all_ascii_chars
+wandb_entity: ssdd
+wandb_log: false
+wandb_project: addition
+wandb_run_name: addition_plain
+warmup_iters: 100
+weight_decay: 0.1
+zero_pad: false

teaching_arithmetic/out/addition_plain/addition_plain/result.csv

@@ -0,0 +1,22 @@
+iter,train_loss,val_loss,val_ppl,test_acc,train_acc,test_acc_ar,test_acc_other
+0,4.558201789855957,4.557050704956055,,0.0,0.0,,
+250,1.7286615371704102,1.7170720100402832,,0.0707070707070707,0.16999999999999998,,
+500,1.4398878812789917,1.4310863018035889,,5.434343434343434,5.96,,
+750,1.2331775426864624,1.2203372716903687,,73.54545454545455,67.83,,
+1000,1.1358768939971924,1.2272201776504517,,75.29292929292929,70.32000000000001,,
+1250,0.5539495348930359,1.7250535488128662,,77.78787878787878,73.18,,
+1500,0.1001376137137413,2.863405227661133,,78.0,74.55000000000001,,
+1750,0.07596367597579956,3.4369161128997803,,78.83838383838383,77.46,,
+2000,0.06951306015253067,3.724381685256958,,78.57575757575758,78.39,,
+2250,0.06603887677192688,3.890638828277588,,79.06060606060606,78.68,,
+2500,0.06290150433778763,3.9894931316375732,,79.94949494949495,82.75,,
+2750,0.06049436703324318,4.099140167236328,,80.77777777777779,81.78,,
+3000,0.05838387832045555,4.249746322631836,,80.53535353535354,83.32000000000001,,
+3250,0.057155631482601166,4.353604316711426,,81.17171717171718,83.25,,
+3500,0.055782441049814224,4.412276268005371,,81.0,83.96000000000001,,
+3750,0.054552655667066574,4.499052047729492,,81.4040404040404,84.98,,
+4000,0.053756218403577805,4.524080276489258,,80.96969696969697,85.02,,
+4250,0.05292206257581711,4.601434707641602,,81.5050505050505,84.89,,
+4500,0.052253320813179016,4.638243675231934,,81.73737373737374,85.17,,
+4750,0.0518411286175251,4.682107448577881,,81.43434343434343,85.61,,
+5000,0.051410287618637085,4.731619358062744,,81.51515151515152,86.03,,

teaching_arithmetic/out/addition_plain/ckpt_10000.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9b6e21682cc49d528b55dd0b7569ffedb9f5a259f20fb30faddc983c6cfedf05
+size 129138715

teaching_arithmetic/out/addition_plain/ckpt_10000_acc.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ac751f03da9a54a32944b82532d2530352c89f6e23e7f53c5933c4697b29e685
+size 129139359

teaching_arithmetic/out/addition_plain/ckpt_10000_final.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b02d470faa1df8f459c2b19670e716be7f44db1a4243b604585ce57648f1c94e
+size 129139681