Upload 3 files

config.json

@@ -0,0 +1,218 @@
+{
+  "architectures": [
+    "SpeechEncoderDecoderModel"
+  ],
+  "decoder": {
+    "_name_or_path": "/gpfs/accounts/lingjzhu_root/lingjzhu1/lingjzhu/embeddings/models/bert-hubert-50/checkpoint-32000",
+    "add_cross_attention": true,
+    "architectures": [
+      "BertForMaskedLM"
+    ],
+    "attention_probs_dropout_prob": 0.1,
+    "bad_words_ids": null,
+    "bos_token_id": null,
+    "chunk_size_feed_forward": 0,
+    "classifier_dropout": null,
+    "cross_attention_hidden_size": null,
+    "decoder_start_token_id": null,
+    "diversity_penalty": 0.0,
+    "do_sample": false,
+    "early_stopping": false,
+    "encoder_no_repeat_ngram_size": 0,
+    "eos_token_id": null,
+    "exponential_decay_length_penalty": null,
+    "finetuning_task": null,
+    "forced_bos_token_id": null,
+    "forced_eos_token_id": null,
+    "gradient_checkpointing": false,
+    "hidden_act": "gelu",
+    "hidden_dropout_prob": 0.1,
+    "hidden_size": 768,
+    "id2label": {
+      "0": "LABEL_0",
+      "1": "LABEL_1"
+    },
+    "initializer_range": 0.02,
+    "intermediate_size": 3072,
+    "is_decoder": true,
+    "is_encoder_decoder": false,
+    "label2id": {
+      "LABEL_0": 0,
+      "LABEL_1": 1
+    },
+    "layer_norm_eps": 1e-12,
+    "length_penalty": 1.0,
+    "max_length": 20,
+    "max_position_embeddings": 512,
+    "min_length": 0,
+    "model_type": "bert",
+    "no_repeat_ngram_size": 0,
+    "num_attention_heads": 12,
+    "num_beam_groups": 1,
+    "num_beams": 1,
+    "num_hidden_layers": 12,
+    "num_return_sequences": 1,
+    "output_attentions": false,
+    "output_hidden_states": false,
+    "output_scores": false,
+    "pad_token_id": 0,
+    "position_embedding_type": "absolute",
+    "prefix": null,
+    "problem_type": null,
+    "pruned_heads": {},
+    "remove_invalid_values": false,
+    "repetition_penalty": 1.0,
+    "return_dict": true,
+    "return_dict_in_generate": false,
+    "sep_token_id": null,
+    "task_specific_params": null,
+    "temperature": 1.0,
+    "tie_encoder_decoder": false,
+    "tie_word_embeddings": true,
+    "tokenizer_class": null,
+    "top_k": 50,
+    "top_p": 1.0,
+    "torch_dtype": "float32",
+    "torchscript": false,
+    "transformers_version": "4.18.0",
+    "type_vocab_size": 2,
+    "typical_p": 1.0,
+    "use_bfloat16": false,
+    "use_cache": true,
+    "vocab_size": 55
+  },
+  "decoder_start_token_id": 54,
+  "encoder": {
+    "_name_or_path": "facebook/hubert-base-ls960",
+    "activation_dropout": 0.1,
+    "add_cross_attention": false,
+    "apply_spec_augment": true,
+    "architectures": [
+      "HubertModel"
+    ],
+    "attention_dropout": 0.1,
+    "bad_words_ids": null,
+    "bos_token_id": 1,
+    "chunk_size_feed_forward": 0,
+    "classifier_proj_size": 256,
+    "conv_bias": false,
+    "conv_dim": [
+      512,
+      512,
+      512,
+      512,
+      512,
+      512,
+      512
+    ],
+    "conv_kernel": [
+      10,
+      3,
+      3,
+      3,
+      3,
+      2,
+      2
+    ],
+    "conv_stride": [
+      5,
+      2,
+      2,
+      2,
+      2,
+      2,
+      2
+    ],
+    "cross_attention_hidden_size": null,
+    "ctc_loss_reduction": "sum",
+    "ctc_zero_infinity": false,
+    "decoder_start_token_id": null,
+    "diversity_penalty": 0.0,
+    "do_sample": false,
+    "do_stable_layer_norm": false,
+    "early_stopping": false,
+    "encoder_no_repeat_ngram_size": 0,
+    "eos_token_id": 2,
+    "exponential_decay_length_penalty": null,
+    "feat_extract_activation": "gelu",
+    "feat_extract_dropout": 0.0,
+    "feat_extract_norm": "group",
+    "feat_proj_dropout": 0.1,
+    "feat_proj_layer_norm": true,
+    "final_dropout": 0.1,
+    "finetuning_task": null,
+    "forced_bos_token_id": null,
+    "forced_eos_token_id": null,
+    "gradient_checkpointing": false,
+    "hidden_act": "gelu",
+    "hidden_dropout": 0.1,
+    "hidden_dropout_prob": 0.1,
+    "hidden_size": 768,
+    "id2label": {
+      "0": "LABEL_0",
+      "1": "LABEL_1"
+    },
+    "initializer_range": 0.02,
+    "intermediate_size": 3072,
+    "is_decoder": false,
+    "is_encoder_decoder": false,
+    "label2id": {
+      "LABEL_0": 0,
+      "LABEL_1": 1
+    },
+    "layer_norm_eps": 1e-05,
+    "layerdrop": 0.1,
+    "length_penalty": 1.0,
+    "mask_feature_length": 10,
+    "mask_feature_min_masks": 0,
+    "mask_feature_prob": 0.0,
+    "mask_time_length": 10,
+    "mask_time_min_masks": 2,
+    "mask_time_prob": 0.05,
+    "max_length": 20,
+    "min_length": 0,
+    "model_type": "hubert",
+    "no_repeat_ngram_size": 0,
+    "num_attention_heads": 12,
+    "num_beam_groups": 1,
+    "num_beams": 1,
+    "num_conv_pos_embedding_groups": 16,
+    "num_conv_pos_embeddings": 128,
+    "num_feat_extract_layers": 7,
+    "num_hidden_layers": 12,
+    "num_return_sequences": 1,
+    "output_attentions": false,
+    "output_hidden_states": false,
+    "output_scores": false,
+    "pad_token_id": 0,
+    "prefix": null,
+    "problem_type": null,
+    "pruned_heads": {},
+    "remove_invalid_values": false,
+    "repetition_penalty": 1.0,
+    "return_dict": true,
+    "return_dict_in_generate": false,
+    "sep_token_id": null,
+    "task_specific_params": null,
+    "temperature": 1.0,
+    "tie_encoder_decoder": false,
+    "tie_word_embeddings": true,
+    "tokenizer_class": "Wav2Vec2CTCTokenizer",
+    "top_k": 50,
+    "top_p": 1.0,
+    "torch_dtype": null,
+    "torchscript": false,
+    "transformers_version": "4.18.0",
+    "typical_p": 1.0,
+    "use_bfloat16": false,
+    "use_weighted_layer_sum": false,
+    "vocab_size": 32
+  },
+  "is_encoder_decoder": true,
+  "model_type": "speech-encoder-decoder",
+  "pad_token_id": 52,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": null,
+  "vocab_size": 55
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1318763d5c9fa693cc015da4b668f4c35287c637a8eb9b29a69fc947c042827c
+size 835486033

results.txt

@@ -0,0 +1,272 @@
+Iteration: 0 - Loss: 6.8719563484191895 
+Pearsonr: 0.02822506823041044; P: 0.5463934512704677 
+Spearmanr: 0.026839185990723664; P: 0.5662751931974506 
+
+Iteration: 200 - Loss: 1.831741213798523 
+Pearsonr: -0.09370338596178601; P: 0.04480606366430184 
+Spearmanr: -0.10795557662763665; P: 0.02070456274632216 
+
+Iteration: 400 - Loss: 1.7502282857894897 
+Pearsonr: -0.12576684371984334; P: 0.006979152523430303 
+Spearmanr: -0.12308542388293962; P: 0.008293741780421948 
+
+Iteration: 600 - Loss: 1.7469736337661743 
+Pearsonr: -0.15532910254058274; P: 0.0008407192821918212 
+Spearmanr: -0.16105387587432093; P: 0.0005328146798399329 
+
+Iteration: 800 - Loss: 1.6827481985092163 
+Pearsonr: -0.17282118408172523; P: 0.000198824226570517 
+Spearmanr: -0.14909205768399567; P: 0.0013581386866252311 
+
+Iteration: 1000 - Loss: 1.7101093530654907 
+Pearsonr: -0.15197685404988917; P: 0.0010903640114218705 
+Spearmanr: -0.10739373384036281; P: 0.021379247605653462 
+
+Iteration: 1200 - Loss: 1.7587130069732666 
+Pearsonr: 0.28936411623208685; P: 2.646598271042884e-10 
+Spearmanr: 0.27571907490147973; P: 1.8803594259465297e-09 
+
+Iteration: 1400 - Loss: 1.7680922746658325 
+Pearsonr: 0.35739628841621623; P: 2.8239402618744422e-15 
+Spearmanr: 0.3613504518722948; P: 1.3254574534859212e-15 
+
+Iteration: 1600 - Loss: 1.7529939413070679 
+Pearsonr: 0.45522121831213513; P: 7.337820650468306e-25 
+Spearmanr: 0.45364953830870564; P: 1.1109282914536672e-24 
+
+Iteration: 1800 - Loss: 1.7283083200454712 
+Pearsonr: 0.44402094991803337; P: 1.345747194942606e-23 
+Spearmanr: 0.45461626181284365; P: 8.610085340093419e-25 
+
+Iteration: 2000 - Loss: 1.7766599655151367 
+Pearsonr: 0.4673103469903363; P: 2.808386302997859e-26 
+Spearmanr: 0.47400817351419716; P: 4.3518934504043215e-27 
+
+Iteration: 2200 - Loss: 1.6536964178085327 
+Pearsonr: 0.4620748800536652; P: 1.1724868112658767e-25 
+Spearmanr: 0.4673898259669975; P: 2.7475908846015254e-26 
+
+Iteration: 2400 - Loss: 1.6404857635498047 
+Pearsonr: 0.4633044472374313; P: 8.40052965277035e-26 
+Spearmanr: 0.46868832216050865; P: 1.9200553428604305e-26 
+
+Iteration: 2600 - Loss: 1.6549359560012817 
+Pearsonr: 0.5011212318698759; P: 1.4855573182320391e-30 
+Spearmanr: 0.5031797417104643; P: 7.866956307638243e-31 
+
+Iteration: 2800 - Loss: 1.6123921871185303 
+Pearsonr: 0.5008548338656781; P: 1.6124312887558582e-30 
+Spearmanr: 0.5050961456011356; P: 4.3360163687960505e-31 
+
+Iteration: 3000 - Loss: 1.7403918504714966 
+Pearsonr: 0.5285856531842982; P: 2.1329234513567175e-34 
+Spearmanr: 0.5307277320255737; P: 1.0333825438755218e-34 
+
+Iteration: 3200 - Loss: 1.6316196918487549 
+Pearsonr: 0.5367563537401875; P: 1.3072854953179037e-35 
+Spearmanr: 0.54106394099427; P: 2.908259698683309e-36 
+
+Iteration: 3400 - Loss: 1.6573631763458252 
+Pearsonr: 0.5446861095889993; P: 8.080139921529646e-37 
+Spearmanr: 0.5491812622961394; P: 1.613142927251775e-37 
+
+Iteration: 3600 - Loss: 1.5965261459350586 
+Pearsonr: 0.5668971666587147; P: 2.2083444300492596e-40 
+Spearmanr: 0.5645931641781715; P: 5.324747818941762e-40 
+
+Iteration: 3800 - Loss: 1.6543303728103638 
+Pearsonr: 0.5553883652436633; P: 1.6743487701529246e-38 
+Spearmanr: 0.5573307761534844; P: 8.160386756301154e-39 
+
+Iteration: 4000 - Loss: 1.6208667755126953 
+Pearsonr: 0.5648862621584205; P: 4.762563074791771e-40 
+Spearmanr: 0.5630012107485691; P: 9.742181119097455e-40 
+
+Iteration: 4200 - Loss: 1.589337706565857 
+Pearsonr: 0.5735581885353966; P: 1.6670124993440217e-41 
+Spearmanr: 0.573256117734025; P: 1.876644869536107e-41 
+
+Iteration: 4400 - Loss: 1.5426558256149292 
+Pearsonr: 0.5656500278289712; P: 3.559088224298744e-40 
+Spearmanr: 0.5651935868991221; P: 4.236218247606656e-40 
+
+Iteration: 4600 - Loss: 1.5967150926589966 
+Pearsonr: 0.5805063592538471; P: 1.0563006313323093e-42 
+Spearmanr: 0.5796340310776816; P: 1.4989643627610055e-42 
+
+Iteration: 4800 - Loss: 1.6179089546203613 
+Pearsonr: 0.5738909908009685; P: 1.4628514856533864e-41 
+Spearmanr: 0.572972128896603; P: 2.0974777532746343e-41 
+
+Iteration: 5000 - Loss: 1.6256439685821533 
+Pearsonr: 0.5781692276548196; P: 2.6915840539003812e-42 
+Spearmanr: 0.5772358119400575; P: 3.902422855798749e-42 
+
+Iteration: 5200 - Loss: 1.5391310453414917 
+Pearsonr: 0.5816199450623889; P: 6.746493735758462e-43 
+Spearmanr: 0.5804478083819812; P: 1.0814449078904429e-42 
+
+Iteration: 5400 - Loss: 1.6562271118164062 
+Pearsonr: 0.5827991794835897; P: 4.1886265624163026e-43 
+Spearmanr: 0.5789342532933607; P: 1.9833432069495964e-42 
+
+Iteration: 5600 - Loss: 1.6008801460266113 
+Pearsonr: 0.5832895002519203; P: 3.4336054487220046e-43 
+Spearmanr: 0.5790409937076929; P: 1.9004988305762952e-42 
+
+Iteration: 5800 - Loss: 1.5070966482162476 
+Pearsonr: 0.5928269792902043; P: 6.716042014277357e-45 
+Spearmanr: 0.5874265863051931; P: 6.332060036437849e-44 
+
+Iteration: 6000 - Loss: 1.472196102142334 
+Pearsonr: 0.5951134354121741; P: 2.5643921587041375e-45 
+Spearmanr: 0.5872130423538708; P: 6.913548987836518e-44 
+
+Iteration: 6200 - Loss: 1.5007520914077759 
+Pearsonr: 0.5987526718275521; P: 5.451553645529932e-46 
+Spearmanr: 0.5871522552344138; P: 7.088548382398432e-44 
+
+Iteration: 6400 - Loss: 1.5846940279006958 
+Pearsonr: 0.5999422639000738; P: 3.2721563231899757e-46 
+Spearmanr: 0.5907427351324793; P: 1.6045876173576617e-44 
+
+Iteration: 6600 - Loss: 1.527746558189392 
+Pearsonr: 0.6033485008559338; P: 7.497213669470443e-47 
+Spearmanr: 0.5898760610409471; P: 2.3005956640995875e-44 
+
+Iteration: 6800 - Loss: 1.5243699550628662 
+Pearsonr: 0.6071802224877151; P: 1.398804600876004e-47 
+Spearmanr: 0.5930837651421408; P: 6.030050544468702e-45 
+
+Iteration: 7000 - Loss: 1.4590983390808105 
+Pearsonr: 0.6018006303294685; P: 1.4677563637101072e-46 
+Spearmanr: 0.5830826112410227; P: 3.7341450217285706e-43 
+
+Iteration: 7200 - Loss: 1.5712698698043823 
+Pearsonr: 0.6062190174244831; P: 2.1359814851275776e-47 
+Spearmanr: 0.5858399983015034; P: 1.2143998195906845e-43 
+
+Iteration: 7400 - Loss: 1.5214884281158447 
+Pearsonr: 0.6099004023741128; P: 4.188689746813027e-48 
+Spearmanr: 0.5920205902168914; P: 9.414211254324742e-45 
+
+Iteration: 7600 - Loss: 1.5258961915969849 
+Pearsonr: 0.6055007133395706; P: 2.9280054732260365e-47 
+Spearmanr: 0.5842234270353607; P: 2.349233529314388e-43 
+
+Iteration: 7800 - Loss: 1.5699869394302368 
+Pearsonr: 0.6093231622997569; P: 5.4153979089710764e-48 
+Spearmanr: 0.58477865393333; P: 1.8736301995904306e-43 
+
+Iteration: 8000 - Loss: 1.5184214115142822 
+Pearsonr: 0.605060782470973; P: 3.5505110155659953e-47 
+Spearmanr: 0.5789763561061104; P: 1.950246861072138e-42 
+
+Iteration: 8200 - Loss: 1.5991960763931274 
+Pearsonr: 0.6112574276498028; P: 2.2852097041251617e-48 
+Spearmanr: 0.5890162041964571; P: 3.2856410965658694e-44 
+
+Iteration: 8400 - Loss: 1.5256214141845703 
+Pearsonr: 0.6166074012573319; P: 2.0367621603375855e-49 
+Spearmanr: 0.5915340407706037; P: 1.1536676024411538e-44 
+
+Iteration: 8600 - Loss: 1.5412988662719727 
+Pearsonr: 0.617253839029917; P: 1.5160061382263392e-49 
+Spearmanr: 0.591598110268266; P: 1.1232132807652539e-44 
+
+Iteration: 8800 - Loss: 1.490647315979004 
+Pearsonr: 0.626078265183986; P: 2.5111495751614948e-51 
+Spearmanr: 0.6011685782752453; P: 1.9289908188639965e-46 
+
+Iteration: 9000 - Loss: 1.5037692785263062 
+Pearsonr: 0.6156840948372987; P: 3.1016210256479906e-49 
+Spearmanr: 0.5897977889452809; P: 2.376561004841454e-44 
+
+Iteration: 9200 - Loss: 1.4089856147766113 
+Pearsonr: 0.6146929210125671; P: 4.863962848935974e-49 
+Spearmanr: 0.5894476475624891; P: 2.74799184682832e-44 
+
+Iteration: 9400 - Loss: 1.5190479755401611 
+Pearsonr: 0.6279227335074822; P: 1.0480660773238495e-51 
+Spearmanr: 0.6001644861576926; P: 2.9738484904043734e-46 
+
+Iteration: 9600 - Loss: 1.5042378902435303 
+Pearsonr: 0.6321982775029772; P: 1.3515629592872948e-52 
+Spearmanr: 0.6067599201801124; P: 1.6835320827161711e-47 
+
+Iteration: 9800 - Loss: 1.4317461252212524 
+Pearsonr: 0.6295689616107754; P: 4.7810976831407934e-52 
+Spearmanr: 0.6009738448759464; P: 2.098169975631767e-46 
+
+Iteration: 10000 - Loss: 1.453094720840454 
+Pearsonr: 0.624768922206764; P: 4.652826148836808e-51 
+Spearmanr: 0.5968096431411777; P: 1.2491797183492255e-45 
+
+Iteration: 10200 - Loss: 1.4644638299942017 
+Pearsonr: 0.623839142537878; P: 7.196875072059396e-51 
+Spearmanr: 0.5975294939306774; P: 9.193912329551836e-46 
+
+Iteration: 10400 - Loss: 1.455410361289978 
+Pearsonr: 0.6235544830058308; P: 8.222633364556974e-51 
+Spearmanr: 0.6023223342144315; P: 1.1708398854462395e-46 
+
+Iteration: 10600 - Loss: 1.4695369005203247 
+Pearsonr: 0.6348209416475994; P: 3.7864436860874804e-53 
+Spearmanr: 0.6116882848101534; P: 1.884118101905883e-48 
+
+Iteration: 10800 - Loss: 1.5022165775299072 
+Pearsonr: 0.6281541414853896; P: 9.388538547537936e-52 
+Spearmanr: 0.6028106510951374; P: 9.472370858441417e-47 
+
+Iteration: 11000 - Loss: 1.4612548351287842 
+Pearsonr: 0.6229978607204307; P: 1.0665748630456236e-50 
+Spearmanr: 0.5995646946633196; P: 3.848544141558783e-46 
+
+Iteration: 11200 - Loss: 1.5569599866867065 
+Pearsonr: 0.6314183844965858; P: 1.968511548311075e-52 
+Spearmanr: 0.6067585946042987; P: 1.6845153805583686e-47 
+
+Iteration: 11400 - Loss: 1.3613736629486084 
+Pearsonr: 0.6408258017284847; P: 1.961802436251644e-54 
+Spearmanr: 0.6155540426188649; P: 3.2905389911833565e-49 
+
+Iteration: 11600 - Loss: 1.5064013004302979 
+Pearsonr: 0.6410379265662344; P: 1.764883697691992e-54 
+Spearmanr: 0.6150733635797523; P: 4.093279039521056e-49 
+
+Iteration: 11800 - Loss: 1.4196375608444214 
+Pearsonr: 0.6364924352181984; P: 1.672038217225987e-53 
+Spearmanr: 0.6118865530782891; P: 1.7238288213773374e-48 
+
+Iteration: 12000 - Loss: 1.53407883644104 
+Pearsonr: 0.636618748374165; P: 1.5715607736816904e-53 
+Spearmanr: 0.6104506388587572; P: 3.2774093494448963e-48 
+
+Iteration: 12200 - Loss: 1.4950836896896362 
+Pearsonr: 0.644366421928408; P: 3.3197046474191645e-55 
+Spearmanr: 0.6152795853027703; P: 3.7275052987909404e-49 
+
+Iteration: 12400 - Loss: 1.3674345016479492 
+Pearsonr: 0.6473769861777058; P: 7.193973392935627e-56 
+Spearmanr: 0.6210564446988981; P: 2.6319194480249493e-50 
+
+Iteration: 12600 - Loss: 1.4719256162643433 
+Pearsonr: 0.6463918659102053; P: 1.1888092558118035e-55 
+Spearmanr: 0.621427542804078; P: 2.2156523070358257e-50 
+
+Iteration: 12800 - Loss: 1.4978338479995728 
+Pearsonr: 0.6400583143580534; P: 2.874518683085562e-54 
+Spearmanr: 0.6165698755507679; P: 2.0719318443916184e-49 
+
+Iteration: 13000 - Loss: 1.4537323713302612 
+Pearsonr: 0.6433952124432938; P: 5.416969365141264e-55 
+Spearmanr: 0.6199290109080128; P: 4.434252845348923e-50 
+
+Iteration: 13200 - Loss: 1.4218829870224 
+Pearsonr: 0.6467281503382728; P: 1.0016975471460382e-55 
+Spearmanr: 0.6260784833531076; P: 2.510890919552011e-51 
+
+Iteration: 13400 - Loss: 1.4264096021652222 
+Pearsonr: 0.6471279084637483; P: 8.169592967434213e-56 
+Spearmanr: 0.6246663032529068; P: 4.882644880048028e-51 
+