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-200/checkpoint-8000",
+    "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": 205
+  },
+  "decoder_start_token_id": 204,
+  "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": 202,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": null,
+  "vocab_size": 205
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:aaadaae4f3fdeb21c3592ad7d806368e25d6f4d521c17f9d642f70f8bf0cc96e
+size 835947409

results.txt

@@ -0,0 +1,272 @@
+Iteration: 0 - Loss: 8.08641242980957 
+Pearsonr: 0.010621631076654455; P: 0.8204651416584556 
+Spearmanr: 0.015827880196904395; P: 0.7352143065765318 
+
+Iteration: 200 - Loss: 2.071859121322632 
+Pearsonr: 0.2703880744597078; P: 3.9306588610586655e-09 
+Spearmanr: 0.24888998853600333; P: 6.553401979822967e-08 
+
+Iteration: 400 - Loss: 2.0651769638061523 
+Pearsonr: 0.42718881100365946; P: 8.742135113954878e-22 
+Spearmanr: 0.40966308232783616; P: 5.2958389742092694e-20 
+
+Iteration: 600 - Loss: 2.117659330368042 
+Pearsonr: 0.5136731670675092; P: 2.8767046353634493e-32 
+Spearmanr: 0.5028092748318626; P: 8.823308311390258e-31 
+
+Iteration: 800 - Loss: 1.945109486579895 
+Pearsonr: 0.5681436159158254; P: 1.3678123197278457e-40 
+Spearmanr: 0.5504079249050792; P: 1.0348476674787196e-37 
+
+Iteration: 1000 - Loss: 2.0464084148406982 
+Pearsonr: 0.5939622212503224; P: 4.167997420071383e-45 
+Spearmanr: 0.5752656275449446; P: 8.514397579285141e-42 
+
+Iteration: 1200 - Loss: 1.9914944171905518 
+Pearsonr: 0.6073071935285385; P: 1.3225910599584228e-47 
+Spearmanr: 0.5881375367999437; P: 4.723981004193902e-44 
+
+Iteration: 1400 - Loss: 1.8374221324920654 
+Pearsonr: 0.6075717386880887; P: 1.1767806305680267e-47 
+Spearmanr: 0.581478727629097; P: 7.141860148677098e-43 
+
+Iteration: 1600 - Loss: 1.849561095237732 
+Pearsonr: 0.6278860892896775; P: 1.0664801455384033e-51 
+Spearmanr: 0.6015524902799576; P: 1.6340943351419174e-46 
+
+Iteration: 1800 - Loss: 1.8758680820465088 
+Pearsonr: 0.6263555276974726; P: 2.2028800259636204e-51 
+Spearmanr: 0.5998857996235254; P: 3.352566746703723e-46 
+
+Iteration: 2000 - Loss: 1.9839155673980713 
+Pearsonr: 0.6351966735114957; P: 3.152286156971959e-53 
+Spearmanr: 0.6017145261425186; P: 1.5234799888366574e-46 
+
+Iteration: 2200 - Loss: 1.756374478340149 
+Pearsonr: 0.6440005255607844; P: 3.993071466828629e-55 
+Spearmanr: 0.6106112229001889; P: 3.0506744187327255e-48 
+
+Iteration: 2400 - Loss: 1.8533791303634644 
+Pearsonr: 0.6408746646700408; P: 1.9145926870945078e-54 
+Spearmanr: 0.6093127898293356; P: 5.440424225527253e-48 
+
+Iteration: 2600 - Loss: 1.8667470216751099 
+Pearsonr: 0.6464866322282338; P: 1.1328237298127798e-55 
+Spearmanr: 0.6103482539078129; P: 3.4306173880739657e-48 
+
+Iteration: 2800 - Loss: 1.7068326473236084 
+Pearsonr: 0.6427583141482234; P: 7.460948718495204e-55 
+Spearmanr: 0.6051805281594114; P: 3.369117421617711e-47 
+
+Iteration: 3000 - Loss: 1.8382970094680786 
+Pearsonr: 0.6552845954028798; P: 1.1919002735048877e-57 
+Spearmanr: 0.6156078862459645; P: 3.2109776184825534e-49 
+
+Iteration: 3200 - Loss: 1.6554993391036987 
+Pearsonr: 0.6600840629923029; P: 9.31205434232349e-59 
+Spearmanr: 0.6201462159734871; P: 4.0109450559252026e-50 
+
+Iteration: 3400 - Loss: 1.7735623121261597 
+Pearsonr: 0.6509680156510307; P: 1.1347375020798484e-56 
+Spearmanr: 0.6138350864017781; P: 7.170435438627515e-49 
+
+Iteration: 3600 - Loss: 1.7398338317871094 
+Pearsonr: 0.649293980761855; P: 2.6924645192144566e-56 
+Spearmanr: 0.6110556695337269; P: 2.5011015044019234e-48 
+
+Iteration: 3800 - Loss: 1.9043127298355103 
+Pearsonr: 0.650289529572843; P: 1.611654278409537e-56 
+Spearmanr: 0.6141181915219909; P: 6.3092199769872e-49 
+
+Iteration: 4000 - Loss: 1.7640738487243652 
+Pearsonr: 0.6582143682985201; P: 2.528049495306161e-58 
+Spearmanr: 0.6177296281424445; P: 1.2193376871567522e-49 
+
+Iteration: 4200 - Loss: 1.6442313194274902 
+Pearsonr: 0.6599942491461693; P: 9.771294549984983e-59 
+Spearmanr: 0.6220306797993137; P: 1.6740322003525815e-50 
+
+Iteration: 4400 - Loss: 1.9100476503372192 
+Pearsonr: 0.6690743597017518; P: 6.905655376873195e-61 
+Spearmanr: 0.629251659360487; P: 5.563991810672497e-52 
+
+Iteration: 4600 - Loss: 1.6869518756866455 
+Pearsonr: 0.6564827532494028; P: 6.334820310839045e-58 
+Spearmanr: 0.6176601300962116; P: 1.2587771742941603e-49 
+
+Iteration: 4800 - Loss: 1.7319031953811646 
+Pearsonr: 0.6681917870593195; P: 1.126143415925422e-60 
+Spearmanr: 0.6292608121458675; P: 5.539719960157941e-52 
+
+Iteration: 5000 - Loss: 1.8187370300292969 
+Pearsonr: 0.6594868772838571; P: 1.282098479701026e-58 
+Spearmanr: 0.6209705347616384; P: 2.738847992410262e-50 
+
+Iteration: 5200 - Loss: 1.7238787412643433 
+Pearsonr: 0.6610702661461342; P: 5.482831953281816e-59 
+Spearmanr: 0.6184022631839132; P: 8.956566723438575e-50 
+
+Iteration: 5400 - Loss: 1.641048789024353 
+Pearsonr: 0.6663983491770775; P: 3.026478036010308e-60 
+Spearmanr: 0.6259620220494749; P: 2.6527929100890802e-51 
+
+Iteration: 5600 - Loss: 1.6363853216171265 
+Pearsonr: 0.6675162716393186; P: 1.6355427976123489e-60 
+Spearmanr: 0.6260125201757113; P: 2.5903119517339975e-51 
+
+Iteration: 5800 - Loss: 1.8371890783309937 
+Pearsonr: 0.670109121477053; P: 3.883815499767754e-61 
+Spearmanr: 0.6294011338141468; P: 5.180495094469664e-52 
+
+Iteration: 6000 - Loss: 1.562391996383667 
+Pearsonr: 0.6747816745692912; P: 2.8047547374897617e-62 
+Spearmanr: 0.6322999156280892; P: 1.2868260493171434e-52 
+
+Iteration: 6200 - Loss: 1.7009657621383667 
+Pearsonr: 0.6722479587455972; P: 1.173286368040994e-61 
+Spearmanr: 0.6319249670407839; P: 1.542119089345593e-52 
+
+Iteration: 6400 - Loss: 1.6922433376312256 
+Pearsonr: 0.671785299890823; P: 1.5213361622013014e-61 
+Spearmanr: 0.628526316899759; P: 7.864138172221659e-52 
+
+Iteration: 6600 - Loss: 1.7369767427444458 
+Pearsonr: 0.6713740274119598; P: 1.915771920005438e-61 
+Spearmanr: 0.6305757833530631; P: 2.951522310848249e-52 
+
+Iteration: 6800 - Loss: 1.629564642906189 
+Pearsonr: 0.6751278806108644; P: 2.304035964492078e-62 
+Spearmanr: 0.6315196564550791; P: 1.8748176415696037e-52 
+
+Iteration: 7000 - Loss: 1.5638651847839355 
+Pearsonr: 0.6782024789726145; P: 3.9706587516969526e-63 
+Spearmanr: 0.6361126504042527; P: 2.0141684229571868e-53 
+
+Iteration: 7200 - Loss: 1.6536705493927002 
+Pearsonr: 0.675255590626147; P: 2.1426683850070987e-62 
+Spearmanr: 0.6364294630237282; P: 1.7244862252630863e-53 
+
+Iteration: 7400 - Loss: 1.6959208250045776 
+Pearsonr: 0.6751852309375689; P: 2.2301301939549945e-62 
+Spearmanr: 0.6330855402270119; P: 8.80000736230832e-53 
+
+Iteration: 7600 - Loss: 1.6595476865768433 
+Pearsonr: 0.6699256892172856; P: 4.301701913212347e-61 
+Spearmanr: 0.626572544395673; P: 1.988056637591422e-51 
+
+Iteration: 7800 - Loss: 1.6718226671218872 
+Pearsonr: 0.6713268786646085; P: 1.9670309997700185e-61 
+Spearmanr: 0.630905472994729; P: 2.5193129523466977e-52 
+
+Iteration: 8000 - Loss: 1.5495575666427612 
+Pearsonr: 0.6712791624950885; P: 2.020293445357269e-61 
+Spearmanr: 0.6315060219609953; P: 1.8871692963918654e-52 
+
+Iteration: 8200 - Loss: 1.6815775632858276 
+Pearsonr: 0.6720321140972193; P: 1.3245369461630289e-61 
+Spearmanr: 0.6311942591541434; P: 2.1927099874473988e-52 
+
+Iteration: 8400 - Loss: 1.6123058795928955 
+Pearsonr: 0.6698522367072551; P: 4.48129299869789e-61 
+Spearmanr: 0.6286222002169504; P: 7.512948637795918e-52 
+
+Iteration: 8600 - Loss: 1.8655686378479004 
+Pearsonr: 0.670825239609283; P: 2.604279028771614e-61 
+Spearmanr: 0.6262207618237786; P: 2.347717740511714e-51 
+
+Iteration: 8800 - Loss: 1.741979956626892 
+Pearsonr: 0.6676089980078708; P: 1.553962137700895e-60 
+Spearmanr: 0.6240043359406054; P: 6.660935693914321e-51 
+
+Iteration: 9000 - Loss: 1.7861990928649902 
+Pearsonr: 0.6753597289101979; P: 2.0194320814037133e-62 
+Spearmanr: 0.6300015565350976; P: 3.8870088500443724e-52 
+
+Iteration: 9200 - Loss: 1.5549899339675903 
+Pearsonr: 0.6710648938401299; P: 2.277675850036259e-61 
+Spearmanr: 0.6266997365511309; P: 1.87195244096068e-51 
+
+Iteration: 9400 - Loss: 1.7516299486160278 
+Pearsonr: 0.6760956778192262; P: 1.3277694056867854e-62 
+Spearmanr: 0.6331347759000925; P: 8.59259186816818e-53 
+
+Iteration: 9600 - Loss: 1.5417461395263672 
+Pearsonr: 0.6724719018549866; P: 1.034477338038094e-61 
+Spearmanr: 0.6300778087057145; P: 3.747591501375187e-52 
+
+Iteration: 9800 - Loss: 1.495241403579712 
+Pearsonr: 0.6721989467238506; P: 1.2060505255107816e-61 
+Spearmanr: 0.6290805338352035; P: 6.0377258365565734e-52 
+
+Iteration: 10000 - Loss: 1.618151068687439 
+Pearsonr: 0.6803738029250779; P: 1.1322321140368235e-63 
+Spearmanr: 0.64053344574296; P: 2.269397030848794e-54 
+
+Iteration: 10200 - Loss: 1.5292233228683472 
+Pearsonr: 0.6791221456257276; P: 2.3368612689757928e-63 
+Spearmanr: 0.638255854004383; P: 7.020276228531535e-54 
+
+Iteration: 10400 - Loss: 1.4918687343597412 
+Pearsonr: 0.6789584266908542; P: 2.5685025993016526e-63 
+Spearmanr: 0.6378274405259249; P: 8.672502626901967e-54 
+
+Iteration: 10600 - Loss: 1.5673266649246216 
+Pearsonr: 0.6726076017661807; P: 9.584335148305353e-62 
+Spearmanr: 0.6302953293844776; P: 3.376568218141803e-52 
+
+Iteration: 10800 - Loss: 1.5710985660552979 
+Pearsonr: 0.6732054851372804; P: 6.843137036839127e-62 
+Spearmanr: 0.6321384478694484; P: 1.3911711318475115e-52 
+
+Iteration: 11000 - Loss: 1.565316081047058 
+Pearsonr: 0.6794492351862846; P: 1.9343985492085727e-63 
+Spearmanr: 0.6376315709187856; P: 9.551294642100249e-54 
+
+Iteration: 11200 - Loss: 1.6766408681869507 
+Pearsonr: 0.6742901790513357; P: 3.7063107434104586e-62 
+Spearmanr: 0.6357503263485065; P: 2.405042033044568e-53 
+
+Iteration: 11400 - Loss: 1.498397707939148 
+Pearsonr: 0.6808975685038877; P: 8.351818097958045e-64 
+Spearmanr: 0.640862819771337; P: 1.925932244187251e-54 
+
+Iteration: 11600 - Loss: 1.442487359046936 
+Pearsonr: 0.6795766280238125; P: 1.7969925970135028e-63 
+Spearmanr: 0.6410959948692334; P: 1.714486655770838e-54 
+
+Iteration: 11800 - Loss: 1.6892503499984741 
+Pearsonr: 0.6753984556160304; P: 1.9754213061437024e-62 
+Spearmanr: 0.6401089458692852; P: 2.803073942635199e-54 
+
+Iteration: 12000 - Loss: 1.5381754636764526 
+Pearsonr: 0.6784983796326016; P: 3.3487176362749904e-63 
+Spearmanr: 0.6388817782790831; P: 5.1520951013619294e-54 
+
+Iteration: 12200 - Loss: 1.6090995073318481 
+Pearsonr: 0.6792403600433304; P: 2.182619170349716e-63 
+Spearmanr: 0.6419734629352486; P: 1.1058006230634839e-54 
+
+Iteration: 12400 - Loss: 1.5205144882202148 
+Pearsonr: 0.6861358428000858; P: 3.843357429201634e-65 
+Spearmanr: 0.646575735915118; P: 1.082573845258996e-55 
+
+Iteration: 12600 - Loss: 1.6619857549667358 
+Pearsonr: 0.676975412804351; P: 8.030454018463582e-63 
+Spearmanr: 0.6407291891047838; P: 2.058573312431572e-54 
+
+Iteration: 12800 - Loss: 1.5637744665145874 
+Pearsonr: 0.6807443189447862; P: 9.13012966631097e-64 
+Spearmanr: 0.6429810267989803; P: 6.671330893304869e-55 
+
+Iteration: 13000 - Loss: 1.529252052307129 
+Pearsonr: 0.6831154067704784; P: 2.2861648655983025e-64 
+Spearmanr: 0.6476491367109304; P: 6.259819872232118e-56 
+
+Iteration: 13200 - Loss: 1.6095237731933594 
+Pearsonr: 0.6807567719868253; P: 9.064281891679384e-64 
+Spearmanr: 0.6409391981872695; P: 1.8539664928785416e-54 
+
+Iteration: 13400 - Loss: 1.3736294507980347 
+Pearsonr: 0.6829489934674531; P: 2.520582231208335e-64 
+Spearmanr: 0.6449012180491192; P: 2.533202460308662e-55 
+