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-100/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": 105
+  },
+  "decoder_start_token_id": 104,
+  "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": 102,
+  "tie_word_embeddings": false,
+  "torch_dtype": "float32",
+  "transformers_version": null,
+  "vocab_size": 105
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78c8b9093bdcddcc349e2aa455ac64b27749f198494462f754ff94ee718e6c82
+size 835639825

results.txt

@@ -0,0 +1,272 @@
+Iteration: 0 - Loss: 7.64247465133667 
+Pearsonr: -0.022832252193482262; P: 0.6256245903890425 
+Spearmanr: -0.019337310602360446; P: 0.6794622528741858 
+
+Iteration: 200 - Loss: 1.9452604055404663 
+Pearsonr: -0.11237811945936477; P: 0.016009877506808122 
+Spearmanr: -0.057375352926347725; P: 0.21986580879088866 
+
+Iteration: 400 - Loss: 1.8948649168014526 
+Pearsonr: 0.10532301784098028; P: 0.024033256638654615 
+Spearmanr: 0.054381810879920306; P: 0.24492074923032167 
+
+Iteration: 600 - Loss: 1.886710524559021 
+Pearsonr: 0.298482860855398; P: 6.722970406195263e-11 
+Spearmanr: 0.27747281170301186; P: 1.4702006148769022e-09 
+
+Iteration: 800 - Loss: 1.9026901721954346 
+Pearsonr: 0.39889409020509253; P: 5.863543225509238e-19 
+Spearmanr: 0.38485234294537074; P: 1.1848083921042916e-17 
+
+Iteration: 1000 - Loss: 1.7541149854660034 
+Pearsonr: 0.4606943489650503; P: 1.7021183019887828e-25 
+Spearmanr: 0.4536824252134171; P: 1.1013533096658058e-24 
+
+Iteration: 1200 - Loss: 1.8975552320480347 
+Pearsonr: 0.46950639593629384; P: 1.5307937616814354e-26 
+Spearmanr: 0.4588002209395024; P: 2.8307426078869717e-25 
+
+Iteration: 1400 - Loss: 1.737377643585205 
+Pearsonr: 0.4960508134320673; P: 6.981963922500576e-30 
+Spearmanr: 0.48755480085837244; P: 8.815557025634631e-29 
+
+Iteration: 1600 - Loss: 1.7466052770614624 
+Pearsonr: 0.526249596218831; P: 4.673467244682937e-34 
+Spearmanr: 0.518525427924469; P: 5.989530072091597e-33 
+
+Iteration: 1800 - Loss: 1.727482557296753 
+Pearsonr: 0.5549506010478117; P: 1.9674764563498308e-38 
+Spearmanr: 0.5463378390530835; P: 4.482578100632254e-37 
+
+Iteration: 2000 - Loss: 1.8628698587417603 
+Pearsonr: 0.567390324436307; P: 1.8275170456385139e-40 
+Spearmanr: 0.5586446111528397; P: 5.00515408929344e-39 
+
+Iteration: 2200 - Loss: 1.7338200807571411 
+Pearsonr: 0.5956509259469079; P: 2.0427048402932726e-45 
+Spearmanr: 0.5835008619715757; P: 3.151336891720383e-43 
+
+Iteration: 2400 - Loss: 1.7214481830596924 
+Pearsonr: 0.6033659198190288; P: 7.4405929386868705e-47 
+Spearmanr: 0.5899331239235943; P: 2.246736788050501e-44 
+
+Iteration: 2600 - Loss: 1.715291976928711 
+Pearsonr: 0.6023728463106371; P: 1.1454714253331585e-46 
+Spearmanr: 0.5875570608388564; P: 6.000920909525325e-44 
+
+Iteration: 2800 - Loss: 1.700711727142334 
+Pearsonr: 0.6124435483563566; P: 1.3423611416767872e-48 
+Spearmanr: 0.5958444977034845; P: 1.8818535907909678e-45 
+
+Iteration: 3000 - Loss: 1.720475196838379 
+Pearsonr: 0.6168267083363987; P: 1.8427562880980246e-49 
+Spearmanr: 0.6009410842165503; P: 2.1280434513074314e-46 
+
+Iteration: 3200 - Loss: 1.8195748329162598 
+Pearsonr: 0.6285916882083233; P: 7.622981924970647e-52 
+Spearmanr: 0.6130363954126917; P: 1.028051777596461e-48 
+
+Iteration: 3400 - Loss: 1.845523715019226 
+Pearsonr: 0.6181520071341202; P: 1.004680721475029e-49 
+Spearmanr: 0.6017337785531462; P: 1.5108411398540332e-46 
+
+Iteration: 3600 - Loss: 1.666135549545288 
+Pearsonr: 0.6293459545476832; P: 5.318910311544119e-52 
+Spearmanr: 0.6136903461474531; P: 7.654809863688238e-49 
+
+Iteration: 3800 - Loss: 1.7646381855010986 
+Pearsonr: 0.6253346624968467; P: 3.565687140897611e-51 
+Spearmanr: 0.6071120183652957; P: 1.4415260128356491e-47 
+
+Iteration: 4000 - Loss: 1.7896883487701416 
+Pearsonr: 0.6284047797261237; P: 8.332787561514094e-52 
+Spearmanr: 0.6068999893577607; P: 1.5827755588198346e-47 
+
+Iteration: 4200 - Loss: 1.638436198234558 
+Pearsonr: 0.6304416555344815; P: 3.147730864234877e-52 
+Spearmanr: 0.6143020678241492; P: 5.805677978450625e-49 
+
+Iteration: 4400 - Loss: 1.6917756795883179 
+Pearsonr: 0.6491768484912954; P: 2.8596833939469116e-56 
+Spearmanr: 0.6278062767422858; P: 1.1077052264033374e-51 
+
+Iteration: 4600 - Loss: 1.6799358129501343 
+Pearsonr: 0.6532440360829803; P: 3.474349589038109e-57 
+Spearmanr: 0.6307138326056619; P: 2.7622501148795776e-52 
+
+Iteration: 4800 - Loss: 1.6511712074279785 
+Pearsonr: 0.6619902748936626; P: 3.339002136714999e-59 
+Spearmanr: 0.6398330367320612; P: 3.2149503782318286e-54 
+
+Iteration: 5000 - Loss: 1.5664293766021729 
+Pearsonr: 0.6585215394558566; P: 2.1465440136957183e-58 
+Spearmanr: 0.635253235418367; P: 3.066425219427498e-53 
+
+Iteration: 5200 - Loss: 1.5850845575332642 
+Pearsonr: 0.6525704146506688; P: 4.937053007058303e-57 
+Spearmanr: 0.6329394743968731; P: 9.445028369023724e-53 
+
+Iteration: 5400 - Loss: 1.7106763124465942 
+Pearsonr: 0.6529056952281443; P: 4.145395743946272e-57 
+Spearmanr: 0.6335178041875955; P: 7.136307637650005e-53 
+
+Iteration: 5600 - Loss: 1.6405019760131836 
+Pearsonr: 0.6555631240766893; P: 1.0292929357241102e-57 
+Spearmanr: 0.6373958077919194; P: 1.0726976409821417e-53 
+
+Iteration: 5800 - Loss: 1.609066128730774 
+Pearsonr: 0.657009096765708; P: 4.794557154159165e-58 
+Spearmanr: 0.6360049000273957; P: 2.123309602394972e-53 
+
+Iteration: 6000 - Loss: 1.5846246480941772 
+Pearsonr: 0.6527277668774474; P: 4.548392652711548e-57 
+Spearmanr: 0.6320587239526526; P: 1.4457450670773346e-52 
+
+Iteration: 6200 - Loss: 1.8170907497406006 
+Pearsonr: 0.6636766950681652; P: 1.3390939628607435e-59 
+Spearmanr: 0.6422044918627801; P: 9.849805604518302e-55 
+
+Iteration: 6400 - Loss: 1.6250044107437134 
+Pearsonr: 0.6632574308440573; P: 1.6815238703048483e-59 
+Spearmanr: 0.6406181563497215; P: 2.175651142869792e-54 
+
+Iteration: 6600 - Loss: 1.6043870449066162 
+Pearsonr: 0.6635132816903551; P: 1.4634384641452328e-59 
+Spearmanr: 0.6416320956018906; P: 1.3117336771830456e-54 
+
+Iteration: 6800 - Loss: 1.5713081359863281 
+Pearsonr: 0.6681877645721368; P: 1.1286519365694093e-60 
+Spearmanr: 0.6434853137121085; P: 5.176785579471148e-55 
+
+Iteration: 7000 - Loss: 1.5540624856948853 
+Pearsonr: 0.6623739050120263; P: 2.7137963703657558e-59 
+Spearmanr: 0.639388021994603; P: 4.009383961771031e-54 
+
+Iteration: 7200 - Loss: 1.638260006904602 
+Pearsonr: 0.6642838911113123; P: 9.62292618434872e-60 
+Spearmanr: 0.6384902915554355; P: 6.252631154635335e-54 
+
+Iteration: 7400 - Loss: 1.5665971040725708 
+Pearsonr: 0.6657470847981182; P: 4.326225658268859e-60 
+Spearmanr: 0.6374454222009467; P: 1.0468188628165724e-53 
+
+Iteration: 7600 - Loss: 1.6532906293869019 
+Pearsonr: 0.6689316496652999; P: 7.474760313003648e-61 
+Spearmanr: 0.6405297846288078; P: 2.2735377968048954e-54 
+
+Iteration: 7800 - Loss: 1.5612772703170776 
+Pearsonr: 0.670306515218574; P: 3.479070983665997e-61 
+Spearmanr: 0.6410840646869101; P: 1.7247228219381281e-54 
+
+Iteration: 8000 - Loss: 1.45542311668396 
+Pearsonr: 0.6724695301933015; P: 1.0358583042638763e-61 
+Spearmanr: 0.6462713584592281; P: 1.2639846897263992e-55 
+
+Iteration: 8200 - Loss: 1.5142748355865479 
+Pearsonr: 0.6759431727598586; P: 1.4484474669684866e-62 
+Spearmanr: 0.6467209180280477; P: 1.0053960123789051e-55 
+
+Iteration: 8400 - Loss: 1.5233200788497925 
+Pearsonr: 0.6743260258565581; P: 3.631794802028464e-62 
+Spearmanr: 0.6449811313338882; P: 2.4327781130894996e-55 
+
+Iteration: 8600 - Loss: 1.5494468212127686 
+Pearsonr: 0.6753893944863307; P: 1.9856326471096438e-62 
+Spearmanr: 0.6475875921195798; P: 6.459929598980597e-56 
+
+Iteration: 8800 - Loss: 1.6037766933441162 
+Pearsonr: 0.6742034039319683; P: 3.8930385313870035e-62 
+Spearmanr: 0.6477106181796233; P: 6.066060631500135e-56 
+
+Iteration: 9000 - Loss: 1.485519528388977 
+Pearsonr: 0.6775934467315273; P: 5.634631467045052e-63 
+Spearmanr: 0.6510482287105599; P: 1.0885667818300776e-56 
+
+Iteration: 9200 - Loss: 1.6029036045074463 
+Pearsonr: 0.6781739218859939; P: 4.0364361863570356e-63 
+Spearmanr: 0.6496008892899672; P: 2.2989534767425237e-56 
+
+Iteration: 9400 - Loss: 1.4569487571716309 
+Pearsonr: 0.6795236081267557; P: 1.852960969020503e-63 
+Spearmanr: 0.6508652992482686; P: 1.1967098656287701e-56 
+
+Iteration: 9600 - Loss: 1.6269348859786987 
+Pearsonr: 0.6836030329903168; P: 1.7168129886483357e-64 
+Spearmanr: 0.6527498261967531; P: 4.496391134616596e-57 
+
+Iteration: 9800 - Loss: 1.4487754106521606 
+Pearsonr: 0.6764751548899254; P: 1.069093721990669e-62 
+Spearmanr: 0.6477517110298481; P: 5.9398876111037285e-56 
+
+Iteration: 10000 - Loss: 1.5608919858932495 
+Pearsonr: 0.6766432136358463; P: 9.711661734535054e-63 
+Spearmanr: 0.6472669921406364; P: 7.609668942872514e-56 
+
+Iteration: 10200 - Loss: 1.478018045425415 
+Pearsonr: 0.6730422288106565; P: 7.503078366421406e-62 
+Spearmanr: 0.6448581683965026; P: 2.5889959027845858e-55 
+
+Iteration: 10400 - Loss: 1.5591259002685547 
+Pearsonr: 0.6746474122523497; P: 3.0268067963771996e-62 
+Spearmanr: 0.6476752694912578; P: 6.176722752175139e-56 
+
+Iteration: 10600 - Loss: 1.5131245851516724 
+Pearsonr: 0.6710759643674664; P: 2.2636137532580706e-61 
+Spearmanr: 0.6392393050128369; P: 4.3161026649060605e-54 
+
+Iteration: 10800 - Loss: 1.4237655401229858 
+Pearsonr: 0.676856863877655; P: 8.594342822528322e-63 
+Spearmanr: 0.6460934788095604; P: 1.3836540519524333e-55 
+
+Iteration: 11000 - Loss: 1.506270170211792 
+Pearsonr: 0.680294911809427; P: 1.18526929468267e-63 
+Spearmanr: 0.6467976120572693; P: 9.66856799197234e-56 
+
+Iteration: 11200 - Loss: 1.5831072330474854 
+Pearsonr: 0.6829126766047493; P: 2.574830998115189e-64 
+Spearmanr: 0.6496082746409292; P: 2.290223983831351e-56 
+
+Iteration: 11400 - Loss: 1.5992683172225952 
+Pearsonr: 0.6771045977995548; P: 7.457744719082818e-63 
+Spearmanr: 0.644957144723926; P: 2.4624989132061544e-55 
+
+Iteration: 11600 - Loss: 1.471413493156433 
+Pearsonr: 0.6790716578828968; P: 2.4059904825980525e-63 
+Spearmanr: 0.6459637617477907; P: 1.4779499665677124e-55 
+
+Iteration: 11800 - Loss: 1.4023113250732422 
+Pearsonr: 0.6801783804762525; P: 1.2681568404026062e-63 
+Spearmanr: 0.6483167219397755; P: 4.4475560434811035e-56 
+
+Iteration: 12000 - Loss: 1.6023857593536377 
+Pearsonr: 0.6811208669213497; P: 7.334240893405566e-64 
+Spearmanr: 0.6500899005199088; P: 1.7866146728494975e-56 
+
+Iteration: 12200 - Loss: 1.5053234100341797 
+Pearsonr: 0.6868447597326187; P: 2.5210672430655244e-65 
+Spearmanr: 0.6541173785778924; P: 2.200164691967787e-57 
+
+Iteration: 12400 - Loss: 1.4871832132339478 
+Pearsonr: 0.6865608449197619; P: 2.9852990799045636e-65 
+Spearmanr: 0.6556674185627186; P: 9.74246637038116e-58 
+
+Iteration: 12600 - Loss: 1.555140733718872 
+Pearsonr: 0.6814229321014913; P: 6.150994976764537e-64 
+Spearmanr: 0.6486873781863506; P: 3.677420611899671e-56 
+
+Iteration: 12800 - Loss: 1.574110507965088 
+Pearsonr: 0.6827753938463443; P: 2.790581729331374e-64 
+Spearmanr: 0.6500460933953986; P: 1.8274622709719802e-56 
+
+Iteration: 13000 - Loss: 1.412144422531128 
+Pearsonr: 0.6816951613447416; P: 5.248119456330189e-64 
+Spearmanr: 0.6494762851634788; P: 2.4513156330917483e-56 
+
+Iteration: 13200 - Loss: 1.5565046072006226 
+Pearsonr: 0.6837594418911727; P: 1.5659371257383546e-64 
+Spearmanr: 0.653652353957913; P: 2.8066481660268577e-57 
+
+Iteration: 13400 - Loss: 1.4073246717453003 
+Pearsonr: 0.682662134095367; P: 2.9820238318453588e-64 
+Spearmanr: 0.6514097321717548; P: 9.025746994527996e-57 
+