upload model

config.json

@@ -0,0 +1,73 @@
+{
+  "_name_or_path": "facebook/hubert-base-ls960",
+  "activation_dropout": 0.1,
+  "apply_spec_augment": true,
+  "architectures": [
+    "HubertForCTC"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "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
+  ],
+  "ctc_loss_reduction": "mean",
+  "ctc_zero_infinity": false,
+  "do_stable_layer_norm": false,
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_dropout": 0.0,
+  "feat_extract_norm": "group",
+  "feat_proj_dropout": 0.0,
+  "final_dropout": 0.1,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.0,
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.0,
+  "mask_feature_length": 10,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_prob": 0.05,
+  "model_type": "hubert",
+  "num_attention_heads": 12,
+  "num_conv_pos_embedding_groups": 16,
+  "num_conv_pos_embeddings": 128,
+  "num_feat_extract_layers": 7,
+  "num_hidden_layers": 12,
+  "pad_token_id": 42,
+  "tokenizer_class": "Wav2Vec2CTCTokenizer",
+  "torch_dtype": "float32",
+  "transformers_version": "4.11.3",
+  "use_weighted_layer_sum": false,
+  "vocab_size": 43
+}

optimizer.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bd96f76c804dad5e562a9a7eaed4f72aeb3ce747a153a28636c9c27c327a269f
+size 721752849

preprocessor_config.json

@@ -0,0 +1,9 @@
+{
+  "do_normalize": true,
+  "feature_extractor_type": "Wav2Vec2FeatureExtractor",
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0.0,
+  "return_attention_mask": true,
+  "sampling_rate": 16000
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e90a0cf2d07736816cf64424b0fbf982331caafacd6b7d853226434080b976ff
+size 377702999

rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0c1d4db70ac7e94817857249df43d9d9b360b5975a0a719dfb90989a7a1a7187
+size 14503

scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0571021828e8451967c6fe0e826edd76fab89ef42c17dd0201160f74f0534f6e
+size 559

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:009d60cfe544495de99d20e1a4bcebd0a0d22527ec6618a0d78081ec93fa965e
+size 623

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "<s>", "eos_token": "</s>", "unk_token": "[UNK]", "pad_token": "[PAD]"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"unk_token": "[UNK]", "bos_token": "<s>", "eos_token": "</s>", "pad_token": "[PAD]", "do_lower_case": false, "word_delimiter_token": "|", "tokenizer_class": "Wav2Vec2CTCTokenizer"}

trainer_state.json

@@ -0,0 +1,289 @@
+{
+  "best_metric": 0.10831519216299057,
+  "best_model_checkpoint": "hubert-base-english-phoneme-v2/checkpoint-3500",
+  "epoch": 26.923076923076923,
+  "global_step": 3500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.97e-05,
+      "loss": 4.5359,
+      "step": 100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.97e-05,
+      "loss": 2.2643,
+      "step": 200
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.969999999999998e-05,
+      "loss": 2.2425,
+      "step": 300
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.0001197,
+      "loss": 2.2134,
+      "step": 400
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00014969999999999998,
+      "loss": 1.7074,
+      "step": 500
+    },
+    {
+      "epoch": 3.85,
+      "eval_cer": 0.709044908285895,
+      "eval_loss": 0.9500909447669983,
+      "eval_runtime": 24.8403,
+      "eval_samples_per_second": 18.599,
+      "eval_steps_per_second": 2.335,
+      "step": 500
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017969999999999998,
+      "loss": 0.7708,
+      "step": 600
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00020969999999999997,
+      "loss": 0.4614,
+      "step": 700
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.0002397,
+      "loss": 0.3393,
+      "step": 800
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.0002697,
+      "loss": 0.2852,
+      "step": 900
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.00029969999999999997,
+      "loss": 0.2549,
+      "step": 1000
+    },
+    {
+      "epoch": 7.69,
+      "eval_cer": 0.15725806451612903,
+      "eval_loss": 0.2167779505252838,
+      "eval_runtime": 24.1643,
+      "eval_samples_per_second": 19.119,
+      "eval_steps_per_second": 2.4,
+      "step": 1000
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 0.00028975862068965515,
+      "loss": 0.2373,
+      "step": 1100
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 0.00027941379310344826,
+      "loss": 0.2001,
+      "step": 1200
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.0002690689655172414,
+      "loss": 0.184,
+      "step": 1300
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 0.00025872413793103443,
+      "loss": 0.1479,
+      "step": 1400
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 0.00024837931034482754,
+      "loss": 0.1328,
+      "step": 1500
+    },
+    {
+      "epoch": 11.54,
+      "eval_cer": 0.13432953826691968,
+      "eval_loss": 0.1408727467060089,
+      "eval_runtime": 23.8901,
+      "eval_samples_per_second": 19.339,
+      "eval_steps_per_second": 2.428,
+      "step": 1500
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 0.00023803448275862068,
+      "loss": 0.1223,
+      "step": 1600
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 0.00022768965517241377,
+      "loss": 0.115,
+      "step": 1700
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 0.00021734482758620688,
+      "loss": 0.1048,
+      "step": 1800
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 0.0944,
+      "step": 1900
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 0.00019665517241379307,
+      "loss": 0.0867,
+      "step": 2000
+    },
+    {
+      "epoch": 15.38,
+      "eval_cer": 0.10468058191018342,
+      "eval_loss": 0.13811400532722473,
+      "eval_runtime": 23.9414,
+      "eval_samples_per_second": 19.297,
+      "eval_steps_per_second": 2.423,
+      "step": 2000
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 0.0001863103448275862,
+      "loss": 0.0738,
+      "step": 2100
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 0.00017596551724137927,
+      "loss": 0.0749,
+      "step": 2200
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 0.0001656206896551724,
+      "loss": 0.0642,
+      "step": 2300
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 0.00015527586206896552,
+      "loss": 0.0643,
+      "step": 2400
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 0.0001449310344827586,
+      "loss": 0.0579,
+      "step": 2500
+    },
+    {
+      "epoch": 19.23,
+      "eval_cer": 0.1101359898798229,
+      "eval_loss": 0.13352537155151367,
+      "eval_runtime": 24.2513,
+      "eval_samples_per_second": 19.051,
+      "eval_steps_per_second": 2.392,
+      "step": 2500
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 0.00013458620689655172,
+      "loss": 0.0564,
+      "step": 2600
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 0.00012424137931034483,
+      "loss": 0.0478,
+      "step": 2700
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 0.00011389655172413792,
+      "loss": 0.044,
+      "step": 2800
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 0.00010355172413793101,
+      "loss": 0.0417,
+      "step": 2900
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 9.320689655172414e-05,
+      "loss": 0.0415,
+      "step": 3000
+    },
+    {
+      "epoch": 23.08,
+      "eval_cer": 0.09875079063883618,
+      "eval_loss": 0.11283650994300842,
+      "eval_runtime": 24.3063,
+      "eval_samples_per_second": 19.007,
+      "eval_steps_per_second": 2.386,
+      "step": 3000
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 8.286206896551724e-05,
+      "loss": 0.0411,
+      "step": 3100
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 7.251724137931034e-05,
+      "loss": 0.033,
+      "step": 3200
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 6.217241379310345e-05,
+      "loss": 0.0318,
+      "step": 3300
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 5.1827586206896546e-05,
+      "loss": 0.0282,
+      "step": 3400
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 4.148275862068966e-05,
+      "loss": 0.0274,
+      "step": 3500
+    },
+    {
+      "epoch": 26.92,
+      "eval_cer": 0.09930423782416192,
+      "eval_loss": 0.10831519216299057,
+      "eval_runtime": 24.5094,
+      "eval_samples_per_second": 18.85,
+      "eval_steps_per_second": 2.366,
+      "step": 3500
+    }
+  ],
+  "max_steps": 3900,
+  "num_train_epochs": 30,
+  "total_flos": 3.19030343120722e+18,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51cde43985a0b309ff16c2f2c073f43e57ffce5c1e19efd52038a40ce294424f
+size 2863

vocab.json

@@ -0,0 +1 @@
+{"ɑ": 0, "aj": 2, "m": 3, "f": 4, "æ": 5, "ə": 6, "z": 7, "p": 8, "ɪ": 9, "n": 10, "ʒ": 11, "k": 12, "h": 13, "j": 14, "ð": 15, "θ": 16, "ʊ": 17, "ow": 18, "ʌ": 19, "ɔ": 20, "ɹ̩": 21, "d": 22, "ŋ": 23, "i": 24, "ɹ": 25, "t": 26, "v": 27, "b": 28, "d͡ʒ": 29, "ɛ": 30, "oj": 31, "l": 32, "u": 33, "s": 34, "w": 35, "ɡ": 36, "ʃ": 37, "aw": 38, "t͡ʃ": 39, "ej": 40, "|": 1, "[UNK]": 41, "[PAD]": 42}