upload model

config.json

@@ -0,0 +1,79 @@
+{
+  "_name_or_path": "facebook/wav2vec2-xls-r-300m",
+  "activation_dropout": 0.0,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Wav2Vec2ForCTC"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "classifier_proj_size": 256,
+  "codevector_dim": 768,
+  "contrastive_logits_temperature": 0.1,
+  "conv_bias": true,
+  "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,
+  "diversity_loss_weight": 0.1,
+  "do_stable_layer_norm": true,
+  "eos_token_id": 2,
+  "feat_extract_activation": "gelu",
+  "feat_extract_dropout": 0.0,
+  "feat_extract_norm": "layer",
+  "feat_proj_dropout": 0.0,
+  "feat_quantizer_dropout": 0.0,
+  "final_dropout": 0.0,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.0,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "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": "wav2vec2",
+  "num_attention_heads": 16,
+  "num_codevector_groups": 2,
+  "num_codevectors_per_group": 320,
+  "num_conv_pos_embedding_groups": 16,
+  "num_conv_pos_embeddings": 128,
+  "num_feat_extract_layers": 7,
+  "num_hidden_layers": 24,
+  "num_negatives": 100,
+  "pad_token_id": 42,
+  "proj_codevector_dim": 768,
+  "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 sha256:9271faf34ade7a1825e5e33d03549794c719cdb5251dd93395e569063fb0ca36
+size 2490411537

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:ab8cbf0c4ca4d9ee905f6e1edd7f32858f67ee46826caa5bef77bd0e55edf602
+size 1262099953

rng_state.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1233c8e11c71db13e502510231ab6138b867a153025d3498e035c655424c9a27
+size 14567

scaler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:90e726ad10a5a831479ad093de807ff5795e97ca6ff62f3665abe81475c35c2e
+size 559

scheduler.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:bc9b08a31d09e16a1debad1cdf01fd7e172e3d8fa1dea22330415ddc352ec9e5
+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 +1,601 @@
+{
+  "best_metric": 0.1282939612865448,
+  "best_model_checkpoint": "XLSR-english-phoneme-v2/checkpoint-7500",
+  "epoch": 57.69230769230769,
+  "global_step": 7500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 5.4253,
+      "step": 100
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 2.4222,
+      "step": 200
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 2.2277,
+      "step": 300
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 2.1449,
+      "step": 400
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 0.00015,
+      "loss": 1.3868,
+      "step": 500
+    },
+    {
+      "epoch": 3.85,
+      "eval_cer": 0.25636523266022826,
+      "eval_loss": 0.6820988059043884,
+      "eval_runtime": 22.2857,
+      "eval_samples_per_second": 20.731,
+      "eval_steps_per_second": 2.603,
+      "step": 500
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 0.5653,
+      "step": 600
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 0.3812,
+      "step": 700
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 0.2902,
+      "step": 800
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 0.00027,
+      "loss": 0.2423,
+      "step": 900
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 0.0003,
+      "loss": 0.2164,
+      "step": 1000
+    },
+    {
+      "epoch": 7.69,
+      "eval_cer": 0.09801261074307606,
+      "eval_loss": 0.20735949277877808,
+      "eval_runtime": 20.162,
+      "eval_samples_per_second": 22.914,
+      "eval_steps_per_second": 2.877,
+      "step": 1000
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 0.00029558823529411763,
+      "loss": 0.2116,
+      "step": 1100
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 0.00029117647058823524,
+      "loss": 0.1857,
+      "step": 1200
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.00028676470588235296,
+      "loss": 0.1561,
+      "step": 1300
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 0.00028235294117647056,
+      "loss": 0.1368,
+      "step": 1400
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 0.0002779411764705882,
+      "loss": 0.1317,
+      "step": 1500
+    },
+    {
+      "epoch": 11.54,
+      "eval_cer": 0.07494612499002315,
+      "eval_loss": 0.16971056163311005,
+      "eval_runtime": 39.1088,
+      "eval_samples_per_second": 11.813,
+      "eval_steps_per_second": 1.483,
+      "step": 1500
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 0.00027352941176470583,
+      "loss": 0.1151,
+      "step": 1600
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 0.0002691176470588235,
+      "loss": 0.1042,
+      "step": 1700
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 0.00026470588235294115,
+      "loss": 0.1044,
+      "step": 1800
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 0.0002602941176470588,
+      "loss": 0.089,
+      "step": 1900
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 0.0002558823529411764,
+      "loss": 0.0856,
+      "step": 2000
+    },
+    {
+      "epoch": 15.38,
+      "eval_cer": 0.0616968632771969,
+      "eval_loss": 0.1550830602645874,
+      "eval_runtime": 20.1049,
+      "eval_samples_per_second": 22.979,
+      "eval_steps_per_second": 2.885,
+      "step": 2000
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 0.0002514705882352941,
+      "loss": 0.0821,
+      "step": 2100
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 0.00024705882352941174,
+      "loss": 0.0816,
+      "step": 2200
+    },
+    {
+      "epoch": 17.69,
+      "learning_rate": 0.0002426470588235294,
+      "loss": 0.0708,
+      "step": 2300
+    },
+    {
+      "epoch": 18.46,
+      "learning_rate": 0.000238235294117647,
+      "loss": 0.0693,
+      "step": 2400
+    },
+    {
+      "epoch": 19.23,
+      "learning_rate": 0.0002338235294117647,
+      "loss": 0.0651,
+      "step": 2500
+    },
+    {
+      "epoch": 19.23,
+      "eval_cer": 0.06121797429962487,
+      "eval_loss": 0.15284913778305054,
+      "eval_runtime": 29.2987,
+      "eval_samples_per_second": 15.769,
+      "eval_steps_per_second": 1.98,
+      "step": 2500
+    },
+    {
+      "epoch": 20.0,
+      "learning_rate": 0.0002294117647058823,
+      "loss": 0.0597,
+      "step": 2600
+    },
+    {
+      "epoch": 20.77,
+      "learning_rate": 0.000225,
+      "loss": 0.0558,
+      "step": 2700
+    },
+    {
+      "epoch": 21.54,
+      "learning_rate": 0.00022058823529411765,
+      "loss": 0.0543,
+      "step": 2800
+    },
+    {
+      "epoch": 22.31,
+      "learning_rate": 0.00021617647058823526,
+      "loss": 0.0527,
+      "step": 2900
+    },
+    {
+      "epoch": 23.08,
+      "learning_rate": 0.00021176470588235295,
+      "loss": 0.0502,
+      "step": 3000
+    },
+    {
+      "epoch": 23.08,
+      "eval_cer": 0.05475297310240242,
+      "eval_loss": 0.15015824139118195,
+      "eval_runtime": 28.2669,
+      "eval_samples_per_second": 16.344,
+      "eval_steps_per_second": 2.052,
+      "step": 3000
+    },
+    {
+      "epoch": 23.85,
+      "learning_rate": 0.00020735294117647056,
+      "loss": 0.0503,
+      "step": 3100
+    },
+    {
+      "epoch": 24.62,
+      "learning_rate": 0.00020294117647058822,
+      "loss": 0.0513,
+      "step": 3200
+    },
+    {
+      "epoch": 25.38,
+      "learning_rate": 0.00019852941176470585,
+      "loss": 0.0469,
+      "step": 3300
+    },
+    {
+      "epoch": 26.15,
+      "learning_rate": 0.0001941176470588235,
+      "loss": 0.0417,
+      "step": 3400
+    },
+    {
+      "epoch": 26.92,
+      "learning_rate": 0.00018970588235294115,
+      "loss": 0.0407,
+      "step": 3500
+    },
+    {
+      "epoch": 26.92,
+      "eval_cer": 0.06385186367627105,
+      "eval_loss": 0.13443736732006073,
+      "eval_runtime": 20.0653,
+      "eval_samples_per_second": 23.025,
+      "eval_steps_per_second": 2.891,
+      "step": 3500
+    },
+    {
+      "epoch": 27.69,
+      "learning_rate": 0.0001852941176470588,
+      "loss": 0.0432,
+      "step": 3600
+    },
+    {
+      "epoch": 28.46,
+      "learning_rate": 0.00018088235294117644,
+      "loss": 0.0379,
+      "step": 3700
+    },
+    {
+      "epoch": 29.23,
+      "learning_rate": 0.0001764705882352941,
+      "loss": 0.0432,
+      "step": 3800
+    },
+    {
+      "epoch": 30.0,
+      "learning_rate": 0.00017205882352941174,
+      "loss": 0.0394,
+      "step": 3900
+    },
+    {
+      "epoch": 30.77,
+      "learning_rate": 0.0001676470588235294,
+      "loss": 0.0374,
+      "step": 4000
+    },
+    {
+      "epoch": 30.77,
+      "eval_cer": 0.058903344241360046,
+      "eval_loss": 0.12998488545417786,
+      "eval_runtime": 16.186,
+      "eval_samples_per_second": 28.543,
+      "eval_steps_per_second": 3.583,
+      "step": 4000
+    },
+    {
+      "epoch": 31.54,
+      "learning_rate": 0.00016323529411764703,
+      "loss": 0.0328,
+      "step": 4100
+    },
+    {
+      "epoch": 32.31,
+      "learning_rate": 0.0001588235294117647,
+      "loss": 0.034,
+      "step": 4200
+    },
+    {
+      "epoch": 33.08,
+      "learning_rate": 0.00015441176470588233,
+      "loss": 0.0334,
+      "step": 4300
+    },
+    {
+      "epoch": 33.85,
+      "learning_rate": 0.00015,
+      "loss": 0.0329,
+      "step": 4400
+    },
+    {
+      "epoch": 34.62,
+      "learning_rate": 0.00014558823529411762,
+      "loss": 0.0285,
+      "step": 4500
+    },
+    {
+      "epoch": 34.62,
+      "eval_cer": 0.04541463803974778,
+      "eval_loss": 0.14117006957530975,
+      "eval_runtime": 25.9649,
+      "eval_samples_per_second": 17.793,
+      "eval_steps_per_second": 2.234,
+      "step": 4500
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 0.00014117647058823528,
+      "loss": 0.0266,
+      "step": 4600
+    },
+    {
+      "epoch": 36.15,
+      "learning_rate": 0.00013676470588235292,
+      "loss": 0.0257,
+      "step": 4700
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 0.00013235294117647058,
+      "loss": 0.0255,
+      "step": 4800
+    },
+    {
+      "epoch": 37.69,
+      "learning_rate": 0.0001279411764705882,
+      "loss": 0.0254,
+      "step": 4900
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 0.00012352941176470587,
+      "loss": 0.0212,
+      "step": 5000
+    },
+    {
+      "epoch": 38.46,
+      "eval_cer": 0.05012371298587277,
+      "eval_loss": 0.14315766096115112,
+      "eval_runtime": 26.1379,
+      "eval_samples_per_second": 17.675,
+      "eval_steps_per_second": 2.219,
+      "step": 5000
+    },
+    {
+      "epoch": 39.23,
+      "learning_rate": 0.0001191176470588235,
+      "loss": 0.0205,
+      "step": 5100
+    },
+    {
+      "epoch": 40.0,
+      "learning_rate": 0.00011470588235294115,
+      "loss": 0.0195,
+      "step": 5200
+    },
+    {
+      "epoch": 40.77,
+      "learning_rate": 0.00011029411764705883,
+      "loss": 0.0209,
+      "step": 5300
+    },
+    {
+      "epoch": 41.54,
+      "learning_rate": 0.00010588235294117647,
+      "loss": 0.0178,
+      "step": 5400
+    },
+    {
+      "epoch": 42.31,
+      "learning_rate": 0.00010147058823529411,
+      "loss": 0.016,
+      "step": 5500
+    },
+    {
+      "epoch": 42.31,
+      "eval_cer": 0.05722723281985793,
+      "eval_loss": 0.14901842176914215,
+      "eval_runtime": 26.3386,
+      "eval_samples_per_second": 17.541,
+      "eval_steps_per_second": 2.202,
+      "step": 5500
+    },
+    {
+      "epoch": 43.08,
+      "learning_rate": 9.705882352941176e-05,
+      "loss": 0.0177,
+      "step": 5600
+    },
+    {
+      "epoch": 43.85,
+      "learning_rate": 9.26470588235294e-05,
+      "loss": 0.0159,
+      "step": 5700
+    },
+    {
+      "epoch": 44.62,
+      "learning_rate": 8.823529411764705e-05,
+      "loss": 0.0181,
+      "step": 5800
+    },
+    {
+      "epoch": 45.38,
+      "learning_rate": 8.38235294117647e-05,
+      "loss": 0.0158,
+      "step": 5900
+    },
+    {
+      "epoch": 46.15,
+      "learning_rate": 7.941176470588235e-05,
+      "loss": 0.0153,
+      "step": 6000
+    },
+    {
+      "epoch": 46.15,
+      "eval_cer": 0.05068241679304015,
+      "eval_loss": 0.1339481920003891,
+      "eval_runtime": 26.3988,
+      "eval_samples_per_second": 17.501,
+      "eval_steps_per_second": 2.197,
+      "step": 6000
+    },
+    {
+      "epoch": 46.92,
+      "learning_rate": 7.5e-05,
+      "loss": 0.0142,
+      "step": 6100
+    },
+    {
+      "epoch": 47.69,
+      "learning_rate": 7.058823529411764e-05,
+      "loss": 0.0138,
+      "step": 6200
+    },
+    {
+      "epoch": 48.46,
+      "learning_rate": 6.617647058823529e-05,
+      "loss": 0.0122,
+      "step": 6300
+    },
+    {
+      "epoch": 49.23,
+      "learning_rate": 6.176470588235294e-05,
+      "loss": 0.0125,
+      "step": 6400
+    },
+    {
+      "epoch": 50.0,
+      "learning_rate": 5.7352941176470576e-05,
+      "loss": 0.0111,
+      "step": 6500
+    },
+    {
+      "epoch": 50.0,
+      "eval_cer": 0.05004389815627744,
+      "eval_loss": 0.13527342677116394,
+      "eval_runtime": 16.9816,
+      "eval_samples_per_second": 27.206,
+      "eval_steps_per_second": 3.415,
+      "step": 6500
+    },
+    {
+      "epoch": 50.77,
+      "learning_rate": 5.294117647058824e-05,
+      "loss": 0.0128,
+      "step": 6600
+    },
+    {
+      "epoch": 51.54,
+      "learning_rate": 4.85735294117647e-05,
+      "loss": 0.0098,
+      "step": 6700
+    },
+    {
+      "epoch": 52.31,
+      "learning_rate": 4.416176470588235e-05,
+      "loss": 0.0114,
+      "step": 6800
+    },
+    {
+      "epoch": 53.08,
+      "learning_rate": 3.975e-05,
+      "loss": 0.0105,
+      "step": 6900
+    },
+    {
+      "epoch": 53.85,
+      "learning_rate": 3.5338235294117645e-05,
+      "loss": 0.0103,
+      "step": 7000
+    },
+    {
+      "epoch": 53.85,
+      "eval_cer": 0.044616489743794396,
+      "eval_loss": 0.1340005248785019,
+      "eval_runtime": 16.089,
+      "eval_samples_per_second": 28.715,
+      "eval_steps_per_second": 3.605,
+      "step": 7000
+    },
+    {
+      "epoch": 54.62,
+      "learning_rate": 3.092647058823529e-05,
+      "loss": 0.0083,
+      "step": 7100
+    },
+    {
+      "epoch": 55.38,
+      "learning_rate": 2.651470588235294e-05,
+      "loss": 0.0085,
+      "step": 7200
+    },
+    {
+      "epoch": 56.15,
+      "learning_rate": 2.2102941176470584e-05,
+      "loss": 0.0082,
+      "step": 7300
+    },
+    {
+      "epoch": 56.92,
+      "learning_rate": 1.7691176470588234e-05,
+      "loss": 0.0074,
+      "step": 7400
+    },
+    {
+      "epoch": 57.69,
+      "learning_rate": 1.3279411764705882e-05,
+      "loss": 0.0072,
+      "step": 7500
+    },
+    {
+      "epoch": 57.69,
+      "eval_cer": 0.04333945247026898,
+      "eval_loss": 0.1282939612865448,
+      "eval_runtime": 16.8303,
+      "eval_samples_per_second": 27.45,
+      "eval_steps_per_second": 3.446,
+      "step": 7500
+    }
+  ],
+  "max_steps": 7800,
+  "num_train_epochs": 60,
+  "total_flos": 2.2858916955207467e+19,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:119c48d622f0b28ec6c8008946f1d38a4e79a06ca6263249509b157e124c7060
+size 2863

vocab.json

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