pretrained xls-r-1b with very weak youtube data

added_tokens.json

@@ -0,0 +1,4 @@
+{
+  "</s>": 126,
+  "<s>": 125
+}

all_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 0.02141648641123937,
+    "total_flos": 8.380982415051637e+18,
+    "train_loss": 2.956299072265625,
+    "train_runtime": 3006.0711,
+    "train_samples": 373544,
+    "train_samples_per_second": 2.661,
+    "train_steps_per_second": 0.665
+}

config.json

@@ -1,4 +1,5 @@
 {
+  "_name_or_path": "facebook/wav2vec2-xls-r-1b",
   "activation_dropout": 0.0,
   "adapter_attn_dim": null,
   "adapter_kernel_size": 3,
@@ -6,12 +7,12 @@
   "add_adapter": false,
   "apply_spec_augment": true,
   "architectures": [
-    "Wav2Vec2ForPreTraining"
+    "Wav2Vec2ForCTC"
   ],
   "attention_dropout": 0.0,
   "bos_token_id": 1,
   "classifier_proj_size": 256,
-  "codevector_dim": 256,
+  "codevector_dim": 1024,
   "contrastive_logits_temperature": 0.1,
   "conv_bias": true,
   "conv_dim": [
@@ -41,7 +42,7 @@
     2,
     2
   ],
-  "ctc_loss_reduction": "sum",
+  "ctc_loss_reduction": "mean",
   "ctc_zero_infinity": false,
   "diversity_loss_weight": 0.1,
   "do_stable_layer_norm": true,
@@ -52,12 +53,12 @@
   "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_dropout_prob": 0.0,
-  "hidden_size": 768,
+  "hidden_size": 1280,
   "initializer_range": 0.02,
-  "intermediate_size": 3072,
+  "intermediate_size": 5120,
   "layer_norm_eps": 1e-05,
   "layerdrop": 0.0,
   "mask_feature_length": 10,
@@ -65,20 +66,20 @@
   "mask_feature_prob": 0.0,
   "mask_time_length": 10,
   "mask_time_min_masks": 2,
-  "mask_time_prob": 0.65,
+  "mask_time_prob": 0.05,
   "model_type": "wav2vec2",
   "num_adapter_layers": 3,
-  "num_attention_heads": 12,
+  "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": 12,
+  "num_hidden_layers": 48,
   "num_negatives": 100,
-  "output_hidden_size": 768,
-  "pad_token_id": 0,
-  "proj_codevector_dim": 256,
+  "output_hidden_size": 1280,
+  "pad_token_id": 124,
+  "proj_codevector_dim": 1024,
   "tdnn_dilation": [
     1,
     2,
@@ -101,8 +102,8 @@
     1
   ],
   "torch_dtype": "float32",
-  "transformers_version": "4.38.2",
+  "transformers_version": "4.46.2",
   "use_weighted_layer_sum": false,
-  "vocab_size": 32,
+  "vocab_size": 127,
   "xvector_output_dim": 512
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1567f48173b19ed1f1e2c9c76fad74d2a6ed662c39128a00be62cca1dd2c9ba7
-size 380246024
+oid sha256:4c82c6976546d4645916faa4952bf332ce4fa1bc1d2eb49611e37db7455349c3
+size 3850741748

preprocessor_config.json

@@ -0,0 +1,10 @@
+{
+  "do_normalize": true,
+  "feature_extractor_type": "Wav2Vec2FeatureExtractor",
+  "feature_size": 1,
+  "padding_side": "right",
+  "padding_value": 0,
+  "processor_class": "Wav2Vec2Processor",
+  "return_attention_mask": true,
+  "sampling_rate": 16000
+}

special_tokens_map.json

@@ -0,0 +1,30 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "[PAD]",
+    "lstrip": true,
+    "normalized": false,
+    "rstrip": true,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "[UNK]",
+    "lstrip": true,
+    "normalized": false,
+    "rstrip": true,
+    "single_word": false
+  }
+}

tokenizer_config.json

@@ -0,0 +1,49 @@
+{
+  "added_tokens_decoder": {
+    "123": {
+      "content": "[UNK]",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "124": {
+      "content": "[PAD]",
+      "lstrip": true,
+      "normalized": false,
+      "rstrip": true,
+      "single_word": false,
+      "special": false
+    },
+    "125": {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "126": {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<s>",
+  "clean_up_tokenization_spaces": false,
+  "do_lower_case": false,
+  "eos_token": "</s>",
+  "ignore_mismatched_sizes": true,
+  "model_max_length": 1000000000000000019884624838656,
+  "pad_token": "[PAD]",
+  "processor_class": "Wav2Vec2Processor",
+  "replace_word_delimiter_char": " ",
+  "target_lang": null,
+  "tokenizer_class": "Wav2Vec2CTCTokenizer",
+  "unk_token": "[UNK]",
+  "word_delimiter_token": "|"
+}

train_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 0.02141648641123937,
+    "total_flos": 8.380982415051637e+18,
+    "train_loss": 2.956299072265625,
+    "train_runtime": 3006.0711,
+    "train_samples": 373544,
+    "train_samples_per_second": 2.661,
+    "train_steps_per_second": 0.665
+}

trainer_state.json

@@ -0,0 +1,70 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.02141648641123937,
+  "eval_steps": 500,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.005354121602809843,
+      "grad_norm": NaN,
+      "learning_rate": 2.5e-06,
+      "loss": 11.8252,
+      "step": 500
+    },
+    {
+      "epoch": 0.010708243205619685,
+      "grad_norm": NaN,
+      "learning_rate": 1.6666666666666669e-06,
+      "loss": 0.0,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01606236480842953,
+      "grad_norm": NaN,
+      "learning_rate": 8.333333333333334e-07,
+      "loss": 0.0,
+      "step": 1500
+    },
+    {
+      "epoch": 0.02141648641123937,
+      "grad_norm": NaN,
+      "learning_rate": 0.0,
+      "loss": 0.0,
+      "step": 2000
+    },
+    {
+      "epoch": 0.02141648641123937,
+      "step": 2000,
+      "total_flos": 8.380982415051637e+18,
+      "train_loss": 2.956299072265625,
+      "train_runtime": 3006.0711,
+      "train_samples_per_second": 2.661,
+      "train_steps_per_second": 0.665
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 2000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.380982415051637e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:38ea9119b51914be79a8bbc365ed90b3e1bd6aecd097c6f150056dbd45e48179
+size 5368

vocab.json

@@ -0,0 +1,127 @@
+{
+  "'": 1,
+  "+": 2,
+  "-": 3,
+  "0": 4,
+  "1": 5,
+  "2": 6,
+  "3": 7,
+  "4": 8,
+  "5": 9,
+  "6": 10,
+  "7": 11,
+  "8": 12,
+  "9": 13,
+  "[PAD]": 124,
+  "[UNK]": 123,
+  "a": 14,
+  "b": 15,
+  "c": 16,
+  "d": 17,
+  "e": 18,
+  "f": 19,
+  "g": 20,
+  "h": 21,
+  "i": 22,
+  "j": 23,
+  "k": 24,
+  "l": 25,
+  "m": 26,
+  "n": 27,
+  "o": 28,
+  "p": 29,
+  "q": 30,
+  "r": 31,
+  "s": 32,
+  "t": 33,
+  "u": 34,
+  "v": 35,
+  "w": 36,
+  "x": 37,
+  "y": 38,
+  "z": 39,
+  "|": 0,
+  "£": 40,
+  "à": 41,
+  "â": 42,
+  "ä": 43,
+  "ç": 44,
+  "ê": 45,
+  "ë": 46,
+  "î": 47,
+  "ï": 48,
+  "ò": 49,
+  "ô": 50,
+  "ø": 51,
+  "ù": 52,
+  "ú": 53,
+  "û": 54,
+  "ć": 55,
+  "ě": 56,
+  "ń": 57,
+  "ř": 58,
+  "ŝ": 59,
+  "š": 60,
+  "ŵ": 61,
+  "ŷ": 62,
+  "ż": 63,
+  "ƒ": 64,
+  "ƙ": 65,
+  "‵": 66,
+  "、": 67,
+  "。": 68,
+  "〇": 69,
+  "い": 70,
+  "え": 71,
+  "か": 72,
+  "こ": 73,
+  "す": 74,
+  "っ": 75,
+  "で": 76,
+  "や": 77,
+  "よ": 78,
+  "ろ": 79,
+  "わ": 80,
+  "ん": 81,
+  "゜": 82,
+  "ジ": 83,
+  "マ": 84,
+  "ー": 85,
+  "万": 86,
+  "不": 87,
+  "中": 88,
+  "了": 89,
+  "事": 90,
+  "他": 91,
+  "但": 92,
+  "你": 93,
+  "再": 94,
+  "啊": 95,
+  "嗯": 96,
+  "在": 97,
+  "妖": 98,
+  "娘": 99,
+  "已": 100,
+  "年": 101,
+  "幸": 102,
+  "愈": 103,
+  "掉": 104,
+  "排": 105,
+  "是": 106,
+  "现": 107,
+  "珠": 108,
+  "痊": 109,
+  "的": 110,
+  "看": 111,
+  "算": 112,
+  "米": 113,
+  "精": 114,
+  "经": 115,
+  "美": 116,
+  "见": 117,
+  "谢": 118,
+  "除": 119,
+  "食": 120,
+  "饭": 121,
+  "�": 122
+}