Upload lm-boosted decoder

added_tokens.json

@@ -0,0 +1 @@
+{"<s>": 41, "</s>": 42}

all_results.json

@@ -0,0 +1,14 @@
+{
+    "epoch": 149.98,
+    "eval_loss": 0.7148692607879639,
+    "eval_runtime": 14.0171,
+    "eval_samples": 336,
+    "eval_samples_per_second": 23.971,
+    "eval_steps_per_second": 2.996,
+    "eval_wer": 0.451,
+    "train_loss": 0.9359948108413003,
+    "train_runtime": 8030.991,
+    "train_samples": 717,
+    "train_samples_per_second": 13.392,
+    "train_steps_per_second": 0.411
+}

alphabet.json

@@ -0,0 +1 @@
+{"labels": [" ", "a", "c", "i", "k", "\u0430", "\u0431", "\u0433", "\u0434", "\u0435", "\u0436", "\u0437", "\u0438", "\u0439", "\u043a", "\u043b", "\u043c", "\u043d", "\u043e", "\u043f", "\u0440", "\u0441", "\u0442", "\u0443", "\u0445", "\u0448", "\u0449", "\u044b", "\u044e", "\u044f", "\u0456", "\u0493", "\u049b", "\u04a3", "\u04af", "\u04b1", "\u04bb", "\u04d9", "\u04e9", "\u2047", "", "<s>", "</s>"], "is_bpe": false}

config.json

@@ -0,0 +1,107 @@
+{
+  "_name_or_path": "facebook/wav2vec2-xls-r-300m",
+  "activation_dropout": 0.0,
+  "adapter_kernel_size": 3,
+  "adapter_stride": 2,
+  "add_adapter": false,
+  "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,
+  "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_min_masks": 0,
+  "mask_feature_prob": 0.0,
+  "mask_time_length": 10,
+  "mask_time_min_masks": 2,
+  "mask_time_prob": 0.25,
+  "model_type": "wav2vec2",
+  "num_adapter_layers": 3,
+  "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,
+  "output_hidden_size": 1024,
+  "pad_token_id": 40,
+  "proj_codevector_dim": 768,
+  "tdnn_dilation": [
+    1,
+    2,
+    3,
+    1,
+    1
+  ],
+  "tdnn_dim": [
+    512,
+    512,
+    512,
+    512,
+    1500
+  ],
+  "tdnn_kernel": [
+    5,
+    3,
+    3,
+    1,
+    1
+  ],
+  "torch_dtype": "float32",
+  "transformers_version": "4.17.0.dev0",
+  "use_weighted_layer_sum": false,
+  "vocab_size": 43,
+  "xvector_output_dim": 512
+}

eval_results.json

@@ -0,0 +1,9 @@
+{
+    "epoch": 149.98,
+    "eval_loss": 0.7148692607879639,
+    "eval_runtime": 14.0171,
+    "eval_samples": 336,
+    "eval_samples_per_second": 23.971,
+    "eval_steps_per_second": 2.996,
+    "eval_wer": 0.451
+}

language_model/5gram.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a1b4e76899c23765f8045a49b0fdd216ac89e687ffde64f3dc4fead2d84b337c
+size 26726842

language_model/attrs.json

@@ -0,0 +1 @@
+{"alpha": 0.5, "beta": 1.5, "unk_score_offset": -10.0, "score_boundary": true}

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": "Wav2Vec2ProcessorWithLM",
+  "return_attention_mask": true,
+  "sampling_rate": 16000
+}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:242eeb153035874f0ec5f10fee8fd7581625ea7916125d721673df269fa204ad
+size 1262099953

special_tokens_map.json

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

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": "|", "special_tokens_map_file": null, "tokenizer_file": null, "name_or_path": "DrishtiSharma/wav2vec2-xls-r-300m-kk-n2", "tokenizer_class": "Wav2Vec2CTCTokenizer", "processor_class": "Wav2Vec2ProcessorWithLM"}

train_results.json

@@ -0,0 +1,8 @@
+{
+    "epoch": 149.98,
+    "train_loss": 0.9359948108413003,
+    "train_runtime": 8030.991,
+    "train_samples": 717,
+    "train_samples_per_second": 13.392,
+    "train_steps_per_second": 0.411
+}

trainer_state.json

@@ -0,0 +1,265 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 149.9777777777778,
+  "global_step": 3300,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.3956000000000004e-05,
+      "loss": 9.6799,
+      "step": 200
+    },
+    {
+      "epoch": 9.09,
+      "eval_loss": 3.611884593963623,
+      "eval_runtime": 14.8755,
+      "eval_samples_per_second": 22.588,
+      "eval_steps_per_second": 2.823,
+      "eval_wer": 1.0,
+      "step": 200
+    },
+    {
+      "epoch": 18.18,
+      "learning_rate": 8.8356e-05,
+      "loss": 3.1332,
+      "step": 400
+    },
+    {
+      "epoch": 18.18,
+      "eval_loss": 2.535223960876465,
+      "eval_runtime": 14.5917,
+      "eval_samples_per_second": 23.027,
+      "eval_steps_per_second": 2.878,
+      "eval_wer": 1.005,
+      "step": 400
+    },
+    {
+      "epoch": 27.27,
+      "learning_rate": 0.000132756,
+      "loss": 1.0465,
+      "step": 600
+    },
+    {
+      "epoch": 27.27,
+      "eval_loss": 0.6169007420539856,
+      "eval_runtime": 14.7692,
+      "eval_samples_per_second": 22.75,
+      "eval_steps_per_second": 2.844,
+      "eval_wer": 0.682,
+      "step": 600
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 0.000177156,
+      "loss": 0.3452,
+      "step": 800
+    },
+    {
+      "epoch": 36.36,
+      "eval_loss": 0.6572489142417908,
+      "eval_runtime": 14.4275,
+      "eval_samples_per_second": 23.289,
+      "eval_steps_per_second": 2.911,
+      "eval_wer": 0.607,
+      "step": 800
+    },
+    {
+      "epoch": 45.44,
+      "learning_rate": 0.000221556,
+      "loss": 0.2575,
+      "step": 1000
+    },
+    {
+      "epoch": 45.44,
+      "eval_loss": 0.6526808142662048,
+      "eval_runtime": 14.9602,
+      "eval_samples_per_second": 22.46,
+      "eval_steps_per_second": 2.807,
+      "eval_wer": 0.578,
+      "step": 1000
+    },
+    {
+      "epoch": 54.53,
+      "learning_rate": 0.0002028886956521739,
+      "loss": 0.2088,
+      "step": 1200
+    },
+    {
+      "epoch": 54.53,
+      "eval_loss": 0.6828188300132751,
+      "eval_runtime": 14.4501,
+      "eval_samples_per_second": 23.252,
+      "eval_steps_per_second": 2.907,
+      "eval_wer": 0.551,
+      "step": 1200
+    },
+    {
+      "epoch": 63.62,
+      "learning_rate": 0.00018358434782608698,
+      "loss": 0.158,
+      "step": 1400
+    },
+    {
+      "epoch": 63.62,
+      "eval_loss": 0.707419753074646,
+      "eval_runtime": 14.2371,
+      "eval_samples_per_second": 23.6,
+      "eval_steps_per_second": 2.95,
+      "eval_wer": 0.5575,
+      "step": 1400
+    },
+    {
+      "epoch": 72.71,
+      "learning_rate": 0.00016428,
+      "loss": 0.1309,
+      "step": 1600
+    },
+    {
+      "epoch": 72.71,
+      "eval_loss": 0.6523335576057434,
+      "eval_runtime": 14.2381,
+      "eval_samples_per_second": 23.599,
+      "eval_steps_per_second": 2.95,
+      "eval_wer": 0.5595,
+      "step": 1600
+    },
+    {
+      "epoch": 81.8,
+      "learning_rate": 0.00014497565217391306,
+      "loss": 0.1074,
+      "step": 1800
+    },
+    {
+      "epoch": 81.8,
+      "eval_loss": 0.7262120842933655,
+      "eval_runtime": 14.2676,
+      "eval_samples_per_second": 23.55,
+      "eval_steps_per_second": 2.944,
+      "eval_wer": 0.5415,
+      "step": 1800
+    },
+    {
+      "epoch": 90.89,
+      "learning_rate": 0.0001256713043478261,
+      "loss": 0.087,
+      "step": 2000
+    },
+    {
+      "epoch": 90.89,
+      "eval_loss": 0.7198619246482849,
+      "eval_runtime": 14.0317,
+      "eval_samples_per_second": 23.946,
+      "eval_steps_per_second": 2.993,
+      "eval_wer": 0.521,
+      "step": 2000
+    },
+    {
+      "epoch": 99.98,
+      "learning_rate": 0.00010636695652173913,
+      "loss": 0.0711,
+      "step": 2200
+    },
+    {
+      "epoch": 99.98,
+      "eval_loss": 0.7113156318664551,
+      "eval_runtime": 14.2007,
+      "eval_samples_per_second": 23.661,
+      "eval_steps_per_second": 2.958,
+      "eval_wer": 0.523,
+      "step": 2200
+    },
+    {
+      "epoch": 109.09,
+      "learning_rate": 8.706260869565218e-05,
+      "loss": 0.0601,
+      "step": 2400
+    },
+    {
+      "epoch": 109.09,
+      "eval_loss": 0.6862908601760864,
+      "eval_runtime": 13.9453,
+      "eval_samples_per_second": 24.094,
+      "eval_steps_per_second": 3.012,
+      "eval_wer": 0.496,
+      "step": 2400
+    },
+    {
+      "epoch": 118.18,
+      "learning_rate": 6.775826086956522e-05,
+      "loss": 0.0451,
+      "step": 2600
+    },
+    {
+      "epoch": 118.18,
+      "eval_loss": 0.6998496055603027,
+      "eval_runtime": 14.078,
+      "eval_samples_per_second": 23.867,
+      "eval_steps_per_second": 2.983,
+      "eval_wer": 0.483,
+      "step": 2600
+    },
+    {
+      "epoch": 127.27,
+      "learning_rate": 4.8453913043478265e-05,
+      "loss": 0.0378,
+      "step": 2800
+    },
+    {
+      "epoch": 127.27,
+      "eval_loss": 0.6970936059951782,
+      "eval_runtime": 14.3534,
+      "eval_samples_per_second": 23.409,
+      "eval_steps_per_second": 2.926,
+      "eval_wer": 0.4615,
+      "step": 2800
+    },
+    {
+      "epoch": 136.36,
+      "learning_rate": 2.9149565217391308e-05,
+      "loss": 0.0319,
+      "step": 3000
+    },
+    {
+      "epoch": 136.36,
+      "eval_loss": 0.7119385600090027,
+      "eval_runtime": 14.2911,
+      "eval_samples_per_second": 23.511,
+      "eval_steps_per_second": 2.939,
+      "eval_wer": 0.4475,
+      "step": 3000
+    },
+    {
+      "epoch": 145.44,
+      "learning_rate": 9.845217391304348e-06,
+      "loss": 0.0305,
+      "step": 3200
+    },
+    {
+      "epoch": 145.44,
+      "eval_loss": 0.7181466221809387,
+      "eval_runtime": 14.0233,
+      "eval_samples_per_second": 23.96,
+      "eval_steps_per_second": 2.995,
+      "eval_wer": 0.459,
+      "step": 3200
+    },
+    {
+      "epoch": 149.98,
+      "step": 3300,
+      "total_flos": 1.7623169369275333e+19,
+      "train_loss": 0.9359948108413003,
+      "train_runtime": 8030.991,
+      "train_samples_per_second": 13.392,
+      "train_steps_per_second": 0.411
+    }
+  ],
+  "max_steps": 3300,
+  "num_train_epochs": 150,
+  "total_flos": 1.7623169369275333e+19,
+  "trial_name": null,
+  "trial_params": null
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:adb0105863dcce00203582f0ef97e1325605ebb083f887a9da4eb3e96f1c91bb
+size 2991

vocab.json

@@ -0,0 +1 @@
+{"a": 1, "c": 2, "i": 3, "k": 4, "а": 5, "б": 6, "г": 7, "д": 8, "е": 9, "ж": 10, "з": 11, "и": 12, "й": 13, "к": 14, "л": 15, "м": 16, "н": 17, "о": 18, "п": 19, "р": 20, "с": 21, "т": 22, "у": 23, "х": 24, "ш": 25, "щ": 26, "ы": 27, "ю": 28, "я": 29, "і": 30, "ғ": 31, "қ": 32, "ң": 33, "ү": 34, "ұ": 35, "һ": 36, "ә": 37, "ө": 38, "|": 0, "[UNK]": 39, "[PAD]": 40}