Upload 15 files

.gitignore

@@ -0,0 +1 @@
+checkpoint-*/

README.md

@@ -0,0 +1,66 @@
+---
+tags:
+- generated_from_trainer
+metrics:
+- wer
+model-index:
+- name: wav2vec2-large-asr-th-2
+  results: []
+---
+
+<!-- This model card has been generated automatically according to the information the Trainer had access to. You
+should probably proofread and complete it, then remove this comment. -->
+
+# wav2vec2-large-asr-th-2
+
+This model was trained from scratch on the None dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.2590
+- Wer: 0.4210
+- Cer: 0.1146
+
+## Model description
+
+More information needed
+
+## Intended uses & limitations
+
+More information needed
+
+## Training and evaluation data
+
+More information needed
+
+## Training procedure
+
+### Training hyperparameters
+
+The following hyperparameters were used during training:
+- learning_rate: 0.0002
+- train_batch_size: 16
+- eval_batch_size: 8
+- seed: 42
+- gradient_accumulation_steps: 3
+- total_train_batch_size: 48
+- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
+- lr_scheduler_type: linear
+- lr_scheduler_warmup_steps: 600
+- training_steps: 4000
+- mixed_precision_training: Native AMP
+
+### Training results
+
+| Training Loss | Epoch | Step | Validation Loss | Wer    | Cer    |
+|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
+| 0.6189        | 0.24  | 1000 | 0.3571          | 0.5060 | 0.1449 |
+| 0.5435        | 0.47  | 2000 | 0.3255          | 0.4712 | 0.1328 |
+| 0.5612        | 0.71  | 3000 | 0.2870          | 0.4464 | 0.1233 |
+| 0.6737        | 0.95  | 4000 | 0.2590          | 0.4210 | 0.1146 |
+
+
+### Framework versions
+
+- Transformers 4.27.3
+- Pytorch 1.13.1+cu116
+- Datasets 2.10.1
+- Tokenizers 0.13.2

added_tokens.json

@@ -0,0 +1,4 @@
+{
+  "</s>": 71,
+  "<s>": 70
+}

alphabet.json

@@ -0,0 +1 @@
+{"labels": ["\u0e11", "\u0e45", "\u0e01", "\u0e07", "\u0e12", "\u0e30", "\u0e4a", "\u0e49", "\u0e0c", "\u0e0b", "\u0e14", "\u0e2f", "\u0e43", "\u0e36", "\u0e0d", "\u0e48", "\u0e32", "\u0e24", "\u0e4b", "\u0e2d", "\u0e2c", "\u0e17", "\u0e42", "\u0e20", "\u0e22", "\u0e47", "\u0e25", "\u0e38", "\u0e40", "\u0e2e", "\u0e1d", "\u0e1b", "\u0e35", "\u0e1a", "\u0e10", "\u0e15", "\u0e16", "\u0e28", "\u0e1f", "\u0e13", "\u0e2b", "\u0e23", " ", "\u0e1e", "\u0e06", "\u0e31", "\u0e04", "\u0e27", "\u0e0f", "\u0e08", "\u0e41", "\u0e21", "\u0e0e", "\u0e09", "\u0e4c", "\u0e29", "\u0e33", "\u0e1c", "\u0e02", "\u0e44", "\u0e39", "\u0e37", "\u0e19", "\u0e0a", "\u0e34", "\u0e18", "\u0e03", "\u0e2a", "\u2047", "", "<s>", "</s>"], "is_bpe": false}

config.json

@@ -0,0 +1,116 @@
+{
+  "_name_or_path": "DylanonWic/wav2vec2-large-asr-th-2",
+  "activation_dropout": 0.0,
+  "adapter_kernel_size": 3,
+  "adapter_stride": 2,
+  "add_adapter": false,
+  "apply_spec_augment": true,
+  "architectures": [
+    "Wav2Vec2ForCTC"
+  ],
+  "attention_dropout": 0.08,
+  "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": 0.1,
+  "feat_quantizer_dropout": 0.0,
+  "final_dropout": 0.0,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout": 0.1,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.1,
+  "mask_channel_length": 10,
+  "mask_channel_min_space": 1,
+  "mask_channel_other": 0.0,
+  "mask_channel_prob": 0.0,
+  "mask_channel_selection": "static",
+  "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_min_space": 1,
+  "mask_time_other": 0.0,
+  "mask_time_prob": 0.1,
+  "mask_time_selection": "static",
+  "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": 69,
+  "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": "4.27.3",
+  "use_weighted_layer_sum": false,
+  "vocab_size": 72,
+  "xvector_output_dim": 512
+}

language_model/attrs.json

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

language_model/internet-crawled.3gram-002.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c42eeb703599fbfa67a1992abf89231027c54dbb8e093cf1c395ae954be6c3b6
+size 2515339944

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:f3db9d140ed973d933448e83b4a3d7c72b7683b7a077f52c258594763abf2139
+size 1262197037

special_tokens_map.json

@@ -0,0 +1,36 @@
+{
+  "additional_special_tokens": [
+    {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "<s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    },
+    {
+      "content": "</s>",
+      "lstrip": false,
+      "normalized": true,
+      "rstrip": false,
+      "single_word": false
+    }
+  ],
+  "bos_token": "<s>",
+  "eos_token": "</s>",
+  "pad_token": "[PAD]",
+  "unk_token": "[UNK]"
+}

tokenizer_config.json

@@ -0,0 +1,14 @@
+{
+  "bos_token": "<s>",
+  "do_lower_case": false,
+  "eos_token": "</s>",
+  "model_max_length": 1000000000000000019884624838656,
+  "num_proc": 2,
+  "pad_token": "[PAD]",
+  "processor_class": "Wav2Vec2ProcessorWithLM",
+  "replace_word_delimiter_char": " ",
+  "special_tokens_map_file": "/root/.cache/huggingface/hub/models--airesearch--wav2vec2-large-xlsr-53-th/snapshots/3155938c549b23eee16b1d4b55dcb161b7fe4bcf/special_tokens_map.json",
+  "tokenizer_class": "Wav2Vec2CTCTokenizer",
+  "unk_token": "[UNK]",
+  "word_delimiter_token": "|"
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ab1acd7b1adc89516ec5569b2e31e361d047913189de035bffe974d3c7e42c8a
+size 3643

vocab.json

@@ -0,0 +1,72 @@
+{
+  "[PAD]": 69,
+  "[UNK]": 68,
+  "|": 42,
+  "ก": 2,
+  "ข": 58,
+  "ฃ": 66,
+  "ค": 46,
+  "ฆ": 44,
+  "ง": 3,
+  "จ": 49,
+  "ฉ": 53,
+  "ช": 63,
+  "ซ": 9,
+  "ฌ": 8,
+  "ญ": 14,
+  "ฎ": 52,
+  "ฏ": 48,
+  "ฐ": 34,
+  "ฑ": 0,
+  "ฒ": 4,
+  "ณ": 39,
+  "ด": 10,
+  "ต": 35,
+  "ถ": 36,
+  "ท": 21,
+  "ธ": 65,
+  "น": 62,
+  "บ": 33,
+  "ป": 31,
+  "ผ": 57,
+  "ฝ": 30,
+  "พ": 43,
+  "ฟ": 38,
+  "ภ": 23,
+  "ม": 51,
+  "ย": 24,
+  "ร": 41,
+  "ฤ": 17,
+  "ล": 26,
+  "ว": 47,
+  "ศ": 37,
+  "ษ": 55,
+  "ส": 67,
+  "ห": 40,
+  "ฬ": 20,
+  "อ": 19,
+  "ฮ": 29,
+  "ฯ": 11,
+  "ะ": 5,
+  "ั": 45,
+  "า": 16,
+  "ำ": 56,
+  "ิ": 64,
+  "ี": 32,
+  "ึ": 13,
+  "ื": 61,
+  "ุ": 27,
+  "ู": 60,
+  "เ": 28,
+  "แ": 50,
+  "โ": 22,
+  "ใ": 12,
+  "ไ": 59,
+  "ๅ": 1,
+  "็": 25,
+  "่": 15,
+  "้": 7,
+  "๊": 6,
+  "๋": 18,
+  "์": 54
+}