BriereAssia/wav2vec2_finetuned_model_86

README.md

@@ -0,0 +1,102 @@
+---
+library_name: transformers
+license: mit
+base_model: facebook/w2v-bert-2.0
+tags:
+- generated_from_trainer
+datasets:
+- common_voice_11_0
+metrics:
+- wer
+- bleu
+model-index:
+- name: w2v3
+  results:
+  - task:
+      name: Automatic Speech Recognition
+      type: automatic-speech-recognition
+    dataset:
+      name: common_voice_11_0
+      type: common_voice_11_0
+      config: ar
+      split: test
+      args: ar
+    metrics:
+    - name: Wer
+      type: wer
+      value: 0.14389960749626846
+    - name: Bleu
+      type: bleu
+      value: 0.6267894360640803
+---
+
+<!-- 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. -->
+
+# w2v3
+
+This model is a fine-tuned version of [facebook/w2v-bert-2.0](https://huggingface.co/facebook/w2v-bert-2.0) on the common_voice_11_0 dataset.
+It achieves the following results on the evaluation set:
+- Loss: 0.1713
+- Wer: 0.1439
+- Cer: 0.0349
+- Bleu: 0.6268
+- Bert Score F1: 0.9719
+
+## 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: 1e-05
+- train_batch_size: 4
+- eval_batch_size: 8
+- seed: 42
+- optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
+- lr_scheduler_type: linear
+- training_steps: 5000
+
+### Training results
+
+| Training Loss | Epoch  | Step | Validation Loss | Wer    | Cer    | Bleu   | Bert Score F1 |
+|:-------------:|:------:|:----:|:---------------:|:------:|:------:|:------:|:-------------:|
+| 3.2221        | 0.0357 | 250  | 0.6698          | 0.5755 | 0.1572 | 0.1621 | 0.8585        |
+| 0.6124        | 0.0713 | 500  | 0.3894          | 0.2558 | 0.0702 | 0.4591 | 0.9370        |
+| 0.3391        | 0.1070 | 750  | 0.3347          | 0.2498 | 0.0672 | 0.4715 | 0.9382        |
+| 0.2579        | 0.1426 | 1000 | 0.2844          | 0.2047 | 0.0547 | 0.5279 | 0.9519        |
+| 0.2819        | 0.1783 | 1250 | 0.2671          | 0.2083 | 0.0538 | 0.5314 | 0.9543        |
+| 0.2042        | 0.2139 | 1500 | 0.2500          | 0.2029 | 0.0502 | 0.5389 | 0.9579        |
+| 0.2262        | 0.2496 | 1750 | 0.2324          | 0.1807 | 0.0466 | 0.5678 | 0.9611        |
+| 0.2001        | 0.2853 | 2000 | 0.2181          | 0.1777 | 0.0434 | 0.5733 | 0.9642        |
+| 0.1609        | 0.3209 | 2250 | 0.2095          | 0.1705 | 0.0430 | 0.5845 | 0.9648        |
+| 0.2138        | 0.3566 | 2500 | 0.2064          | 0.1561 | 0.0395 | 0.6074 | 0.9678        |
+| 0.1736        | 0.3922 | 2750 | 0.2018          | 0.1630 | 0.0398 | 0.5954 | 0.9675        |
+| 0.1622        | 0.4279 | 3000 | 0.2000          | 0.1602 | 0.0389 | 0.6004 | 0.9682        |
+| 0.1737        | 0.4636 | 3250 | 0.2017          | 0.1557 | 0.0383 | 0.6093 | 0.9685        |
+| 0.1525        | 0.4992 | 3500 | 0.1844          | 0.1460 | 0.0362 | 0.6230 | 0.9696        |
+| 0.1615        | 0.5349 | 3750 | 0.1947          | 0.1496 | 0.0365 | 0.6171 | 0.9704        |
+| 0.1558        | 0.5705 | 4000 | 0.1806          | 0.1488 | 0.0364 | 0.6184 | 0.9705        |
+| 0.125         | 0.6062 | 4250 | 0.1792          | 0.1468 | 0.0352 | 0.6234 | 0.9716        |
+| 0.1281        | 0.6418 | 4500 | 0.1785          | 0.1460 | 0.0353 | 0.6230 | 0.9717        |
+| 0.1321        | 0.6775 | 4750 | 0.1732          | 0.1453 | 0.0350 | 0.6245 | 0.9718        |
+| 0.1356        | 0.7132 | 5000 | 0.1713          | 0.1439 | 0.0349 | 0.6268 | 0.9719        |
+
+
+### Framework versions
+
+- Transformers 4.49.0
+- Pytorch 2.5.1+cu121
+- Datasets 3.3.1
+- Tokenizers 0.21.0

config.json

@@ -0,0 +1,82 @@
+{
+  "_name_or_path": "facebook/w2v-bert-2.0",
+  "activation_dropout": 0.0,
+  "adapter_act": "relu",
+  "adapter_kernel_size": 3,
+  "adapter_stride": 2,
+  "add_adapter": true,
+  "apply_spec_augment": false,
+  "architectures": [
+    "Wav2Vec2BertForCTC"
+  ],
+  "attention_dropout": 0.0,
+  "bos_token_id": 1,
+  "classifier_proj_size": 768,
+  "codevector_dim": 768,
+  "conformer_conv_dropout": 0.1,
+  "contrastive_logits_temperature": 0.1,
+  "conv_depthwise_kernel_size": 31,
+  "ctc_loss_reduction": "mean",
+  "ctc_zero_infinity": false,
+  "diversity_loss_weight": 0.1,
+  "eos_token_id": 2,
+  "feat_proj_dropout": 0.0,
+  "feat_quantizer_dropout": 0.0,
+  "feature_projection_input_dim": 160,
+  "final_dropout": 0.1,
+  "hidden_act": "swish",
+  "hidden_dropout": 0.0,
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 4096,
+  "layer_norm_eps": 1e-05,
+  "layerdrop": 0.0,
+  "left_max_position_embeddings": 64,
+  "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": 0.0,
+  "max_source_positions": 5000,
+  "model_type": "wav2vec2-bert",
+  "num_adapter_layers": 1,
+  "num_attention_heads": 16,
+  "num_codevector_groups": 2,
+  "num_codevectors_per_group": 320,
+  "num_hidden_layers": 24,
+  "num_negatives": 100,
+  "output_hidden_size": 1024,
+  "pad_token_id": 74,
+  "position_embeddings_type": "relative_key",
+  "proj_codevector_dim": 768,
+  "right_max_position_embeddings": 8,
+  "rotary_embedding_base": 10000,
+  "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": "4.49.0",
+  "use_intermediate_ffn_before_adapter": false,
+  "use_weighted_layer_sum": false,
+  "vocab_size": 77,
+  "xvector_output_dim": 512
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4990c79313d597fe147e31917b8a8e07cbf050f2832ba941c14f0a967dc4db54
+size 2423130260

preprocessor_config.json

@@ -0,0 +1,11 @@
+{
+  "feature_extractor_type": "SeamlessM4TFeatureExtractor",
+  "feature_size": 80,
+  "num_mel_bins": 80,
+  "padding_side": "right",
+  "padding_value": 1,
+  "processor_class": "Wav2Vec2BertProcessor",
+  "return_attention_mask": true,
+  "sampling_rate": 16000,
+  "stride": 2
+}

training_args.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:718ab1836dab9b9949b83db5b99ac96ddb6fabebe306d84a3e5d2ca2da015545
+size 5240