Ir para o conteúdo
ModelVerificadoExterno
Baixar

Whisper Tanglish Voice Generalized 5000

This model is a fine-tuned version of Badri0510/whisper-tanglish-5e5-8000-experiment on the None dataset. It achieves the following results on the evaluation set: Loss: 0.0006 Wer: 0.2142

Fontehuggingface.co/Badri0510/whisper-tanglish-voice-generalized-5000/tree/689722eb364ae42bd32c1b8d85842b0795cd07d3
TipoText modelModelo baseBadri0510/whisper-tanglish-5e5-8000-experimentVersãoPublicador@Badri0510Cnota

Fonte do modelo

Tipo
Text model
Tamanho do arquivo
278 MB
Modelo base
Badri0510/whisper-tanglish-5e5-8000-experiment
Versão
vmodel
Atualizado
2 de out. de 2026
Publicador
@Badri0510
Downloads
22
Fonte
Hugging Face

Descrição da fonte

whisper-tanglish-voice-generalized-5000

This model is a fine-tuned version of Badri0510/whisper-tanglish-5e5-8000-experiment on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0006
  • Wer: 0.2142

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: 5e-05
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 2
  • total_train_batch_size: 16
  • optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
  • lr_scheduler_type: cosine
  • lr_scheduler_warmup_steps: 200
  • training_steps: 10000
  • mixed_precision_training: Native AMP

Training results

Training LossEpochStepValidation LossWer
0.07820.055000.04220.2990
0.11720.110000.04630.2849
0.09891.037215000.03020.2789
0.13721.087220000.02860.2632
0.06172.024425000.01810.2471
0.09902.074430000.01970.2850
0.04993.012435000.02300.2724
0.05363.062440000.01600.2547
0.04554.001345000.00990.2417
0.02644.051350000.00870.2301
0.03174.101355000.00500.2236
0.01275.038860000.00450.2206
0.01555.088865000.00370.2223
0.00706.027270000.00220.2174
0.00586.077275000.00180.2164
0.00237.013580000.00130.2148
0.00297.063585000.00060.2141
0.00278.002590000.00060.2141
0.00248.052595000.00060.2142
0.00988.1025100000.00060.2142

Framework versions

  • Transformers 5.0.0
  • Pytorch 2.10.0+cu128
  • Datasets 4.0.0
  • Tokenizers 0.22.2

Fontes

1 fonte
Hugging FacePrimárioVersão 689722eb364ae42bd32c1b8d85842b0795cd07d3
huggingface.co/Badri0510/whisper-tanglish-voice-generalized-5000/tree/689722eb364ae42bd32c1b8d85842b0795cd07d3

Verificado 2 de out.

Artefatos de modelo

1 artefato

last-checkpoint/model.safetensors

safetensors · 278 MB · SHA-256 f8f6f5f089f0…f33e · Hugging Face

Baixar

Trechos de fonte

3 trechos
Readme11 de jul.huggingface.co/Badri0510/whisper-tanglish-voice-generalized-5000

--- library_name: transformers base_model: Badri0510/whisper-tanglish-5e5-8000-experiment tags: - generated_from_trainer metrics: - wer model-index: - name: whisper-tanglish-voice-generalized-5000 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. --> # whisper-tanglish-voice-generalized-5000 This model is a fine-tuned version of [Badri0510/whisper-tanglish-5e5-8000-experiment](https://huggingface.co/Badri0510/whisper-tanglish-5e5-8000-experiment) on the None dataset. It achieves the following results on the evaluation set: - Loss: 0.0006 - Wer: 0.2142 ## 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: 5e-05 - train_batch_size: 8 - eval_batch_size: 8 - seed: 42 - gradient_accumulation_steps: 2 - total_train_batch_size: 16 - optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments - lr_scheduler_type: cosine - lr_scheduler_warmup_steps: 200 - training_steps: 10000 - mixed_precision_training: Native AMP ### Training results | Training Loss | Epoch | Step | Validation Loss | Wer | |:-------------:|:------:|:-----:|:---------------:|:------:| | 0.0782 | 0.05 | 500 | 0.0422 | 0.2990 | | 0.1172 | 0.1 | 1000 | 0.0463 | 0.2849 | | 0.0989 | 1.0372 | 1500 | 0.0302 | 0.2789 | | 0.1372 | 1.0872 | 2000 | 0.0286 | 0.2632 | | 0.0617 | 2.0244 | 2500 | 0.0181 | 0.2471 | | 0.0990 | 2.0744 | 3000 | 0.0197 | 0.2850 | | 0.0499 | 3.0124 | 3500 | 0.0230 | 0.2724 | | 0.0536 | 3.0624 | 4000 | 0.0160 | 0.2547 | | 0.0455 | 4.0013 | 4500 | 0.0099 | 0.2417 | | 0.0264 | 4.0513 | 5000 | 0.0087 | 0.2301 | | 0.0317 | 4.1013 | 5500 | 0.0050 | 0.2236 | | 0.0127 | 5.0388 | 6000 | 0.0045 | 0.2206 | | 0.0155 | 5.0888 | 6500 | 0.0037 | 0.2223 | | 0.0070 | 6.0272 | 7000 | 0.0022 | 0.2174 | | 0.0058 | 6.0772 | 7500 | 0.0018 | 0.2164 | | 0.0023 | 7.0135 | 8000 | 0.0013 | 0.2148 | | 0.0029 | 7.0635 | 8500 | 0.0006 | 0.2141 | | 0.0027 | 8.0025 | 9000 | 0.0006 | 0.2141 | | 0.0024 | 8.0525 | 9500 | 0.0006 | 0.2142 | | 0.0098 | 8.10...

Note2 de out.huggingface.co/Badri0510/whisper-tanglish-voice-generalized-5000

Source context: 22 downloads · 0 likes · Pipeline automatic-speech-recognition · Library transformers · Repo Badri0510/whisper-tanglish-voice-generalized-5000

Note17 de jul.huggingface.co/Badri0510/whisper-tanglish-voice-generalized-5000

Source context: 1 downloads · 0 likes · Pipeline automatic-speech-recognition · Library transformers · Repo Badri0510/whisper-tanglish-voice-generalized-5000