Facebook Mms 1b All Km Indomain20 Withac | Sweet Tea Studio
Resources / Facebook Mms 1b All Km Indomain20 Withac Facebook Mms 1b All Km Indomain20 Withac This model is a fine-tuned version of facebook/mms-1b-all on the None dataset. It achieves the following results on the evaluation set: Loss: 0.1921 Wer: 0.1624 Model description
Verified source
Kind automatic-speech-recognition Base model facebook/mms-1b-all Version v49c0d4d42c1be0e9fdfe5d954732953d9b0dc420 License cc-by-nc-4.0 Publisher @Shagufta C grade Model source
Kind automatic-speech-recognition
Base model facebook/mms-1b-all
Version v49c0d4d42c1be0e9fdfe5d954732953d9b0dc420
License cc-by-nc-4.0
Parameters 1B
Tasks Video
Source Hugging Face --- library_name: transformers license: cc-by-nc-4.0 base_model: facebook/mms-1b-all tags: - generated_from_trainer metrics: - wer model-index: - name: facebook-mms-1b-all-km-indomain20-withac results: [] --- # facebook-mms-1b-all-km-indomain20-withac This model is a fine-tuned version of [facebook/mms-1b-all](https://huggingface.co/facebook/mms-1b-all) on the None dataset. It achieves the following results on the evaluation set: - Loss: 0.1921 - Wer: 0.1624 ## 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: 1.5e-05 - train_batch_size: 4 - eval_batch_size: 4 - seed: 42 - gradient_accumulation_steps: 4 - 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: 150 - num_epochs: 20 - mixed_precision_training: Native AMP ### Training results | Training Loss | Epoch | Step | Validation Loss | Wer | |:-------------:|:-------:|:----:|:---------------:|:------:| | 11.6650 | 0.9445 | 200 | 2.9165 | 1.0 | | 1.9822 | 1.8878 | 400 | 0.3216 | 0.2960 | | 1.4406 | 2.8312 | 600 | 0.2689 | 0.2619 | | 1.2779 | 3.7745 | 800 | 0.2356 | 0.2238 | | 1.1193 | 4.7178 | 1000 | 0.2190 | 0.2080 | | 1.0423 | 5.6612 | 1200 | 0.2154 | 0.2038 | | 0.9447 | 6.6045 | 1400 | 0.2070 | 0.1944 | | 0.8678 | 7.5478 | 1600 | 0.2039 | 0.1877 | | 0.8488 | 8.4911 | 1800 | 0.1984 | 0.1817 | | 0.7765 | 9.4345 | 2000 | 0.2037 | 0.1793 | | 0.7410 | 10.3778 | 2200 | 0.1944 | 0.1726 | | 0.7008 | 11.3211 | 2400 | 0.1983 | 0.1720 | | 0.7001 | 12.2645 | 2600 | 0.1961 | 0.1741 | | 0.6553 | 13.2078 | 2800 | 0.1967 | 0.1684 | | 0.6469 | 14.1511 | 3000 | 0.1953 | 0.1682 | | 0.6271 | 15.0945 | 3200 | 0.1921 | 0.1641 | | 0.6198 | 16.0378 | 3400 | 0.1921 | 0.1628 | | 0.6432 | 16.9823 | 3600 | 0.1924 | 0.1639 | | 0.6262 | 17.9256 | 3800 | 0.1926 | 0.1627 | | 0.5928 | 18.8689 | 4000 | 0.1926 | 0.1624 | | 0.5891 | 19.8123 | 4200 | 0.1921 | 0.1624 | | 0.5891 | 20.0 | 4240 | 0.1921 | 0.1624 | ### Framework versions - Transformers 5.9.0 - Pytorch 2.11.0+cu128 - Datasets 4.0.0 - Tokenizers 0.22.2
Sources & provenance
1 active source Source evidence
3 excerpts This model is a fine-tuned version of facebook/mms-1b-all on the None dataset. It achieves the following results on the evaluation set: Loss: 0.1921 Wer: 0.1624 Model description
Jul 11
libraryname: transformers license: cc-by-nc-4.0 basemodel: facebook/mms-1b-all tags: generatedfromtrainer metrics: wer model-index: name: facebook-mms-1b-all-km-indomain20-withac results: []
Shagufta/facebook-mms-1b-all-km-indomain20-withac
