whisper-large-v3-base-small-yt-os
This model is a fine-tuned version of openai/whisper-large-v3 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.2389
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: 16
- eval_batch_size: 16
- seed: 42
- gradient_accumulation_steps: 16
- total_train_batch_size: 256
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 1
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
0.2985 | 0.03 | 300 | 0.3361 |
0.2869 | 0.06 | 600 | 0.3195 |
0.2795 | 0.09 | 900 | 0.3062 |
0.2779 | 0.12 | 1200 | 0.2890 |
0.2752 | 0.15 | 1500 | 0.2973 |
0.2818 | 0.18 | 1800 | 0.2881 |
0.2713 | 0.21 | 2100 | 0.2874 |
0.2693 | 0.24 | 2400 | 0.2807 |
0.2657 | 0.28 | 2700 | 0.2731 |
0.2652 | 0.31 | 3000 | 0.2726 |
0.2583 | 0.34 | 3300 | 0.2680 |
0.2607 | 0.37 | 3600 | 0.2767 |
0.2616 | 0.4 | 3900 | 0.2660 |
0.2566 | 0.43 | 4200 | 0.2684 |
0.2527 | 0.46 | 4500 | 0.2525 |
0.257 | 0.49 | 4800 | 0.2582 |
0.2558 | 0.52 | 5100 | 0.2641 |
0.2572 | 0.55 | 5400 | 0.2545 |
0.2485 | 0.58 | 5700 | 0.2620 |
0.2479 | 0.61 | 6000 | 0.2551 |
0.2494 | 0.64 | 6300 | 0.2490 |
0.2466 | 0.67 | 6600 | 0.2496 |
0.25 | 0.7 | 6900 | 0.2492 |
0.2513 | 0.73 | 7200 | 0.2509 |
0.246 | 0.76 | 7500 | 0.2515 |
0.2506 | 0.8 | 7800 | 0.2428 |
0.2483 | 0.83 | 8100 | 0.2367 |
0.2501 | 0.86 | 8400 | 0.2399 |
0.2464 | 0.89 | 8700 | 0.2491 |
0.2402 | 0.92 | 9000 | 0.2448 |
0.246 | 0.95 | 9300 | 0.2418 |
0.251 | 0.98 | 9600 | 0.2389 |
Framework versions
- Transformers 4.37.2
- Pytorch 2.0.1+cu117
- Datasets 2.18.0
- Tokenizers 0.15.2
- Downloads last month
- 0
This model does not have enough activity to be deployed to Inference API (serverless) yet.
Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.