File size: 3,581 Bytes
ea7acb7 33c380c ea7acb7 33c380c |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 |
---
tags:
- generated_from_trainer
base_model: Lakoc/DeCRED_small_cv_2
datasets:
- common_voice_13_0
metrics:
- wer
model-index:
- name: DeCRED_linear_mixing_tuning
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. -->
# DeCRED_linear_mixing_tuning
This model is a fine-tuned version of [Lakoc/DeCRED_small_cv_2](https://huggingface.co/Lakoc/DeCRED_small_cv_2) on the common_voice_13_0 dataset.
It achieves the following results on the evaluation set:
- Loss: 1.0590
- Cer: 0.0632
- Wer: 0.1471
- Mer: 0.1444
- Wil: 0.2408
- Wip: 0.7592
- Hits: 23158
- Substitutions: 2931
- Deletions: 484
- Insertions: 494
## 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: 0.05
- train_batch_size: 256
- eval_batch_size: 64
- seed: 42
- distributed_type: multi-GPU
- gradient_accumulation_steps: 4
- total_train_batch_size: 1024
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 50.0
### Training results
| Training Loss | Epoch | Step | Validation Loss | Cer | Wer | Mer | Wil | Wip | Hits | Substitutions | Deletions | Insertions |
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|:------:|:------:|:------:|:-----:|:-------------:|:---------:|:----------:|
| 1.185 | 2.67 | 20 | 1.1215 | 0.0646 | 0.1520 | 0.1492 | 0.2479 | 0.7521 | 23030 | 3013 | 530 | 496 |
| 1.0639 | 5.33 | 40 | 1.0717 | 0.0634 | 0.1480 | 0.1453 | 0.2420 | 0.7580 | 23124 | 2939 | 510 | 483 |
| 1.0957 | 8.0 | 60 | 1.0612 | 0.0628 | 0.1464 | 0.1438 | 0.2402 | 0.7598 | 23162 | 2929 | 482 | 480 |
| 1.0936 | 10.67 | 80 | 1.0595 | 0.0631 | 0.1469 | 0.1442 | 0.2407 | 0.7593 | 23158 | 2934 | 481 | 488 |
| 1.0804 | 13.33 | 100 | 1.0591 | 0.0630 | 0.1468 | 0.1441 | 0.2405 | 0.7595 | 23164 | 2929 | 480 | 492 |
| 1.1044 | 16.0 | 120 | 1.0591 | 0.0631 | 0.1469 | 0.1442 | 0.2405 | 0.7595 | 23162 | 2929 | 482 | 492 |
| 1.0836 | 18.67 | 140 | 1.0589 | 0.0631 | 0.1468 | 0.1442 | 0.2405 | 0.7595 | 23163 | 2929 | 481 | 492 |
| 1.0924 | 21.33 | 160 | 1.0590 | 0.0632 | 0.1471 | 0.1444 | 0.2408 | 0.7592 | 23158 | 2931 | 484 | 494 |
| 1.1048 | 24.0 | 180 | 1.0590 | 0.0632 | 0.1471 | 0.1444 | 0.2408 | 0.7592 | 23158 | 2931 | 484 | 494 |
| 1.0858 | 26.67 | 200 | 1.0589 | 0.0631 | 0.1470 | 0.1443 | 0.2407 | 0.7593 | 23160 | 2929 | 484 | 494 |
| 1.0953 | 29.33 | 220 | 1.0589 | 0.0631 | 0.1468 | 0.1442 | 0.2405 | 0.7595 | 23163 | 2929 | 481 | 492 |
| 1.1308 | 32.0 | 240 | 1.0590 | 0.0632 | 0.1471 | 0.1444 | 0.2408 | 0.7592 | 23158 | 2931 | 484 | 494 |
### Framework versions
- Transformers 4.40.0.dev0
- Pytorch 2.2.0+rocm5.6
- Datasets 2.18.0
- Tokenizers 0.15.2
### Wandb run
https://wandb.ai/butspeechfit/decred_commonvoice_en/runs/DeCRED_linear_mixing_tuning |