Commit
•
d8cc25f
1
Parent(s):
b3270d3
update model card README.md
Browse files
README.md
CHANGED
|
@@ -2,19 +2,9 @@
|
|
| 2 |
license: apache-2.0
|
| 3 |
tags:
|
| 4 |
- generated_from_trainer
|
| 5 |
-
metrics:
|
| 6 |
-
- wer
|
| 7 |
-
- cer
|
| 8 |
model-index:
|
| 9 |
- name: wav2vec2-large-asr-th
|
| 10 |
results: []
|
| 11 |
-
datasets:
|
| 12 |
-
- mozilla-foundation/common_voice_6_1
|
| 13 |
-
- mozilla-foundation/common_voice_10_0
|
| 14 |
-
language:
|
| 15 |
-
- th
|
| 16 |
-
library_name: transformers
|
| 17 |
-
pipeline_tag: automatic-speech-recognition
|
| 18 |
---
|
| 19 |
|
| 20 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
@@ -22,11 +12,7 @@ should probably proofread and complete it, then remove this comment. -->
|
|
| 22 |
|
| 23 |
# wav2vec2-large-asr-th
|
| 24 |
|
| 25 |
-
This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on
|
| 26 |
-
It achieves the following results on the evaluation set:
|
| 27 |
-
- Loss: 0.7110
|
| 28 |
-
- Wer: 0.418 ±5%
|
| 29 |
-
- Cer: 0.15 ±5%
|
| 30 |
|
| 31 |
## Model description
|
| 32 |
|
|
@@ -54,34 +40,12 @@ The following hyperparameters were used during training:
|
|
| 54 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 55 |
- lr_scheduler_type: linear
|
| 56 |
- lr_scheduler_warmup_steps: 200
|
| 57 |
-
- training_steps:
|
| 58 |
- mixed_precision_training: Native AMP
|
| 59 |
|
| 60 |
-
### Training results
|
| 61 |
-
|
| 62 |
-
| Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
|
| 63 |
-
|:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
|
| 64 |
-
| 3.5509 | 2.35 | 200 | 3.5608 | 1.0 | 1.0000 |
|
| 65 |
-
| 3.507 | 4.71 | 400 | 3.4854 | 1.0 | 1.0000 |
|
| 66 |
-
| 3.3614 | 7.06 | 600 | 3.2711 | 1.0 | 1.0000 |
|
| 67 |
-
| 1.5151 | 9.41 | 800 | 1.1078 | 0.7485 | 0.8674 |
|
| 68 |
-
| 0.9279 | 11.76 | 1000 | 0.7934 | 0.6052 | 0.8534 |
|
| 69 |
-
| 0.7193 | 14.12 | 1200 | 0.7220 | 0.5466 | 0.8491 |
|
| 70 |
-
| 0.5668 | 16.47 | 1400 | 0.6828 | 0.5127 | 0.8459 |
|
| 71 |
-
| 0.4963 | 18.82 | 1600 | 0.6487 | 0.5071 | 0.8451 |
|
| 72 |
-
| 0.4301 | 21.18 | 1800 | 0.6668 | 0.4946 | 0.8442 |
|
| 73 |
-
| 0.3881 | 23.53 | 2000 | 0.6685 | 0.4806 | 0.8434 |
|
| 74 |
-
| 0.3628 | 25.88 | 2200 | 0.6911 | 0.4836 | 0.8433 |
|
| 75 |
-
| 0.3711 | 28.24 | 2400 | 0.7008 | 0.4795 | 0.8430 |
|
| 76 |
-
| 0.351 | 30.59 | 2600 | 0.6974 | 0.4697 | 0.8424 |
|
| 77 |
-
| 0.2799 | 32.94 | 2800 | 0.7090 | 0.4705 | 0.8421 |
|
| 78 |
-
| 0.2814 | 35.29 | 3000 | 0.7110 | 0.4690 | 0.8418 |
|
| 79 |
-
| 0.2707 | 37.65 | 3200 | 0.7090 | 0.4681 | 0.8418 |
|
| 80 |
-
|
| 81 |
-
|
| 82 |
### Framework versions
|
| 83 |
|
| 84 |
- Transformers 4.26.1
|
| 85 |
- Pytorch 1.13.1+cu116
|
| 86 |
- Datasets 2.9.0
|
| 87 |
-
- Tokenizers 0.13.2
|
|
|
|
| 2 |
license: apache-2.0
|
| 3 |
tags:
|
| 4 |
- generated_from_trainer
|
|
|
|
|
|
|
|
|
|
| 5 |
model-index:
|
| 6 |
- name: wav2vec2-large-asr-th
|
| 7 |
results: []
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
---
|
| 9 |
|
| 10 |
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
|
|
|
| 12 |
|
| 13 |
# wav2vec2-large-asr-th
|
| 14 |
|
| 15 |
+
This model is a fine-tuned version of [facebook/wav2vec2-large-xlsr-53](https://huggingface.co/facebook/wav2vec2-large-xlsr-53) on an unknown dataset.
|
|
|
|
|
|
|
|
|
|
|
|
|
| 16 |
|
| 17 |
## Model description
|
| 18 |
|
|
|
|
| 40 |
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 41 |
- lr_scheduler_type: linear
|
| 42 |
- lr_scheduler_warmup_steps: 200
|
| 43 |
+
- training_steps: 3500
|
| 44 |
- mixed_precision_training: Native AMP
|
| 45 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 46 |
### Framework versions
|
| 47 |
|
| 48 |
- Transformers 4.26.1
|
| 49 |
- Pytorch 1.13.1+cu116
|
| 50 |
- Datasets 2.9.0
|
| 51 |
+
- Tokenizers 0.13.2
|