Hugging Face's logo

FelixK7
/
whisper-large-lv

Automatic Speech Recognition
Transformers
Safetensors
Latvian
whisper
hf-asr-leaderboard
Generated from Trainer
Inference Endpoints
Model card Files Community
FelixK7 commited on
Commit
75826f9
•
1 Parent(s): 9fd34be

End of training

Browse files
Files changed (3) hide show
  1. README.md +15 -13
  2. model-00001-of-00002.safetensors +1 -1
  3. model-00002-of-00002.safetensors +1 -1
README.md CHANGED
@@ -21,8 +21,8 @@ should probably proofread and complete it, then remove this comment. -->
21
 
22
  This model is a fine-tuned version of [AiLab-IMCS-UL/whisper-large-v3-lv-late-cv19](https://huggingface.co/AiLab-IMCS-UL/whisper-large-v3-lv-late-cv19) on the Recorded Voice dataset.
23
  It achieves the following results on the evaluation set:
24
- - Loss: 0.1193
25
- - Wer: 0.0
26
 
27
  ## Model description
28
 
@@ -42,27 +42,29 @@ More information needed
42
 
43
  The following hyperparameters were used during training:
44
  - learning_rate: 1e-05
45
- - train_batch_size: 32
46
- - eval_batch_size: 32
47
  - seed: 42
48
- - gradient_accumulation_steps: 2
49
  - total_train_batch_size: 64
50
  - optimizer: Use adamw_hf with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
51
  - lr_scheduler_type: linear
52
- - lr_scheduler_warmup_steps: 1
53
- - training_steps: 100
54
 
55
  ### Training results
56
 
57
- | Training Loss | Epoch | Step | Validation Loss | Wer |
58
- |:-------------:|:-----:|:----:|:---------------:|:---:|
59
- | 0.0215 | 50.0 | 50 | 0.1396 | 0.0 |
60
- | 0.0173 | 100.0 | 100 | 0.1193 | 0.0 |
 
 
61
 
62
 
63
  ### Framework versions
64
 
65
  - Transformers 4.46.1
66
  - Pytorch 2.3.1
67
- - Datasets 3.0.2
68
- - Tokenizers 0.20.1
 
21
 
22
  This model is a fine-tuned version of [AiLab-IMCS-UL/whisper-large-v3-lv-late-cv19](https://huggingface.co/AiLab-IMCS-UL/whisper-large-v3-lv-late-cv19) on the Recorded Voice dataset.
23
  It achieves the following results on the evaluation set:
24
+ - Loss: 0.4380
25
+ - Wer: 25.9259
26
 
27
  ## Model description
28
 
 
42
 
43
  The following hyperparameters were used during training:
44
  - learning_rate: 1e-05
45
+ - train_batch_size: 16
46
+ - eval_batch_size: 16
47
  - seed: 42
48
+ - gradient_accumulation_steps: 4
49
  - total_train_batch_size: 64
50
  - optimizer: Use adamw_hf with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
51
  - lr_scheduler_type: linear
52
+ - lr_scheduler_warmup_steps: 500
53
+ - training_steps: 5000
54
 
55
  ### Training results
56
 
57
+ | Training Loss | Epoch | Step | Validation Loss | Wer |
58
+ |:-------------:|:--------:|:----:|:---------------:|:-------:|
59
+ | 0.036 | 48.7805 | 1000 | 0.4135 | 19.7531 |
60
+ | 0.0196 | 97.5610 | 2000 | 0.4086 | 22.2222 |
61
+ | 0.0126 | 146.3415 | 3000 | 0.4710 | 23.4568 |
62
+ | 0.0117 | 195.1220 | 4000 | 0.4380 | 25.9259 |
63
 
64
 
65
  ### Framework versions
66
 
67
  - Transformers 4.46.1
68
  - Pytorch 2.3.1
69
+ - Datasets 3.1.0
70
+ - Tokenizers 0.20.2
model-00001-of-00002.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:34ddd50d6fcf594a8b8c8935524d753fdd7db1782eb0c4fa85f2e29a29fc898f
3
  size 4993448880
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:06d472d25ee785019188064f399d7eef1f63079c6fe1cafbc2ed0f47bbc56162
3
  size 4993448880
model-00002-of-00002.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:1b01ac544421fba61734ca30f8dbb135372457dec05d8d7f7031c896e888ce45
3
  size 1180663192
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1cc2bcef33463b17d70eaaf3f60bf3f7af9fbfc4cdb71539affea53d8a974739
3
  size 1180663192