Hugging Face's logo

asr-malayalam
/
malayalam_combined_

TensorBoard
Safetensors
wav2vec2-bert
Generated from Trainer
Model card Files Metrics Community
akrishnan commited on
Commit
5f840e0
•
1 Parent(s): 8146240

End of training

Browse files
Files changed (2) hide show
  1. README.md +29 -54
  2. model.safetensors +1 -1
README.md CHANGED
@@ -13,13 +13,13 @@ model-index:
13
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
14
  should probably proofread and complete it, then remove this comment. -->
15
 
16
- [<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/krishnan-aravind/huggingface/runs/8awzxwte)
17
  # malayalam_combined_
18
 
19
  This model is a fine-tuned version of [facebook/w2v-bert-2.0](https://huggingface.co/facebook/w2v-bert-2.0) on the None dataset.
20
  It achieves the following results on the evaluation set:
21
- - Loss: 0.4324
22
- - Wer: 0.4182
23
 
24
  ## Model description
25
 
@@ -40,69 +40,44 @@ More information needed
40
  The following hyperparameters were used during training:
41
  - learning_rate: 5e-05
42
  - train_batch_size: 16
43
- - eval_batch_size: 8
44
  - seed: 42
45
  - gradient_accumulation_steps: 2
46
  - total_train_batch_size: 32
47
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
48
  - lr_scheduler_type: linear
49
  - lr_scheduler_warmup_steps: 50
50
- - num_epochs: 10
51
  - mixed_precision_training: Native AMP
52
 
53
  ### Training results
54
 
55
  | Training Loss | Epoch | Step | Validation Loss | Wer |
56
  |:-------------:|:------:|:-----:|:---------------:|:------:|
57
- | 0.8323 | 0.2031 | 500 | 0.8530 | 0.6902 |
58
- | 0.7407 | 0.4063 | 1000 | 0.7524 | 0.6254 |
59
- | 0.6897 | 0.6094 | 1500 | 0.6791 | 0.6136 |
60
- | 0.6347 | 0.8125 | 2000 | 0.6630 | 0.5799 |
61
- | 0.5888 | 1.0156 | 2500 | 0.6322 | 0.5833 |
62
- | 0.5818 | 1.2188 | 3000 | 0.6162 | 0.5866 |
63
- | 0.5732 | 1.4219 | 3500 | 0.6035 | 0.5519 |
64
- | 0.5451 | 1.6250 | 4000 | 0.5993 | 0.5303 |
65
- | 0.534 | 1.8282 | 4500 | 0.5844 | 0.5404 |
66
- | 0.5032 | 2.0313 | 5000 | 0.5632 | 0.5245 |
67
- | 0.4977 | 2.2344 | 5500 | 0.5445 | 0.5323 |
68
- | 0.5048 | 2.4375 | 6000 | 0.5302 | 0.5090 |
69
- | 0.5039 | 2.6407 | 6500 | 0.5409 | 0.5172 |
70
- | 0.4874 | 2.8438 | 7000 | 0.5326 | 0.5255 |
71
- | 0.4498 | 3.0469 | 7500 | 0.5139 | 0.4958 |
72
- | 0.4802 | 3.2501 | 8000 | 0.5199 | 0.5168 |
73
- | 0.452 | 3.4532 | 8500 | 0.5030 | 0.4990 |
74
- | 0.4526 | 3.6563 | 9000 | 0.4942 | 0.4841 |
75
- | 0.438 | 3.8594 | 9500 | 0.4896 | 0.4908 |
76
- | 0.4285 | 4.0626 | 10000 | 0.5156 | 0.4869 |
77
- | 0.4132 | 4.2657 | 10500 | 0.4971 | 0.4805 |
78
- | 0.4163 | 4.4688 | 11000 | 0.4845 | 0.4805 |
79
- | 0.3999 | 4.6719 | 11500 | 0.4804 | 0.4778 |
80
- | 0.3975 | 4.8751 | 12000 | 0.4743 | 0.4732 |
81
- | 0.3823 | 5.0782 | 12500 | 0.5002 | 0.4637 |
82
- | 0.3893 | 5.2813 | 13000 | 0.4793 | 0.4666 |
83
- | 0.3789 | 5.4845 | 13500 | 0.4742 | 0.4564 |
84
- | 0.3718 | 5.6876 | 14000 | 0.4731 | 0.4606 |
85
- | 0.3813 | 5.8907 | 14500 | 0.4609 | 0.4639 |
86
- | 0.3498 | 6.0938 | 15000 | 0.4645 | 0.4532 |
87
- | 0.348 | 6.2970 | 15500 | 0.4662 | 0.4508 |
88
- | 0.3263 | 6.5001 | 16000 | 0.4635 | 0.4493 |
89
- | 0.3404 | 6.7032 | 16500 | 0.4522 | 0.4399 |
90
- | 0.3352 | 6.9064 | 17000 | 0.4439 | 0.4459 |
91
- | 0.3192 | 7.1095 | 17500 | 0.4490 | 0.4486 |
92
- | 0.3035 | 7.3126 | 18000 | 0.4474 | 0.4381 |
93
- | 0.3175 | 7.5157 | 18500 | 0.4431 | 0.4374 |
94
- | 0.3161 | 7.7189 | 19000 | 0.4458 | 0.4344 |
95
- | 0.3244 | 7.9220 | 19500 | 0.4384 | 0.4377 |
96
- | 0.2795 | 8.1251 | 20000 | 0.4444 | 0.4326 |
97
- | 0.2975 | 8.3283 | 20500 | 0.4403 | 0.4307 |
98
- | 0.2854 | 8.5314 | 21000 | 0.4352 | 0.4278 |
99
- | 0.2856 | 8.7345 | 21500 | 0.4344 | 0.4288 |
100
- | 0.2764 | 8.9376 | 22000 | 0.4356 | 0.4239 |
101
- | 0.2546 | 9.1408 | 22500 | 0.4426 | 0.4229 |
102
- | 0.2703 | 9.3439 | 23000 | 0.4336 | 0.4237 |
103
- | 0.2438 | 9.5470 | 23500 | 0.4358 | 0.4218 |
104
- | 0.2612 | 9.7502 | 24000 | 0.4337 | 0.4195 |
105
- | 0.2698 | 9.9533 | 24500 | 0.4324 | 0.4182 |
106
 
107
 
108
  ### Framework versions
 
13
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
14
  should probably proofread and complete it, then remove this comment. -->
15
 
16
+ [<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/krishnan-aravind/huggingface/runs/e0c2wxc6)
17
  # malayalam_combined_
18
 
19
  This model is a fine-tuned version of [facebook/w2v-bert-2.0](https://huggingface.co/facebook/w2v-bert-2.0) on the None dataset.
20
  It achieves the following results on the evaluation set:
21
+ - Loss: 0.5153
22
+ - Wer: 0.5077
23
 
24
  ## Model description
25
 
 
40
  The following hyperparameters were used during training:
41
  - learning_rate: 5e-05
42
  - train_batch_size: 16
43
+ - eval_batch_size: 16
44
  - seed: 42
45
  - gradient_accumulation_steps: 2
46
  - total_train_batch_size: 32
47
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
48
  - lr_scheduler_type: linear
49
  - lr_scheduler_warmup_steps: 50
50
+ - num_epochs: 25
51
  - mixed_precision_training: Native AMP
52
 
53
  ### Training results
54
 
55
  | Training Loss | Epoch | Step | Validation Loss | Wer |
56
  |:-------------:|:------:|:-----:|:---------------:|:------:|
57
+ | 0.8243 | 0.2031 | 500 | 0.8413 | 0.6658 |
58
+ | 0.7336 | 0.4063 | 1000 | 0.7351 | 0.6251 |
59
+ | 0.6824 | 0.6094 | 1500 | 0.6786 | 0.5956 |
60
+ | 0.6489 | 0.8125 | 2000 | 0.6836 | 0.6075 |
61
+ | 0.585 | 1.0156 | 2500 | 0.6295 | 0.5864 |
62
+ | 0.5917 | 1.2188 | 3000 | 0.6166 | 0.5579 |
63
+ | 0.56 | 1.4219 | 3500 | 0.6006 | 0.5646 |
64
+ | 0.5736 | 1.6250 | 4000 | 0.6268 | 0.5643 |
65
+ | 0.5821 | 1.8282 | 4500 | 0.6216 | 0.5786 |
66
+ | 0.5505 | 2.0313 | 5000 | 0.5705 | 0.5379 |
67
+ | 0.5065 | 2.2344 | 5500 | 0.5864 | 0.5460 |
68
+ | 0.5004 | 2.4375 | 6000 | 0.5555 | 0.5259 |
69
+ | 0.5327 | 2.6407 | 6500 | 0.5539 | 0.5255 |
70
+ | 0.5148 | 2.8438 | 7000 | 0.5584 | 0.5457 |
71
+ | 0.4751 | 3.0469 | 7500 | 0.5389 | 0.5208 |
72
+ | 0.4779 | 3.2501 | 8000 | 0.5284 | 0.5102 |
73
+ | 0.4874 | 3.4532 | 8500 | 0.5300 | 0.5084 |
74
+ | 0.4955 | 3.6563 | 9000 | 0.5248 | 0.5125 |
75
+ | 0.4961 | 3.8594 | 9500 | 0.5116 | 0.5061 |
76
+ | 0.4449 | 4.0626 | 10000 | 0.5257 | 0.5122 |
77
+ | 0.48 | 4.2657 | 10500 | 0.5254 | 0.5046 |
78
+ | 0.4513 | 4.4688 | 11000 | 0.5364 | 0.5232 |
79
+ | 0.4698 | 4.6719 | 11500 | 0.5293 | 0.5106 |
80
+ | 0.4674 | 4.8751 | 12000 | 0.5153 | 0.5077 |
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
81
 
82
 
83
  ### Framework versions
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:07e48a3e33475e3b2af4e811d18dd565d6a6c0b1102384ce6e095af4aacd4157
3
  size 2423199960
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1c80e7d2b3bcd80794a838ea7a69d53e5e32c247cb6bcf6c9e61c671143203e8
3
  size 2423199960