Hugging Face's logo

b-brave
/
asr2_medium_v0.9

PEFT
TensorBoard
Safetensors
Italian
Generated from Trainer
Eval Results
Model card Files Metrics Community
miosipof commited on
Commit
53b3bbf
·
verified ·
1 Parent(s): 72b6a08

End of training

Browse files
Files changed (5) hide show
  1. README.md +20 -18
  2. adapter_config.json +2 -2
  3. adapter_model.safetensors +1 -1
  4. runs/Feb15_11-46-13_e7511933f02a/events.out.tfevents.1739619975.e7511933f02a +3 -0
  5. training_args.bin +1 -1
README.md CHANGED
@@ -24,7 +24,7 @@ model-index:
24
  args: default
25
  metrics:
26
  - type: wer
27
- value: 76.07449856733524
28
  name: Wer
29
  ---
30
 
@@ -35,9 +35,9 @@ should probably proofread and complete it, then remove this comment. -->
35
 
36
  This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the b-brave-clean dataset.
37
  It achieves the following results on the evaluation set:
38
- - Loss: 0.7540
39
- - Wer: 76.0745
40
- - Cer: 60.6514
41
  - Lr: 0.0000
42
 
43
  ## Model description
@@ -57,7 +57,7 @@ More information needed
57
  ### Training hyperparameters
58
 
59
  The following hyperparameters were used during training:
60
- - learning_rate: 0.0001
61
  - train_batch_size: 8
62
  - eval_batch_size: 8
63
  - seed: 42
@@ -66,23 +66,25 @@ The following hyperparameters were used during training:
66
  - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
67
  - lr_scheduler_type: linear
68
  - lr_scheduler_warmup_ratio: 0.3
69
- - num_epochs: 10
70
  - mixed_precision_training: Native AMP
71
 
72
  ### Training results
73
 
74
- | Training Loss | Epoch | Step | Validation Loss | Wer | Cer | Lr |
75
- |:-------------:|:-----:|:----:|:---------------:|:--------:|:-------:|:------:|
76
- | 3.7378 | 1.0 | 251 | 3.4955 | 122.6361 | 91.8308 | 0.0000 |
77
- | 1.1797 | 2.0 | 502 | 1.1534 | 80.6590 | 50.3546 | 0.0001 |
78
- | 0.9184 | 3.0 | 753 | 0.9837 | 71.3467 | 45.7578 | 0.0001 |
79
- | 0.6271 | 4.0 | 1004 | 0.8690 | 90.2579 | 72.3667 | 0.0001 |
80
- | 0.4524 | 5.0 | 1255 | 0.8207 | 79.3696 | 66.6141 | 0.0001 |
81
- | 0.3238 | 6.0 | 1506 | 0.7704 | 80.2292 | 68.0063 | 0.0001 |
82
- | 0.2246 | 7.0 | 1757 | 0.7517 | 77.9370 | 61.9648 | 0.0000 |
83
- | 0.1642 | 8.0 | 2008 | 0.7447 | 56.5903 | 38.1928 | 0.0000 |
84
- | 0.1172 | 9.0 | 2259 | 0.7502 | 79.0831 | 62.9892 | 0.0000 |
85
- | 0.0787 | 10.0 | 2510 | 0.7540 | 76.0745 | 60.6514 | 0.0000 |
 
 
86
 
87
 
88
  ### Framework versions
 
24
  args: default
25
  metrics:
26
  - type: wer
27
+ value: 41.833810888252145
28
  name: Wer
29
  ---
30
 
 
35
 
36
  This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the b-brave-clean dataset.
37
  It achieves the following results on the evaluation set:
38
+ - Loss: 0.6398
39
+ - Wer: 41.8338
40
+ - Cer: 30.4702
41
  - Lr: 0.0000
42
 
43
  ## Model description
 
57
  ### Training hyperparameters
58
 
59
  The following hyperparameters were used during training:
60
+ - learning_rate: 0.0003
61
  - train_batch_size: 8
62
  - eval_batch_size: 8
63
  - seed: 42
 
66
  - optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
67
  - lr_scheduler_type: linear
68
  - lr_scheduler_warmup_ratio: 0.3
69
+ - num_epochs: 12
70
  - mixed_precision_training: Native AMP
71
 
72
  ### Training results
73
 
74
+ | Training Loss | Epoch | Step | Validation Loss | Wer | Cer | Lr |
75
+ |:-------------:|:-------:|:----:|:---------------:|:--------:|:--------:|:------:|
76
+ | 3.6633 | 1.0 | 168 | 1.8570 | 149.4269 | 97.5309 | 0.0001 |
77
+ | 1.2316 | 2.0 | 336 | 0.9484 | 66.4756 | 45.5477 | 0.0002 |
78
+ | 0.8773 | 3.0 | 504 | 0.8240 | 117.3352 | 102.2590 | 0.0002 |
79
+ | 0.5747 | 4.0 | 672 | 0.7314 | 76.3610 | 56.0809 | 0.0003 |
80
+ | 0.3652 | 5.0 | 840 | 0.6545 | 104.1547 | 90.4124 | 0.0002 |
81
+ | 0.2632 | 6.0 | 1008 | 0.6305 | 50.5731 | 34.6467 | 0.0002 |
82
+ | 0.1568 | 7.0 | 1176 | 0.5926 | 90.1146 | 80.4833 | 0.0002 |
83
+ | 0.104 | 8.0 | 1344 | 0.6388 | 47.5645 | 33.4121 | 0.0001 |
84
+ | 0.0356 | 9.0 | 1512 | 0.6073 | 42.8367 | 46.9398 | 0.0001 |
85
+ | 0.0189 | 10.0 | 1680 | 0.6361 | 42.5501 | 47.1500 | 0.0001 |
86
+ | 0.0089 | 11.0 | 1848 | 0.6385 | 41.9771 | 30.4965 | 0.0000 |
87
+ | 0.0108 | 11.9313 | 2004 | 0.6398 | 41.8338 | 30.4702 | 0.0000 |
88
 
89
 
90
  ### Framework versions
adapter_config.json CHANGED
@@ -26,9 +26,9 @@
26
  "rank_pattern": {},
27
  "revision": null,
28
  "target_modules": [
29
- "v_proj",
30
  "k_proj",
31
- "q_proj"
 
32
  ],
33
  "task_type": null,
34
  "use_dora": false,
 
26
  "rank_pattern": {},
27
  "revision": null,
28
  "target_modules": [
 
29
  "k_proj",
30
+ "q_proj",
31
+ "v_proj"
32
  ],
33
  "task_type": null,
34
  "use_dora": false,
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:29b3ff2a85199993eeb2fda54e10814abc0ca9417f24b00f07d5346fd489beb4
3
  size 56685000
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f442d4908fdc53f54e3e67301f7ce91b81b488428248eb69e10e8e4568a33eed
3
  size 56685000
runs/Feb15_11-46-13_e7511933f02a/events.out.tfevents.1739619975.e7511933f02a ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:97c11ec82b5e087d905a224a26593341f67908ea5d6c8f5340f7c2814d1eb556
3
+ size 28972
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:77c07e6a08fe5fb8627c0da1abb6413c4aa8aa59b19f4bcf09d29412b54263a8
3
  size 5496
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1b90c1f79a11de63d81de4c2fe0cb2aecd4a1346e022c19471e037c6cbfa462
3
  size 5496