Fahd1199 commited on
Commit
839680e
·
verified ·
1 Parent(s): 2dd06e9

End of training

Browse files
README.md CHANGED
@@ -2,24 +2,29 @@
2
  language:
3
  - dj
4
  license: apache-2.0
5
- base_model: openai/whisper-medium
6
  tags:
7
  - generated_from_trainer
8
  datasets:
9
- - Fahd1199/MergeV2
10
  model-index:
11
- - name: Whisper medium Darija
12
  results: []
13
  ---
14
 
15
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
16
  should probably proofread and complete it, then remove this comment. -->
17
 
18
- # Whisper medium Darija
19
 
20
- This model is a fine-tuned version of [openai/whisper-medium](https://huggingface.co/openai/whisper-medium) on the MergeV2 dataset.
21
  It achieves the following results on the evaluation set:
22
- - Loss: 5.4036
 
 
 
 
 
23
 
24
  ## Model description
25
 
@@ -38,37 +43,19 @@ More information needed
38
  ### Training hyperparameters
39
 
40
  The following hyperparameters were used during training:
41
- - learning_rate: 0.001
42
  - train_batch_size: 8
43
  - eval_batch_size: 8
44
  - seed: 42
45
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
46
- - lr_scheduler_type: linear
47
- - lr_scheduler_warmup_steps: 50
48
- - num_epochs: 3
49
  - mixed_precision_training: Native AMP
50
 
51
- ### Training results
52
-
53
- | Training Loss | Epoch | Step | Validation Loss |
54
- |:-------------:|:------:|:----:|:---------------:|
55
- | 1.4992 | 0.2415 | 100 | 1.2243 |
56
- | 0.8767 | 0.4831 | 200 | 1.6230 |
57
- | 0.7651 | 0.7246 | 300 | 1.5771 |
58
- | 0.8255 | 0.9662 | 400 | 2.2942 |
59
- | 1.0079 | 1.2077 | 500 | 1.5600 |
60
- | 1.6485 | 1.4493 | 600 | 3.9792 |
61
- | 1.9125 | 1.6908 | 700 | 4.1289 |
62
- | 1.2813 | 1.9324 | 800 | 4.8182 |
63
- | 0.8415 | 2.1739 | 900 | 4.9184 |
64
- | 0.6234 | 2.4155 | 1000 | 5.0654 |
65
- | 0.5348 | 2.6570 | 1100 | 5.2614 |
66
- | 0.4577 | 2.8986 | 1200 | 5.4036 |
67
-
68
-
69
  ### Framework versions
70
 
71
- - Transformers 4.43.4
72
  - Pytorch 2.2.1+cu121
73
  - Datasets 2.20.0
74
  - Tokenizers 0.19.1
 
2
  language:
3
  - dj
4
  license: apache-2.0
5
+ base_model: openai/whisper-small
6
  tags:
7
  - generated_from_trainer
8
  datasets:
9
+ - fahd1199/datasetdarijatest
10
  model-index:
11
+ - name: Whisper Small Darija
12
  results: []
13
  ---
14
 
15
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
16
  should probably proofread and complete it, then remove this comment. -->
17
 
18
+ # Whisper Small Darija
19
 
20
+ This model is a fine-tuned version of [openai/whisper-small](https://huggingface.co/openai/whisper-small) on the DatasetDarijaTest dataset.
21
  It achieves the following results on the evaluation set:
22
+ - eval_loss: 0.7074
23
+ - eval_runtime: 57.1463
24
+ - eval_samples_per_second: 2.625
25
+ - eval_steps_per_second: 0.332
26
+ - epoch: 1.9324
27
+ - step: 800
28
 
29
  ## Model description
30
 
 
43
  ### Training hyperparameters
44
 
45
  The following hyperparameters were used during training:
46
+ - learning_rate: 0.0001
47
  - train_batch_size: 8
48
  - eval_batch_size: 8
49
  - seed: 42
50
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
51
+ - lr_scheduler_type: cosine_with_restarts
52
+ - lr_scheduler_warmup_steps: 200
53
+ - num_epochs: 2
54
  - mixed_precision_training: Native AMP
55
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
56
  ### Framework versions
57
 
58
+ - Transformers 4.44.0
59
  - Pytorch 2.2.1+cu121
60
  - Datasets 2.20.0
61
  - Tokenizers 0.19.1
adapter_config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "base_model_name_or_path": "openai/whisper-medium",
3
  "bias": "none",
4
  "fan_in_fan_out": false,
5
  "inference_mode": true,
@@ -13,9 +13,7 @@
13
  "q_proj",
14
  "v_proj",
15
  "k_proj",
16
- "out_proj",
17
- "fc1",
18
- "fc2"
19
  ],
20
  "task_type": null
21
  }
 
1
  {
2
+ "base_model_name_or_path": "openai/whisper-large-v3",
3
  "bias": "none",
4
  "fan_in_fan_out": false,
5
  "inference_mode": true,
 
13
  "q_proj",
14
  "v_proj",
15
  "k_proj",
16
+ "out_proj"
 
 
17
  ],
18
  "task_type": null
19
  }
adapter_model.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a9f276dbcfca04d96c05e0034aa3a47391adae7db46f7dc297943ea97a4bc163
3
- size 138694026
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fe64dd9eae20dc396b4f7d2a123a56a37f263a1ee75a027dbc2d46e3e60cd19f
3
+ size 126113802
preprocessor_config.json CHANGED
@@ -1,7 +1,7 @@
1
  {
2
  "chunk_length": 30,
3
  "feature_extractor_type": "WhisperFeatureExtractor",
4
- "feature_size": 80,
5
  "hop_length": 160,
6
  "n_fft": 400,
7
  "n_samples": 480000,
 
1
  {
2
  "chunk_length": 30,
3
  "feature_extractor_type": "WhisperFeatureExtractor",
4
+ "feature_size": 128,
5
  "hop_length": 160,
6
  "n_fft": 400,
7
  "n_samples": 480000,
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ee4aba78f27ab5e9af7c7dc8fe1e1692930ac83155d1967efb153538b106327d
3
  size 5304
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f4e3eac49ae7bfae7d06533090373e54f8bb025e7bffd92059bc3d74f9c51020
3
  size 5304