Hugging Face's logo

warmestman
/
whisper-large-v3-mn-cv-fleurs

Automatic Speech Recognition
Transformers
TensorBoard
Safetensors
Mongolian
whisper
hf-asr-leaderboard
Generated from Trainer
Eval Results
Inference Endpoints
Model card Files Metrics Community
warmestman commited on
Commit
c5bb807
•
1 Parent(s): dfed547

End of training

Browse files
Files changed (5) hide show
  1. README.md +16 -20
  2. generation_config.json +1 -0
  3. model-00001-of-00002.safetensors +1 -1
  4. model-00002-of-00002.safetensors +1 -1
  5. runs/Feb20_05-13-43_nrz8795syv/events.out.tfevents.1708406031.nrz8795syv.6055.3 +2 -2
README.md CHANGED
@@ -26,7 +26,7 @@ model-index:
26
  metrics:
27
  - name: Wer
28
  type: wer
29
- value: 31.994939772289754
30
  ---
31
 
32
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
@@ -36,8 +36,8 @@ should probably proofread and complete it, then remove this comment. -->
36
 
37
  This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the Common Voice 16.1 & FLEURS dataset.
38
  It achieves the following results on the evaluation set:
39
- - Loss: 0.5662
40
- - Wer: 31.9949
41
 
42
  ## Model description
43
 
@@ -56,30 +56,26 @@ More information needed
56
  ### Training hyperparameters
57
 
58
  The following hyperparameters were used during training:
59
- - learning_rate: 0.0001
60
- - train_batch_size: 16
61
- - eval_batch_size: 8
62
  - seed: 42
63
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
64
  - lr_scheduler_type: linear
65
- - lr_scheduler_warmup_steps: 500
66
- - training_steps: 10000
67
  - mixed_precision_training: Native AMP
68
 
69
  ### Training results
70
 
71
- | Training Loss | Epoch | Step | Validation Loss | Wer |
72
- |:-------------:|:-----:|:-----:|:---------------:|:-------:|
73
- | 0.0691 | 5.99 | 1000 | 0.4597 | 41.5049 |
74
- | 0.0183 | 11.98 | 2000 | 0.4996 | 38.2982 |
75
- | 0.012 | 17.96 | 3000 | 0.5328 | 38.5402 |
76
- | 0.0091 | 23.95 | 4000 | 0.5619 | 38.1277 |
77
- | 0.004 | 29.94 | 5000 | 0.5439 | 35.2236 |
78
- | 0.0019 | 35.93 | 6000 | 0.5731 | 35.3941 |
79
- | 0.001 | 41.92 | 7000 | 0.5309 | 33.3755 |
80
- | 0.0002 | 47.9 | 8000 | 0.5391 | 32.3140 |
81
- | 0.0 | 53.89 | 9000 | 0.5543 | 32.1984 |
82
- | 0.0 | 59.88 | 10000 | 0.5662 | 31.9949 |
83
 
84
 
85
  ### Framework versions
 
26
  metrics:
27
  - name: Wer
28
  type: wer
29
+ value: 37.049667235025574
30
  ---
31
 
32
  <!-- This model card has been generated automatically according to the information the Trainer had access to. You
 
36
 
37
  This model is a fine-tuned version of [openai/whisper-large-v3](https://huggingface.co/openai/whisper-large-v3) on the Common Voice 16.1 & FLEURS dataset.
38
  It achieves the following results on the evaluation set:
39
+ - Loss: 0.3245
40
+ - Wer: 37.0497
41
 
42
  ## Model description
43
 
 
56
  ### Training hyperparameters
57
 
58
  The following hyperparameters were used during training:
59
+ - learning_rate: 1e-05
60
+ - train_batch_size: 8
61
+ - eval_batch_size: 4
62
  - seed: 42
63
  - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
64
  - lr_scheduler_type: linear
65
+ - lr_scheduler_warmup_steps: 40
66
+ - num_epochs: 2
67
  - mixed_precision_training: Native AMP
68
 
69
  ### Training results
70
 
71
+ | Training Loss | Epoch | Step | Validation Loss | Wer |
72
+ |:-------------:|:-----:|:----:|:---------------:|:-------:|
73
+ | 0.4691 | 0.3 | 100 | 0.5472 | 57.2191 |
74
+ | 0.3191 | 0.6 | 200 | 0.4417 | 49.0237 |
75
+ | 0.2677 | 0.9 | 300 | 0.3791 | 43.3530 |
76
+ | 0.1486 | 1.2 | 400 | 0.3560 | 40.1188 |
77
+ | 0.1387 | 1.5 | 500 | 0.3430 | 37.8912 |
78
+ | 0.1396 | 1.8 | 600 | 0.3245 | 37.0497 |
 
 
 
 
79
 
80
 
81
  ### Framework versions
generation_config.json CHANGED
@@ -161,6 +161,7 @@
161
  "<|yue|>": 50358,
162
  "<|zh|>": 50260
163
  },
 
164
  "max_initial_timestamp_index": 50,
165
  "max_length": 448,
166
  "no_timestamps_token_id": 50364,
 
161
  "<|yue|>": 50358,
162
  "<|zh|>": 50260
163
  },
164
+ "language": "mn",
165
  "max_initial_timestamp_index": 50,
166
  "max_length": 448,
167
  "no_timestamps_token_id": 50364,
model-00001-of-00002.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e971a5194ba027488dfa2604d64eaaa0f1c8bd7fd5fd461d4190e15878642d1f
3
  size 4993448880
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1aac66e698dcf30209216b7f930c52cd515e2ed6146ccba33206f934b7658132
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:aa895cc7bf9bae9de82a668e609b652e1e49d98ddece27b00eb3fac6bba1a884
3
  size 1180663192
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:82d764675948fd85443bd97e56d50fe97154179cdadb016dd65c5d5400cafcb5
3
  size 1180663192
runs/Feb20_05-13-43_nrz8795syv/events.out.tfevents.1708406031.nrz8795syv.6055.3 CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a29d842667e82b3f304122e1e4c14c2a6ce9a400e12f7dd7d0ea570ca42dbb21
3
- size 11085
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f674c9305a1aa9326fad4f42f9b6b9dae26a40c81a9f4f82be5eff1ec0511b96
3
+ size 11753