Hugging Face's logo

NbAiLabBeta
/
nb-whisper-large

Automatic Speech Recognition
Transformers
PyTorch
JAX
TensorBoard
ONNX
Safetensors
4 languages
whisper
audio
asr
hf-asr-leaderboard
Inference Endpoints
Model card Files Metrics Community
pere commited on
Commit
fa9caa8
•
1 Parent(s): 94abb9c

Saving weights and logs of step 35000 - epoch 0

Browse files
Files changed (7) hide show
  1. README.md +4 -4
  2. config.json +1 -1
  3. flax_model.msgpack +1 -1
  4. predictions/validation_clean_stortinget_no/step_35000.md +0 -0
  5. predictions/validation_nst/step_35000.md +1 -1
  6. runs/Jan10_10-18-43_t1v-n-a66b4568-w-7/events.out.tfevents.1704881923.t1v-n-a66b4568-w-7.91471.0.v2 +3 -0
  7. training_state.bin +2 -2
README.md CHANGED
@@ -42,13 +42,13 @@ The following hyperparameters were used during training:
42
  - per_device_train_batch_size: 8
43
  - total_train_batch_size_per_node: 32
44
  - total_train_batch_size: 1024
45
- - total_optimization_steps: 50,000
46
- - starting_optimization_step: None
47
  - finishing_optimization_step: 50,000
48
  - num_train_dataset_workers: 32
49
  - num_hosts: 32
50
  - total_num_training_examples: 51,200,000
51
- - steps_per_epoch: 7482
52
  - num_beams: None
53
  - weight_decay: 0.01
54
  - adam_beta1: 0.9
@@ -69,7 +69,7 @@ The following hyperparameters were used during training:
69
  | 20000 | 0.4324 | 0.5187 | 2.3518 | 0.6945 | 2.9561 | 0.7857 | 0.7357 | 8.9839 | 5.7154 | 11.8610 | 6.1664 |
70
  | 25000 | 0.4307 | 0.5158 | 2.3028 | 0.6712 | 2.9343 | 0.7711 | 0.7228 | 9.1284 | 5.8704 | 11.9915 | 6.3161 |
71
  | 30000 | 0.4312 | 0.5108 | 2.2810 | 0.6656 | 2.8690 | 0.7564 | 0.7428 | 8.9010 | 5.6726 | 11.8349 | 6.1305 |
72
- | 35000 | 0.4294 | 0.4957 | 2.2375 | 0.6870 | 2.8417 | 0.7821 | 0.7530 | 8.9294 | 5.6666 | 11.7732 | 6.1110 |
73
 
74
 
75
  ### Framework versions
 
42
  - per_device_train_batch_size: 8
43
  - total_train_batch_size_per_node: 32
44
  - total_train_batch_size: 1024
45
+ - total_optimization_steps: 15,000
46
+ - starting_optimization_step: 35,000
47
  - finishing_optimization_step: 50,000
48
  - num_train_dataset_workers: 32
49
  - num_hosts: 32
50
  - total_num_training_examples: 51,200,000
51
+ - steps_per_epoch: _To be computed after first epoch_
52
  - num_beams: None
53
  - weight_decay: 0.01
54
  - adam_beta1: 0.9
 
69
  | 20000 | 0.4324 | 0.5187 | 2.3518 | 0.6945 | 2.9561 | 0.7857 | 0.7357 | 8.9839 | 5.7154 | 11.8610 | 6.1664 |
70
  | 25000 | 0.4307 | 0.5158 | 2.3028 | 0.6712 | 2.9343 | 0.7711 | 0.7228 | 9.1284 | 5.8704 | 11.9915 | 6.3161 |
71
  | 30000 | 0.4312 | 0.5108 | 2.2810 | 0.6656 | 2.8690 | 0.7564 | 0.7428 | 8.9010 | 5.6726 | 11.8349 | 6.1305 |
72
+ | 35000 | 0.4299 | 0.4908 | 2.2320 | 0.6768 | 2.8417 | 0.7729 | 0.7513 | 8.8015 | 5.6123 | 11.6854 | 6.0642 |
73
 
74
 
75
  ### Framework versions
config.json CHANGED
@@ -1,5 +1,5 @@
1
  {
2
- "_name_or_path": "NbAiLab/nb-whisper-large-v3-RC4",
3
  "activation_dropout": 0.1,
4
  "activation_function": "gelu",
5
  "apply_spec_augment": false,
 
1
  {
2
+ "_name_or_path": "../../../nb-whisper-large-v0.8-vad3",
3
  "activation_dropout": 0.1,
4
  "activation_function": "gelu",
5
  "apply_spec_augment": false,
flax_model.msgpack CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:5587cc4b2c6d432ebe48d8bb87232e0db65b5c8be8ff87ceb93b0d6b4a9f505d
3
  size 3087027463
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9edd3aeb5e9d5f0c7d1e5cd2a15e321917844bf963edba9fb3486ca628b9d129
3
  size 3087027463
predictions/validation_clean_stortinget_no/step_35000.md CHANGED
The diff for this file is too large to render. See raw diff
 
predictions/validation_nst/step_35000.md CHANGED
@@ -2,7 +2,7 @@
2
 
3
  | STEP| loss | wer |cer|
4
  | ---| --- | --- |--- |
5
- | **35000**| 0.429 | 2.237 |0.687 |
6
 
7
  | target | prediction |
8
  |:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
 
2
 
3
  | STEP| loss | wer |cer|
4
  | ---| --- | --- |--- |
5
+ | **35000**| 0.430 | 2.232 |0.677 |
6
 
7
  | target | prediction |
8
  |:-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------|:-----------------------------------------------------------------------------------------------------------------------------------------------------------------------------|
runs/Jan10_10-18-43_t1v-n-a66b4568-w-7/events.out.tfevents.1704881923.t1v-n-a66b4568-w-7.91471.0.v2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:069b7c00385cca4e48d3949c81b3583fa86c5acf98c71837bb15d7c2df69939b
3
+ size 159483
training_state.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7f61e85aae06c429c1a19c9567352b4384613e81df264087cefef85265eb2aea
3
- size 4608
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8478fa617c3e9090af8312175443281b9a6198473ff55fea68cf25576da418c5
3
+ size 4612