Hugging Face's logo

mskhattori
/
jdrt_byclass_rinnna_hubert_asr_2

Automatic Speech Recognition
Transformers
PyTorch
hubert
Generated from Trainer
Inference Endpoints
Model card Files Community
Masaki Hattori commited on
Commit
6a9faba
•
1 Parent(s): 8e3fa5f

jdrt_byclass_rinnna_hubert_asr_2

Browse files
Files changed (5) hide show
  1. README.md +111 -0
  2. config.json +72 -0
  3. preprocessor_config.json +9 -0
  4. pytorch_model.bin +3 -0
  5. training_args.bin +3 -0
README.md ADDED
@@ -0,0 +1,111 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model: rinna/japanese-hubert-base
4
+ tags:
5
+ - generated_from_trainer
6
+ metrics:
7
+ - wer
8
+ model-index:
9
+ - name: jdrt_byclass_rinnna_hubert_asr_2
10
+ results: []
11
+ ---
12
+
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
+ # jdrt_byclass_rinnna_hubert_asr_2
17
+
18
+ This model is a fine-tuned version of [rinna/japanese-hubert-base](https://huggingface.co/rinna/japanese-hubert-base) on the None dataset.
19
+ It achieves the following results on the evaluation set:
20
+ - Loss: 0.4574
21
+ - Wer: 0.4461
22
+ - Cer: 0.3253
23
+
24
+ ## Model description
25
+
26
+ More information needed
27
+
28
+ ## Intended uses & limitations
29
+
30
+ More information needed
31
+
32
+ ## Training and evaluation data
33
+
34
+ More information needed
35
+
36
+ ## Training procedure
37
+
38
+ ### Training hyperparameters
39
+
40
+ The following hyperparameters were used during training:
41
+ - learning_rate: 5e-05
42
+ - train_batch_size: 256
43
+ - eval_batch_size: 256
44
+ - seed: 42
45
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
46
+ - lr_scheduler_type: cosine
47
+ - lr_scheduler_warmup_steps: 260
48
+ - num_epochs: 50
49
+
50
+ ### Training results
51
+
52
+ | Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
53
+ |:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
54
+ | 11.8822 | 1.0 | 53 | 8.0828 | 0.9838 | 0.9911 |
55
+ | 6.3754 | 2.0 | 106 | 4.6774 | 0.9838 | 0.9911 |
56
+ | 4.5264 | 3.0 | 159 | 3.9963 | 0.9838 | 0.9911 |
57
+ | 3.7867 | 4.0 | 212 | 3.3575 | 0.9838 | 0.9911 |
58
+ | 3.2022 | 5.0 | 265 | 2.9007 | 0.9838 | 0.9911 |
59
+ | 2.8134 | 6.0 | 318 | 2.7079 | 0.9838 | 0.9911 |
60
+ | 2.708 | 7.0 | 371 | 2.6779 | 0.9838 | 0.9911 |
61
+ | 2.6923 | 8.0 | 424 | 2.6719 | 0.9838 | 0.9911 |
62
+ | 2.6555 | 9.0 | 477 | 2.4997 | 0.9838 | 0.9911 |
63
+ | 2.3308 | 10.0 | 530 | 2.1709 | 0.9960 | 0.9522 |
64
+ | 2.1951 | 11.0 | 583 | 2.0579 | 0.9960 | 0.9522 |
65
+ | 2.0032 | 12.0 | 636 | 1.8255 | 0.9989 | 0.9302 |
66
+ | 1.8394 | 13.0 | 689 | 1.6870 | 0.9999 | 0.9021 |
67
+ | 1.6562 | 14.0 | 742 | 1.4793 | 0.9999 | 0.8139 |
68
+ | 1.5057 | 15.0 | 795 | 1.2942 | 0.9997 | 0.7813 |
69
+ | 1.4022 | 16.0 | 848 | 1.2146 | 0.9997 | 0.7594 |
70
+ | 1.2945 | 17.0 | 901 | 1.1512 | 0.9941 | 0.7395 |
71
+ | 1.2062 | 18.0 | 954 | 1.0732 | 0.9844 | 0.6726 |
72
+ | 1.1396 | 19.0 | 1007 | 1.0457 | 0.9694 | 0.6559 |
73
+ | 1.0419 | 20.0 | 1060 | 0.8567 | 0.5929 | 0.4214 |
74
+ | 0.8799 | 21.0 | 1113 | 0.7172 | 0.5273 | 0.3951 |
75
+ | 0.8249 | 22.0 | 1166 | 0.6980 | 0.5204 | 0.4048 |
76
+ | 0.7649 | 23.0 | 1219 | 0.6157 | 0.4923 | 0.3737 |
77
+ | 0.7401 | 24.0 | 1272 | 0.6171 | 0.4992 | 0.3722 |
78
+ | 0.7074 | 25.0 | 1325 | 0.5948 | 0.4835 | 0.3669 |
79
+ | 0.6543 | 26.0 | 1378 | 0.5520 | 0.4759 | 0.3537 |
80
+ | 0.6164 | 27.0 | 1431 | 0.5512 | 0.4745 | 0.3467 |
81
+ | 0.5758 | 28.0 | 1484 | 0.5286 | 0.4677 | 0.3416 |
82
+ | 0.5648 | 29.0 | 1537 | 0.5099 | 0.4644 | 0.3447 |
83
+ | 0.578 | 30.0 | 1590 | 0.5262 | 0.4701 | 0.3530 |
84
+ | 0.5343 | 31.0 | 1643 | 0.5025 | 0.4621 | 0.3463 |
85
+ | 0.5236 | 32.0 | 1696 | 0.4927 | 0.4588 | 0.3411 |
86
+ | 0.5256 | 33.0 | 1749 | 0.4971 | 0.4614 | 0.3434 |
87
+ | 0.5309 | 34.0 | 1802 | 0.4882 | 0.4553 | 0.3400 |
88
+ | 0.4938 | 35.0 | 1855 | 0.4663 | 0.4497 | 0.3309 |
89
+ | 0.4874 | 36.0 | 1908 | 0.4734 | 0.4508 | 0.3316 |
90
+ | 0.4519 | 37.0 | 1961 | 0.4642 | 0.4483 | 0.3289 |
91
+ | 0.5137 | 38.0 | 2014 | 0.4655 | 0.4490 | 0.3303 |
92
+ | 0.4832 | 39.0 | 2067 | 0.4613 | 0.4474 | 0.3282 |
93
+ | 0.4581 | 40.0 | 2120 | 0.4565 | 0.4450 | 0.3236 |
94
+ | 0.4542 | 41.0 | 2173 | 0.4550 | 0.4461 | 0.3262 |
95
+ | 0.4478 | 42.0 | 2226 | 0.4548 | 0.4442 | 0.3247 |
96
+ | 0.448 | 43.0 | 2279 | 0.4523 | 0.4450 | 0.3238 |
97
+ | 0.4468 | 44.0 | 2332 | 0.4560 | 0.4455 | 0.3224 |
98
+ | 0.4681 | 45.0 | 2385 | 0.4599 | 0.4468 | 0.3253 |
99
+ | 0.455 | 46.0 | 2438 | 0.4589 | 0.4472 | 0.3246 |
100
+ | 0.4479 | 47.0 | 2491 | 0.4583 | 0.4465 | 0.3245 |
101
+ | 0.4533 | 48.0 | 2544 | 0.4582 | 0.4469 | 0.3253 |
102
+ | 0.4486 | 49.0 | 2597 | 0.4575 | 0.4462 | 0.3253 |
103
+ | 0.4466 | 50.0 | 2650 | 0.4574 | 0.4461 | 0.3253 |
104
+
105
+
106
+ ### Framework versions
107
+
108
+ - Transformers 4.34.0.dev0
109
+ - Pytorch 2.0.1+cu118
110
+ - Datasets 2.14.5
111
+ - Tokenizers 0.13.3
config.json ADDED
@@ -0,0 +1,72 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "rinna/japanese-hubert-base",
3
+ "activation_dropout": 0.1,
4
+ "apply_spec_augment": true,
5
+ "architectures": [
6
+ "HubertForCTC"
7
+ ],
8
+ "attention_dropout": 0.1,
9
+ "bos_token_id": 1,
10
+ "classifier_proj_size": 256,
11
+ "conv_bias": false,
12
+ "conv_dim": [
13
+ 512,
14
+ 512,
15
+ 512,
16
+ 512,
17
+ 512,
18
+ 512,
19
+ 512
20
+ ],
21
+ "conv_kernel": [
22
+ 10,
23
+ 3,
24
+ 3,
25
+ 3,
26
+ 3,
27
+ 2,
28
+ 2
29
+ ],
30
+ "conv_stride": [
31
+ 5,
32
+ 2,
33
+ 2,
34
+ 2,
35
+ 2,
36
+ 2,
37
+ 2
38
+ ],
39
+ "ctc_loss_reduction": "mean",
40
+ "ctc_zero_infinity": false,
41
+ "do_stable_layer_norm": false,
42
+ "eos_token_id": 2,
43
+ "feat_extract_activation": "gelu",
44
+ "feat_extract_norm": "group",
45
+ "feat_proj_dropout": 0.1,
46
+ "feat_proj_layer_norm": true,
47
+ "final_dropout": 0.1,
48
+ "hidden_act": "gelu",
49
+ "hidden_dropout": 0,
50
+ "hidden_size": 768,
51
+ "initializer_range": 0.02,
52
+ "intermediate_size": 3072,
53
+ "layer_norm_eps": 1e-05,
54
+ "layerdrop": 0.1,
55
+ "mask_feature_length": 10,
56
+ "mask_feature_min_masks": 0,
57
+ "mask_feature_prob": 0.0,
58
+ "mask_time_length": 10,
59
+ "mask_time_min_masks": 2,
60
+ "mask_time_prob": 0.05,
61
+ "model_type": "hubert",
62
+ "num_attention_heads": 12,
63
+ "num_conv_pos_embedding_groups": 16,
64
+ "num_conv_pos_embeddings": 128,
65
+ "num_feat_extract_layers": 7,
66
+ "num_hidden_layers": 12,
67
+ "pad_token_id": 28,
68
+ "torch_dtype": "float32",
69
+ "transformers_version": "4.34.0.dev0",
70
+ "use_weighted_layer_sum": false,
71
+ "vocab_size": 29
72
+ }
preprocessor_config.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "do_normalize": true,
3
+ "feature_extractor_type": "Wav2Vec2FeatureExtractor",
4
+ "feature_size": 1,
5
+ "padding_side": "right",
6
+ "padding_value": 0.0,
7
+ "return_attention_mask": true,
8
+ "sampling_rate": 16000
9
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ce138f33909c9f4784589f06de30fa10b0ee265d231bdbb1bbe65ad0a619a5bd
3
+ size 377649057
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c4712b520abc1536421c411c214980d15f7ee55b49d90fa0a792055a41ae2904
3
+ size 4027