Hugging Face's logo

8688chris
/
Helldivers2ASR

Automatic Speech Recognition
Transformers
Safetensors
wav2vec2
Generated from Trainer
Model card Files
xet
Community
8688chris commited on
Commit
9326639
·
verified ·
1 Parent(s): 48930bb

Model save

Browse files
Files changed (4) hide show
  1. README.md +59 -57
  2. config.json +3 -3
  3. model.safetensors +1 -1
  4. training_args.bin +1 -1
README.md CHANGED
@@ -18,9 +18,9 @@ should probably proofread and complete it, then remove this comment. -->
18
 
19
  This model is a fine-tuned version of [facebook/wav2vec2-base-960h](https://huggingface.co/facebook/wav2vec2-base-960h) on an unknown dataset.
20
  It achieves the following results on the evaluation set:
21
- - Loss: 100.7699
22
- - Wer: 0.1916
23
- - Cer: 0.8008
24
 
25
  ## Model description
26
 
@@ -39,13 +39,15 @@ More information needed
39
  ### Training hyperparameters
40
 
41
  The following hyperparameters were used during training:
42
- - learning_rate: 0.0001
43
- - train_batch_size: 32
44
- - eval_batch_size: 32
45
  - seed: 42
 
 
46
  - optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
47
  - lr_scheduler_type: linear
48
- - lr_scheduler_warmup_steps: 50
49
  - num_epochs: 50
50
  - mixed_precision_training: Native AMP
51
 
@@ -53,56 +55,56 @@ The following hyperparameters were used during training:
53
 
54
  | Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
55
  |:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
56
- | 1776.0296 | 1.0 | 148 | 994.9070 | 0.5253 | 0.8441 |
57
- | 1258.6077 | 2.0 | 296 | 759.4000 | 0.4627 | 0.8343 |
58
- | 1088.4589 | 3.0 | 444 | 652.5190 | 0.4167 | 0.8290 |
59
- | 942.0106 | 4.0 | 592 | 557.9450 | 0.3848 | 0.8242 |
60
- | 870.7333 | 5.0 | 740 | 511.7816 | 0.3518 | 0.8209 |
61
- | 786.5758 | 6.0 | 888 | 454.8054 | 0.3420 | 0.8184 |
62
- | 751.542 | 7.0 | 1036 | 400.5589 | 0.3274 | 0.8170 |
63
- | 691.6483 | 8.0 | 1184 | 370.5620 | 0.3110 | 0.8148 |
64
- | 650.9937 | 9.0 | 1332 | 350.6898 | 0.2972 | 0.8130 |
65
- | 600.0241 | 10.0 | 1480 | 333.2052 | 0.2938 | 0.8121 |
66
- | 591.0826 | 11.0 | 1628 | 310.2686 | 0.2829 | 0.8107 |
67
- | 567.3793 | 12.0 | 1776 | 290.4415 | 0.2734 | 0.8100 |
68
- | 540.7039 | 13.0 | 1924 | 285.3511 | 0.2702 | 0.8096 |
69
- | 527.8851 | 14.0 | 2072 | 254.9483 | 0.2613 | 0.8086 |
70
- | 479.033 | 15.0 | 2220 | 233.2339 | 0.2519 | 0.8078 |
71
- | 462.2839 | 16.0 | 2368 | 234.3993 | 0.2536 | 0.8078 |
72
- | 440.3757 | 17.0 | 2516 | 215.7443 | 0.2427 | 0.8065 |
73
- | 403.512 | 18.0 | 2664 | 198.2435 | 0.2427 | 0.8062 |
74
- | 416.854 | 19.0 | 2812 | 201.3992 | 0.2418 | 0.8060 |
75
- | 414.6919 | 20.0 | 2960 | 187.9181 | 0.2387 | 0.8055 |
76
- | 377.038 | 21.0 | 3108 | 186.3279 | 0.2292 | 0.8048 |
77
- | 361.3397 | 22.0 | 3256 | 174.6297 | 0.2263 | 0.8043 |
78
- | 370.364 | 23.0 | 3404 | 170.0526 | 0.2266 | 0.8040 |
79
- | 334.322 | 24.0 | 3552 | 160.5238 | 0.2220 | 0.8038 |
80
- | 371.9682 | 25.0 | 3700 | 156.7221 | 0.2186 | 0.8035 |
81
- | 319.2351 | 26.0 | 3848 | 141.5811 | 0.2174 | 0.8033 |
82
- | 321.3611 | 27.0 | 3996 | 155.3030 | 0.2171 | 0.8034 |
83
- | 293.7229 | 28.0 | 4144 | 155.8836 | 0.2140 | 0.8032 |
84
- | 295.0259 | 29.0 | 4292 | 150.5117 | 0.2145 | 0.8028 |
85
- | 326.6667 | 30.0 | 4440 | 132.2860 | 0.2117 | 0.8029 |
86
- | 304.087 | 31.0 | 4588 | 135.3334 | 0.2122 | 0.8027 |
87
- | 290.372 | 32.0 | 4736 | 127.9463 | 0.2059 | 0.8021 |
88
- | 300.9416 | 33.0 | 4884 | 124.8618 | 0.2028 | 0.8020 |
89
- | 268.9617 | 34.0 | 5032 | 121.5129 | 0.2019 | 0.8018 |
90
- | 303.5479 | 35.0 | 5180 | 116.0215 | 0.2048 | 0.8018 |
91
- | 255.0924 | 36.0 | 5328 | 121.9995 | 0.2010 | 0.8018 |
92
- | 241.0434 | 37.0 | 5476 | 116.4092 | 0.1987 | 0.8016 |
93
- | 265.054 | 38.0 | 5624 | 114.8559 | 0.2028 | 0.8018 |
94
- | 239.6707 | 39.0 | 5772 | 110.8761 | 0.1964 | 0.8014 |
95
- | 242.3196 | 40.0 | 5920 | 106.6581 | 0.1939 | 0.8011 |
96
- | 244.1651 | 41.0 | 6068 | 107.1305 | 0.1944 | 0.8011 |
97
- | 265.9968 | 42.0 | 6216 | 99.1424 | 0.1947 | 0.8009 |
98
- | 261.463 | 43.0 | 6364 | 103.3172 | 0.1927 | 0.8010 |
99
- | 232.4012 | 44.0 | 6512 | 109.4188 | 0.1933 | 0.8010 |
100
- | 238.1314 | 45.0 | 6660 | 100.4500 | 0.1918 | 0.8009 |
101
- | 229.5179 | 46.0 | 6808 | 99.3243 | 0.1933 | 0.8009 |
102
- | 230.4529 | 47.0 | 6956 | 95.2599 | 0.1916 | 0.8007 |
103
- | 248.9069 | 48.0 | 7104 | 96.6804 | 0.1901 | 0.8007 |
104
- | 242.4876 | 49.0 | 7252 | 98.2327 | 0.1916 | 0.8007 |
105
- | 241.9431 | 50.0 | 7400 | 100.7699 | 0.1916 | 0.8008 |
106
 
107
 
108
  ### Framework versions
 
18
 
19
  This model is a fine-tuned version of [facebook/wav2vec2-base-960h](https://huggingface.co/facebook/wav2vec2-base-960h) on an unknown dataset.
20
  It achieves the following results on the evaluation set:
21
+ - Loss: 45.7976
22
+ - Wer: 0.0486
23
+ - Cer: 0.0203
24
 
25
  ## Model description
26
 
 
39
  ### Training hyperparameters
40
 
41
  The following hyperparameters were used during training:
42
+ - learning_rate: 5e-05
43
+ - train_batch_size: 16
44
+ - eval_batch_size: 16
45
  - seed: 42
46
+ - gradient_accumulation_steps: 2
47
+ - total_train_batch_size: 32
48
  - optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
49
  - lr_scheduler_type: linear
50
+ - lr_scheduler_warmup_ratio: 0.1
51
  - num_epochs: 50
52
  - mixed_precision_training: Native AMP
53
 
 
55
 
56
  | Training Loss | Epoch | Step | Validation Loss | Wer | Cer |
57
  |:-------------:|:-----:|:----:|:---------------:|:------:|:------:|
58
+ | 1069.0904 | 1.0 | 111 | 690.8810 | 0.5789 | 0.3204 |
59
+ | 812.3311 | 2.0 | 222 | 524.6868 | 0.4769 | 0.2529 |
60
+ | 677.891 | 3.0 | 333 | 423.8824 | 0.4040 | 0.2043 |
61
+ | 584.6176 | 4.0 | 444 | 338.5651 | 0.3429 | 0.1709 |
62
+ | 510.1064 | 5.0 | 555 | 284.0197 | 0.3016 | 0.1467 |
63
+ | 462.5595 | 6.0 | 666 | 245.5092 | 0.2615 | 0.1233 |
64
+ | 398.9107 | 7.0 | 777 | 205.1819 | 0.2198 | 0.1036 |
65
+ | 375.3196 | 8.0 | 888 | 183.1224 | 0.2049 | 0.0943 |
66
+ | 338.0957 | 9.0 | 999 | 163.4431 | 0.1834 | 0.0832 |
67
+ | 305.4381 | 10.0 | 1110 | 158.5278 | 0.1636 | 0.0759 |
68
+ | 297.0845 | 11.0 | 1221 | 140.0732 | 0.1603 | 0.0715 |
69
+ | 280.9225 | 12.0 | 1332 | 128.1153 | 0.1453 | 0.0633 |
70
+ | 274.2178 | 13.0 | 1443 | 115.9890 | 0.1283 | 0.0581 |
71
+ | 238.5611 | 14.0 | 1554 | 112.8672 | 0.1271 | 0.0572 |
72
+ | 233.9152 | 15.0 | 1665 | 105.2606 | 0.1178 | 0.0517 |
73
+ | 222.8375 | 16.0 | 1776 | 105.7286 | 0.1093 | 0.0522 |
74
+ | 218.3437 | 17.0 | 1887 | 100.8668 | 0.1089 | 0.0480 |
75
+ | 208.0329 | 18.0 | 1998 | 96.2257 | 0.1020 | 0.0457 |
76
+ | 199.382 | 19.0 | 2109 | 85.2498 | 0.0939 | 0.0433 |
77
+ | 198.6175 | 20.0 | 2220 | 82.4100 | 0.0927 | 0.0425 |
78
+ | 173.409 | 21.0 | 2331 | 78.6151 | 0.0842 | 0.0377 |
79
+ | 168.3968 | 22.0 | 2442 | 79.7964 | 0.0830 | 0.0382 |
80
+ | 171.8005 | 23.0 | 2553 | 70.0593 | 0.0773 | 0.0336 |
81
+ | 157.5571 | 24.0 | 2664 | 67.0374 | 0.0713 | 0.0308 |
82
+ | 157.863 | 25.0 | 2775 | 61.6305 | 0.0684 | 0.0299 |
83
+ | 153.1922 | 26.0 | 2886 | 64.7148 | 0.0676 | 0.0305 |
84
+ | 158.7052 | 27.0 | 2997 | 59.4825 | 0.0676 | 0.0286 |
85
+ | 149.8696 | 28.0 | 3108 | 59.3228 | 0.0615 | 0.0290 |
86
+ | 145.0604 | 29.0 | 3219 | 57.5999 | 0.0595 | 0.0273 |
87
+ | 138.6555 | 30.0 | 3330 | 56.0322 | 0.0587 | 0.0273 |
88
+ | 145.2049 | 31.0 | 3441 | 51.0521 | 0.0559 | 0.0262 |
89
+ | 137.6945 | 32.0 | 3552 | 52.0388 | 0.0510 | 0.0236 |
90
+ | 130.1284 | 33.0 | 3663 | 51.2634 | 0.0587 | 0.0256 |
91
+ | 127.2604 | 34.0 | 3774 | 47.7200 | 0.0510 | 0.0239 |
92
+ | 124.2158 | 35.0 | 3885 | 47.9998 | 0.0530 | 0.0233 |
93
+ | 116.4625 | 36.0 | 3996 | 50.8454 | 0.0547 | 0.0236 |
94
+ | 125.1983 | 37.0 | 4107 | 47.3148 | 0.0490 | 0.0212 |
95
+ | 110.8833 | 38.0 | 4218 | 47.1674 | 0.0522 | 0.0229 |
96
+ | 109.9017 | 39.0 | 4329 | 46.9405 | 0.0506 | 0.0224 |
97
+ | 116.4361 | 40.0 | 4440 | 49.4927 | 0.0482 | 0.0221 |
98
+ | 117.1769 | 41.0 | 4551 | 46.7733 | 0.0474 | 0.0212 |
99
+ | 108.7644 | 42.0 | 4662 | 45.9081 | 0.0490 | 0.0209 |
100
+ | 117.5977 | 43.0 | 4773 | 43.4770 | 0.0462 | 0.0202 |
101
+ | 125.1272 | 44.0 | 4884 | 41.2925 | 0.0482 | 0.0206 |
102
+ | 110.6224 | 45.0 | 4995 | 47.2408 | 0.0498 | 0.0211 |
103
+ | 110.3162 | 46.0 | 5106 | 46.2208 | 0.0437 | 0.0196 |
104
+ | 107.2695 | 47.0 | 5217 | 44.8825 | 0.0433 | 0.0199 |
105
+ | 111.1748 | 48.0 | 5328 | 45.3059 | 0.0425 | 0.0191 |
106
+ | 108.1643 | 49.0 | 5439 | 45.5355 | 0.0449 | 0.0199 |
107
+ | 107.1299 | 50.0 | 5550 | 45.7976 | 0.0486 | 0.0203 |
108
 
109
 
110
  ### Framework versions
config.json CHANGED
@@ -8,7 +8,7 @@
8
  "architectures": [
9
  "Wav2Vec2ForCTC"
10
  ],
11
- "attention_dropout": 0.1,
12
  "bos_token_id": 1,
13
  "classifier_proj_size": 256,
14
  "codevector_dim": 256,
@@ -52,10 +52,10 @@
52
  "feat_extract_norm": "group",
53
  "feat_proj_dropout": 0.1,
54
  "feat_quantizer_dropout": 0.0,
55
- "final_dropout": 0.1,
56
  "gradient_checkpointing": false,
57
  "hidden_act": "gelu",
58
- "hidden_dropout": 0.1,
59
  "hidden_dropout_prob": 0.1,
60
  "hidden_size": 768,
61
  "initializer_range": 0.02,
 
8
  "architectures": [
9
  "Wav2Vec2ForCTC"
10
  ],
11
+ "attention_dropout": 0.15,
12
  "bos_token_id": 1,
13
  "classifier_proj_size": 256,
14
  "codevector_dim": 256,
 
52
  "feat_extract_norm": "group",
53
  "feat_proj_dropout": 0.1,
54
  "feat_quantizer_dropout": 0.0,
55
+ "final_dropout": 0.15,
56
  "gradient_checkpointing": false,
57
  "hidden_act": "gelu",
58
+ "hidden_dropout": 0.15,
59
  "hidden_dropout_prob": 0.1,
60
  "hidden_size": 768,
61
  "initializer_range": 0.02,
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8d27b649e89d6111b98a5bbddb6aaded8099c78bd5a0ed16d31639cc88b3d256
3
  size 377611120
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9d082a13dba083c29cf4ce1b363999139bd7b6f1ea6a66438d29c87b15527e05
3
  size 377611120
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:73990d8e0e7b9d2ea6cdcc7f3d1f15b3e18a9c297a24dd92bd29a6f11ef0f83c
3
  size 5432
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a57a6e8d1830acadd3a1d456c013d98f91a365e0dab0f2788f50f863ef47287c
3
  size 5432