8688chris commited on
Commit
f34e4d4
·
verified ·
1 Parent(s): ca5fd39

Model save

Browse files
Files changed (5) hide show
  1. README.md +40 -41
  2. config.json +8 -8
  3. model.safetensors +2 -2
  4. tokenizer_config.json +1 -2
  5. training_args.bin +2 -2
README.md CHANGED
@@ -1,7 +1,6 @@
1
  ---
2
- library_name: transformers
3
  license: apache-2.0
4
- base_model: facebook/wav2vec2-base-960h
5
  tags:
6
  - generated_from_trainer
7
  metrics:
@@ -16,10 +15,10 @@ should probably proofread and complete it, then remove this comment. -->
16
 
17
  # Helldivers2ASR_V3
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: 46.4982
22
- - Wer: 0.0197
23
 
24
  ## Model description
25
 
@@ -38,11 +37,11 @@ More information needed
38
  ### Training hyperparameters
39
 
40
  The following hyperparameters were used during training:
41
- - learning_rate: 7e-05
42
- - train_batch_size: 32
43
- - eval_batch_size: 32
44
  - seed: 42
45
- - optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
46
  - lr_scheduler_type: constant
47
  - num_epochs: 30
48
  - mixed_precision_training: Native AMP
@@ -51,41 +50,41 @@ The following hyperparameters were used during training:
51
 
52
  | Training Loss | Epoch | Step | Validation Loss | Wer |
53
  |:-------------:|:-----:|:----:|:---------------:|:------:|
54
- | 1568.7385 | 1.0 | 164 | 771.7803 | 0.3712 |
55
- | 1068.8207 | 2.0 | 328 | 590.2993 | 0.2698 |
56
- | 860.8321 | 3.0 | 492 | 448.1273 | 0.2125 |
57
- | 747.6548 | 4.0 | 656 | 366.1845 | 0.1805 |
58
- | 631.9949 | 5.0 | 820 | 274.8642 | 0.1421 |
59
- | 571.2026 | 6.0 | 984 | 241.3273 | 0.1242 |
60
- | 509.5936 | 7.0 | 1148 | 216.0463 | 0.1093 |
61
- | 435.6571 | 8.0 | 1312 | 176.9057 | 0.0875 |
62
- | 411.174 | 9.0 | 1476 | 165.9831 | 0.0833 |
63
- | 364.6828 | 10.0 | 1640 | 138.6788 | 0.0712 |
64
- | 345.6405 | 11.0 | 1804 | 136.7265 | 0.0675 |
65
- | 315.1662 | 12.0 | 1968 | 114.3673 | 0.0570 |
66
- | 284.2125 | 13.0 | 2132 | 98.1911 | 0.0520 |
67
- | 278.4242 | 14.0 | 2296 | 94.5904 | 0.0457 |
68
- | 238.8668 | 15.0 | 2460 | 86.1799 | 0.0465 |
69
- | 235.9656 | 16.0 | 2624 | 79.9327 | 0.0444 |
70
- | 224.2061 | 17.0 | 2788 | 66.6601 | 0.0331 |
71
- | 209.7441 | 18.0 | 2952 | 61.5226 | 0.0307 |
72
- | 197.9857 | 19.0 | 3116 | 59.2185 | 0.0286 |
73
- | 189.3057 | 20.0 | 3280 | 55.8568 | 0.0278 |
74
- | 201.0874 | 21.0 | 3444 | 58.7346 | 0.0268 |
75
- | 152.8864 | 22.0 | 3608 | 60.1501 | 0.0226 |
76
- | 161.5389 | 23.0 | 3772 | 61.3434 | 0.0273 |
77
- | 151.5226 | 24.0 | 3936 | 46.0442 | 0.0208 |
78
- | 154.1182 | 25.0 | 4100 | 49.2783 | 0.0200 |
79
- | 157.7021 | 26.0 | 4264 | 39.4705 | 0.0202 |
80
- | 143.6331 | 27.0 | 4428 | 44.7556 | 0.0234 |
81
- | 136.6361 | 28.0 | 4592 | 39.5889 | 0.0192 |
82
- | 133.5365 | 29.0 | 4756 | 42.7637 | 0.0173 |
83
- | 117.5687 | 30.0 | 4920 | 46.4982 | 0.0197 |
84
 
85
 
86
  ### Framework versions
87
 
88
- - Transformers 4.57.3
89
  - Pytorch 2.5.1+cu121
90
  - Datasets 3.6.0
91
- - Tokenizers 0.22.1
 
1
  ---
 
2
  license: apache-2.0
3
+ base_model: facebook/wav2vec2-large-960h
4
  tags:
5
  - generated_from_trainer
6
  metrics:
 
15
 
16
  # Helldivers2ASR_V3
17
 
18
+ This model is a fine-tuned version of [facebook/wav2vec2-large-960h](https://huggingface.co/facebook/wav2vec2-large-960h) on an unknown dataset.
19
  It achieves the following results on the evaluation set:
20
+ - Loss: 100.2962
21
+ - Wer: 0.2096
22
 
23
  ## Model description
24
 
 
37
  ### Training hyperparameters
38
 
39
  The following hyperparameters were used during training:
40
+ - learning_rate: 0.0005
41
+ - train_batch_size: 16
42
+ - eval_batch_size: 16
43
  - seed: 42
44
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
45
  - lr_scheduler_type: constant
46
  - num_epochs: 30
47
  - mixed_precision_training: Native AMP
 
50
 
51
  | Training Loss | Epoch | Step | Validation Loss | Wer |
52
  |:-------------:|:-----:|:----:|:---------------:|:------:|
53
+ | 1044.7378 | 1.0 | 328 | 674.2581 | 0.7134 |
54
+ | 788.3286 | 2.0 | 656 | 551.7861 | 0.5443 |
55
+ | 686.6791 | 3.0 | 984 | 475.5942 | 0.4852 |
56
+ | 630.9951 | 4.0 | 1312 | 385.9394 | 0.4615 |
57
+ | 591.5513 | 5.0 | 1640 | 375.8221 | 0.4019 |
58
+ | 536.7542 | 6.0 | 1968 | 359.7422 | 0.4190 |
59
+ | 503.6451 | 7.0 | 2296 | 351.6354 | 0.3961 |
60
+ | 469.5973 | 8.0 | 2624 | 303.6820 | 0.3580 |
61
+ | 437.5308 | 9.0 | 2952 | 242.7078 | 0.3969 |
62
+ | 422.3093 | 10.0 | 3280 | 271.1854 | 0.3764 |
63
+ | 409.4765 | 11.0 | 3608 | 281.5080 | 0.3541 |
64
+ | 384.7355 | 12.0 | 3936 | 232.6271 | 0.3331 |
65
+ | 368.876 | 13.0 | 4264 | 205.0227 | 0.3607 |
66
+ | 359.576 | 14.0 | 4592 | 206.7654 | 0.3628 |
67
+ | 347.9149 | 15.0 | 4920 | 203.3771 | 0.3131 |
68
+ | 324.0592 | 16.0 | 5248 | 195.5896 | 0.3134 |
69
+ | 313.765 | 17.0 | 5576 | 187.8961 | 0.2963 |
70
+ | 308.2892 | 18.0 | 5904 | 175.5116 | 0.2926 |
71
+ | 298.7906 | 19.0 | 6232 | 204.8492 | 0.2947 |
72
+ | 276.8503 | 20.0 | 6560 | 174.5410 | 0.2763 |
73
+ | 273.6627 | 21.0 | 6888 | 169.8062 | 0.2813 |
74
+ | 262.0091 | 22.0 | 7216 | 151.0183 | 0.2493 |
75
+ | 256.9988 | 23.0 | 7544 | 146.9712 | 0.2522 |
76
+ | 253.0775 | 24.0 | 7872 | 138.4501 | 0.2474 |
77
+ | 250.4051 | 25.0 | 8200 | 107.8654 | 0.2112 |
78
+ | 247.1768 | 26.0 | 8528 | 113.1651 | 0.2062 |
79
+ | 230.0392 | 27.0 | 8856 | 138.3462 | 0.2564 |
80
+ | 232.4302 | 28.0 | 9184 | 128.4431 | 0.2469 |
81
+ | 225.1965 | 29.0 | 9512 | 108.4411 | 0.2661 |
82
+ | 220.6905 | 30.0 | 9840 | 100.2962 | 0.2096 |
83
 
84
 
85
  ### Framework versions
86
 
87
+ - Transformers 4.44.0
88
  - Pytorch 2.5.1+cu121
89
  - Datasets 3.6.0
90
+ - Tokenizers 0.19.1
config.json CHANGED
@@ -1,4 +1,5 @@
1
  {
 
2
  "activation_dropout": 0.1,
3
  "adapter_attn_dim": null,
4
  "adapter_kernel_size": 3,
@@ -45,7 +46,6 @@
45
  "ctc_zero_infinity": false,
46
  "diversity_loss_weight": 0.1,
47
  "do_stable_layer_norm": false,
48
- "dtype": "float32",
49
  "eos_token_id": 2,
50
  "feat_extract_activation": "gelu",
51
  "feat_extract_dropout": 0.0,
@@ -53,13 +53,12 @@
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,
62
- "intermediate_size": 3072,
63
  "layer_norm_eps": 1e-05,
64
  "layerdrop": 0.1,
65
  "mask_feature_length": 10,
@@ -70,15 +69,15 @@
70
  "mask_time_prob": 0.05,
71
  "model_type": "wav2vec2",
72
  "num_adapter_layers": 3,
73
- "num_attention_heads": 12,
74
  "num_codevector_groups": 2,
75
  "num_codevectors_per_group": 320,
76
  "num_conv_pos_embedding_groups": 16,
77
  "num_conv_pos_embeddings": 128,
78
  "num_feat_extract_layers": 7,
79
- "num_hidden_layers": 12,
80
  "num_negatives": 100,
81
- "output_hidden_size": 768,
82
  "pad_token_id": 0,
83
  "proj_codevector_dim": 256,
84
  "tdnn_dilation": [
@@ -102,7 +101,8 @@
102
  1,
103
  1
104
  ],
105
- "transformers_version": "4.57.3",
 
106
  "use_weighted_layer_sum": false,
107
  "vocab_size": 32,
108
  "xvector_output_dim": 512
 
1
  {
2
+ "_name_or_path": "facebook/wav2vec2-large-960h",
3
  "activation_dropout": 0.1,
4
  "adapter_attn_dim": null,
5
  "adapter_kernel_size": 3,
 
46
  "ctc_zero_infinity": false,
47
  "diversity_loss_weight": 0.1,
48
  "do_stable_layer_norm": false,
 
49
  "eos_token_id": 2,
50
  "feat_extract_activation": "gelu",
51
  "feat_extract_dropout": 0.0,
 
53
  "feat_proj_dropout": 0.1,
54
  "feat_quantizer_dropout": 0.0,
55
  "final_dropout": 0.15,
 
56
  "hidden_act": "gelu",
57
  "hidden_dropout": 0.15,
58
  "hidden_dropout_prob": 0.1,
59
+ "hidden_size": 1024,
60
  "initializer_range": 0.02,
61
+ "intermediate_size": 4096,
62
  "layer_norm_eps": 1e-05,
63
  "layerdrop": 0.1,
64
  "mask_feature_length": 10,
 
69
  "mask_time_prob": 0.05,
70
  "model_type": "wav2vec2",
71
  "num_adapter_layers": 3,
72
+ "num_attention_heads": 16,
73
  "num_codevector_groups": 2,
74
  "num_codevectors_per_group": 320,
75
  "num_conv_pos_embedding_groups": 16,
76
  "num_conv_pos_embeddings": 128,
77
  "num_feat_extract_layers": 7,
78
+ "num_hidden_layers": 24,
79
  "num_negatives": 100,
80
+ "output_hidden_size": 1024,
81
  "pad_token_id": 0,
82
  "proj_codevector_dim": 256,
83
  "tdnn_dilation": [
 
101
  1,
102
  1
103
  ],
104
+ "torch_dtype": "float32",
105
+ "transformers_version": "4.44.0",
106
  "use_weighted_layer_sum": false,
107
  "vocab_size": 32,
108
  "xvector_output_dim": 512
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:50f75135f3205e071dffd7a5c9092a6319d6153256e8171dd3951b85377acaa1
3
- size 377611120
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:cf39c3bad3692b8ada5e3dbea6704a07b1768bafeeb113bc0fc2160f0feba24a
3
+ size 1261897368
tokenizer_config.json CHANGED
@@ -34,11 +34,10 @@
34
  }
35
  },
36
  "bos_token": "<s>",
37
- "clean_up_tokenization_spaces": false,
38
  "do_lower_case": false,
39
  "do_normalize": true,
40
  "eos_token": "</s>",
41
- "extra_special_tokens": {},
42
  "model_max_length": 1000000000000000019884624838656,
43
  "pad_token": "<pad>",
44
  "processor_class": "Wav2Vec2Processor",
 
34
  }
35
  },
36
  "bos_token": "<s>",
37
+ "clean_up_tokenization_spaces": true,
38
  "do_lower_case": false,
39
  "do_normalize": true,
40
  "eos_token": "</s>",
 
41
  "model_max_length": 1000000000000000019884624838656,
42
  "pad_token": "<pad>",
43
  "processor_class": "Wav2Vec2Processor",
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:bb8d8ecf665cf5a646bf6c6c5ea1422b8f2ac7fd9a95644e4e45fa189a41e155
3
- size 5432
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4c6e5da415c451c4bb69a5677f24ce26f9e584b30080767c1bfc04bf9fccee97
3
+ size 5176