jialicheng commited on
Commit
f57d706
1 Parent(s): 345c175

Upload folder using huggingface_hub

Browse files
README.md ADDED
@@ -0,0 +1,86 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model: openai/whisper-tiny
4
+ tags:
5
+ - audio-classification
6
+ - generated_from_trainer
7
+ datasets:
8
+ - superb
9
+ metrics:
10
+ - accuracy
11
+ model-index:
12
+ - name: superb_ks_42
13
+ results:
14
+ - task:
15
+ name: Audio Classification
16
+ type: audio-classification
17
+ dataset:
18
+ name: superb
19
+ type: superb
20
+ config: ks
21
+ split: validation
22
+ args: ks
23
+ metrics:
24
+ - name: Accuracy
25
+ type: accuracy
26
+ value: 0.9801412180052956
27
+ ---
28
+
29
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
30
+ should probably proofread and complete it, then remove this comment. -->
31
+
32
+ # superb_ks_42
33
+
34
+ This model is a fine-tuned version of [openai/whisper-tiny](https://huggingface.co/openai/whisper-tiny) on the superb dataset.
35
+ It achieves the following results on the evaluation set:
36
+ - Loss: 0.1434
37
+ - Accuracy: 0.9801
38
+
39
+ ## Model description
40
+
41
+ More information needed
42
+
43
+ ## Intended uses & limitations
44
+
45
+ More information needed
46
+
47
+ ## Training and evaluation data
48
+
49
+ More information needed
50
+
51
+ ## Training procedure
52
+
53
+ ### Training hyperparameters
54
+
55
+ The following hyperparameters were used during training:
56
+ - learning_rate: 5e-05
57
+ - train_batch_size: 32
58
+ - eval_batch_size: 4
59
+ - seed: 42
60
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
61
+ - lr_scheduler_type: linear
62
+ - lr_scheduler_warmup_ratio: 0.1
63
+ - num_epochs: 10
64
+
65
+ ### Training results
66
+
67
+ | Training Loss | Epoch | Step | Validation Loss | Accuracy |
68
+ |:-------------:|:-----:|:-----:|:---------------:|:--------:|
69
+ | 0.8486 | 1.0 | 1597 | 0.2380 | 0.9385 |
70
+ | 0.0986 | 2.0 | 3194 | 0.1777 | 0.9598 |
71
+ | 0.0773 | 3.0 | 4791 | 0.1249 | 0.9738 |
72
+ | 0.0532 | 4.0 | 6388 | 0.1078 | 0.9782 |
73
+ | 0.0472 | 5.0 | 7985 | 0.1258 | 0.9766 |
74
+ | 0.0322 | 6.0 | 9582 | 0.1365 | 0.9772 |
75
+ | 0.0258 | 7.0 | 11179 | 0.1338 | 0.9798 |
76
+ | 0.0231 | 8.0 | 12776 | 0.1447 | 0.9796 |
77
+ | 0.0171 | 9.0 | 14373 | 0.1435 | 0.9797 |
78
+ | 0.0142 | 10.0 | 15970 | 0.1434 | 0.9801 |
79
+
80
+
81
+ ### Framework versions
82
+
83
+ - Transformers 4.40.1
84
+ - Pytorch 2.3.0+cu121
85
+ - Datasets 2.19.0
86
+ - Tokenizers 0.19.1
all_results.json ADDED
@@ -0,0 +1,19 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 10.0,
3
+ "eval_accuracy": 0.9801412180052956,
4
+ "eval_loss": 0.14339740574359894,
5
+ "eval_runtime": 25.5075,
6
+ "eval_samples_per_second": 266.51,
7
+ "eval_steps_per_second": 1.059,
8
+ "test_accuracy": 0.9425511197663097,
9
+ "test_loss": 0.35165315866470337,
10
+ "test_runtime": 16.1313,
11
+ "test_samples_per_second": 190.995,
12
+ "test_steps_per_second": 0.806,
13
+ "total_flos": 5.6903389435008e+18,
14
+ "train_accuracy": 0.9971816651661644,
15
+ "train_loss": 0.009092943742871284,
16
+ "train_runtime": 118.8228,
17
+ "train_samples_per_second": 430.002,
18
+ "train_steps_per_second": 107.505
19
+ }
config.json ADDED
@@ -0,0 +1,183 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "openai/whisper-tiny",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
+ "architectures": [
7
+ "WhisperForAudioClassification"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "begin_suppress_tokens": [
11
+ 220,
12
+ 50257
13
+ ],
14
+ "bos_token_id": 50257,
15
+ "classifier_proj_size": 256,
16
+ "d_model": 384,
17
+ "decoder_attention_heads": 6,
18
+ "decoder_ffn_dim": 1536,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 4,
21
+ "decoder_start_token_id": 50258,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 6,
24
+ "encoder_ffn_dim": 1536,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 4,
27
+ "eos_token_id": 50257,
28
+ "finetuning_task": "audio-classification",
29
+ "forced_decoder_ids": [
30
+ [
31
+ 1,
32
+ 50259
33
+ ],
34
+ [
35
+ 2,
36
+ 50359
37
+ ],
38
+ [
39
+ 3,
40
+ 50363
41
+ ]
42
+ ],
43
+ "id2label": {
44
+ "0": "yes",
45
+ "1": "no",
46
+ "10": "_silence_",
47
+ "11": "_unknown_",
48
+ "2": "up",
49
+ "3": "down",
50
+ "4": "left",
51
+ "5": "right",
52
+ "6": "on",
53
+ "7": "off",
54
+ "8": "stop",
55
+ "9": "go"
56
+ },
57
+ "init_std": 0.02,
58
+ "is_encoder_decoder": true,
59
+ "label2id": {
60
+ "_silence_": "10",
61
+ "_unknown_": "11",
62
+ "down": "3",
63
+ "go": "9",
64
+ "left": "4",
65
+ "no": "1",
66
+ "off": "7",
67
+ "on": "6",
68
+ "right": "5",
69
+ "stop": "8",
70
+ "up": "2",
71
+ "yes": "0"
72
+ },
73
+ "mask_feature_length": 10,
74
+ "mask_feature_min_masks": 0,
75
+ "mask_feature_prob": 0.0,
76
+ "mask_time_length": 10,
77
+ "mask_time_min_masks": 2,
78
+ "mask_time_prob": 0.05,
79
+ "max_length": 448,
80
+ "max_source_positions": 1500,
81
+ "max_target_positions": 448,
82
+ "median_filter_width": 7,
83
+ "model_type": "whisper",
84
+ "num_hidden_layers": 4,
85
+ "num_mel_bins": 80,
86
+ "pad_token_id": 50257,
87
+ "scale_embedding": false,
88
+ "suppress_tokens": [
89
+ 1,
90
+ 2,
91
+ 7,
92
+ 8,
93
+ 9,
94
+ 10,
95
+ 14,
96
+ 25,
97
+ 26,
98
+ 27,
99
+ 28,
100
+ 29,
101
+ 31,
102
+ 58,
103
+ 59,
104
+ 60,
105
+ 61,
106
+ 62,
107
+ 63,
108
+ 90,
109
+ 91,
110
+ 92,
111
+ 93,
112
+ 359,
113
+ 503,
114
+ 522,
115
+ 542,
116
+ 873,
117
+ 893,
118
+ 902,
119
+ 918,
120
+ 922,
121
+ 931,
122
+ 1350,
123
+ 1853,
124
+ 1982,
125
+ 2460,
126
+ 2627,
127
+ 3246,
128
+ 3253,
129
+ 3268,
130
+ 3536,
131
+ 3846,
132
+ 3961,
133
+ 4183,
134
+ 4667,
135
+ 6585,
136
+ 6647,
137
+ 7273,
138
+ 9061,
139
+ 9383,
140
+ 10428,
141
+ 10929,
142
+ 11938,
143
+ 12033,
144
+ 12331,
145
+ 12562,
146
+ 13793,
147
+ 14157,
148
+ 14635,
149
+ 15265,
150
+ 15618,
151
+ 16553,
152
+ 16604,
153
+ 18362,
154
+ 18956,
155
+ 20075,
156
+ 21675,
157
+ 22520,
158
+ 26130,
159
+ 26161,
160
+ 26435,
161
+ 28279,
162
+ 29464,
163
+ 31650,
164
+ 32302,
165
+ 32470,
166
+ 36865,
167
+ 42863,
168
+ 47425,
169
+ 49870,
170
+ 50254,
171
+ 50258,
172
+ 50358,
173
+ 50359,
174
+ 50360,
175
+ 50361,
176
+ 50362
177
+ ],
178
+ "torch_dtype": "float32",
179
+ "transformers_version": "4.40.1",
180
+ "use_cache": true,
181
+ "use_weighted_layer_sum": false,
182
+ "vocab_size": 51865
183
+ }
eval_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 10.0,
3
+ "eval_accuracy": 0.9801412180052956,
4
+ "eval_loss": 0.14339740574359894,
5
+ "eval_runtime": 25.5075,
6
+ "eval_samples_per_second": 266.51,
7
+ "eval_steps_per_second": 1.059
8
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3c086863c912b31cc9c131b86767909c602ac015e739a312ea613fba40dc67a0
3
+ size 33247448
pred_logit_eval.npy ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aefafe3da4ef4fcff4ec4f263f97deffd0141b159979d057295dea08f4c1575d
3
+ size 326432
pred_logit_test.npy ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5332a4400c78ff7066be6eb5200f74d446913ad2ac75ce803aef28da62f89464
3
+ size 148016
pred_logit_train.npy ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3b7d7f26ba2c46ecedccdde45f4596e104bc624bf4c50ffe8f32f8b8600e84d3
3
+ size 2452640
preprocessor_config.json ADDED
@@ -0,0 +1,14 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "chunk_length": 30,
3
+ "feature_extractor_type": "WhisperFeatureExtractor",
4
+ "feature_size": 80,
5
+ "hop_length": 160,
6
+ "n_fft": 400,
7
+ "n_samples": 480000,
8
+ "nb_max_frames": 3000,
9
+ "padding_side": "right",
10
+ "padding_value": 0.0,
11
+ "processor_class": "WhisperProcessor",
12
+ "return_attention_mask": false,
13
+ "sampling_rate": 16000
14
+ }
test_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 10.0,
3
+ "test_accuracy": 0.9425511197663097,
4
+ "test_loss": 0.35165315866470337,
5
+ "test_runtime": 16.1313,
6
+ "test_samples_per_second": 190.995,
7
+ "test_steps_per_second": 0.806
8
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 10.0,
3
+ "train_accuracy": 0.9971816651661644,
4
+ "train_loss": 0.009092943742871284,
5
+ "train_runtime": 118.8228,
6
+ "train_samples_per_second": 430.002,
7
+ "train_steps_per_second": 107.505
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,225 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.9801412180052956,
3
+ "best_model_checkpoint": "audio/train/checkpoint/whisper-tiny/superb_ks_42/checkpoint-15970",
4
+ "epoch": 10.0,
5
+ "eval_steps": 500,
6
+ "global_step": 15970,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.6261740763932373,
13
+ "grad_norm": 17.53887176513672,
14
+ "learning_rate": 3.1308703819661866e-05,
15
+ "loss": 0.8486,
16
+ "step": 1000
17
+ },
18
+ {
19
+ "epoch": 1.0,
20
+ "eval_accuracy": 0.9385113268608414,
21
+ "eval_loss": 0.23804208636283875,
22
+ "eval_runtime": 29.9994,
23
+ "eval_samples_per_second": 226.605,
24
+ "eval_steps_per_second": 0.9,
25
+ "step": 1597
26
+ },
27
+ {
28
+ "epoch": 1.2523481527864746,
29
+ "grad_norm": 0.1602136492729187,
30
+ "learning_rate": 4.859806581785292e-05,
31
+ "loss": 0.1349,
32
+ "step": 2000
33
+ },
34
+ {
35
+ "epoch": 1.878522229179712,
36
+ "grad_norm": 0.37793028354644775,
37
+ "learning_rate": 4.51193209490016e-05,
38
+ "loss": 0.0986,
39
+ "step": 3000
40
+ },
41
+ {
42
+ "epoch": 2.0,
43
+ "eval_accuracy": 0.9598411297440423,
44
+ "eval_loss": 0.1777295619249344,
45
+ "eval_runtime": 27.7034,
46
+ "eval_samples_per_second": 245.385,
47
+ "eval_steps_per_second": 0.975,
48
+ "step": 3194
49
+ },
50
+ {
51
+ "epoch": 2.504696305572949,
52
+ "grad_norm": 15.802234649658203,
53
+ "learning_rate": 4.1640576080150286e-05,
54
+ "loss": 0.0773,
55
+ "step": 4000
56
+ },
57
+ {
58
+ "epoch": 3.0,
59
+ "eval_accuracy": 0.9738158281847602,
60
+ "eval_loss": 0.12492171674966812,
61
+ "eval_runtime": 28.5966,
62
+ "eval_samples_per_second": 237.721,
63
+ "eval_steps_per_second": 0.944,
64
+ "step": 4791
65
+ },
66
+ {
67
+ "epoch": 3.1308703819661865,
68
+ "grad_norm": 10.594197273254395,
69
+ "learning_rate": 3.8161831211298965e-05,
70
+ "loss": 0.0669,
71
+ "step": 5000
72
+ },
73
+ {
74
+ "epoch": 3.757044458359424,
75
+ "grad_norm": 0.004856509622186422,
76
+ "learning_rate": 3.468308634244764e-05,
77
+ "loss": 0.0532,
78
+ "step": 6000
79
+ },
80
+ {
81
+ "epoch": 4.0,
82
+ "eval_accuracy": 0.9782288908502501,
83
+ "eval_loss": 0.1077924370765686,
84
+ "eval_runtime": 30.2849,
85
+ "eval_samples_per_second": 224.468,
86
+ "eval_steps_per_second": 0.892,
87
+ "step": 6388
88
+ },
89
+ {
90
+ "epoch": 4.383218534752661,
91
+ "grad_norm": 0.8512524962425232,
92
+ "learning_rate": 3.120434147359633e-05,
93
+ "loss": 0.0472,
94
+ "step": 7000
95
+ },
96
+ {
97
+ "epoch": 5.0,
98
+ "eval_accuracy": 0.9766107678729038,
99
+ "eval_loss": 0.12581859529018402,
100
+ "eval_runtime": 25.8895,
101
+ "eval_samples_per_second": 262.577,
102
+ "eval_steps_per_second": 1.043,
103
+ "step": 7985
104
+ },
105
+ {
106
+ "epoch": 5.009392611145898,
107
+ "grad_norm": 0.2138238549232483,
108
+ "learning_rate": 2.7725596604745007e-05,
109
+ "loss": 0.0411,
110
+ "step": 8000
111
+ },
112
+ {
113
+ "epoch": 5.635566687539136,
114
+ "grad_norm": 0.004156901966780424,
115
+ "learning_rate": 2.4246851735893692e-05,
116
+ "loss": 0.0322,
117
+ "step": 9000
118
+ },
119
+ {
120
+ "epoch": 6.0,
121
+ "eval_accuracy": 0.9771991762283024,
122
+ "eval_loss": 0.13651490211486816,
123
+ "eval_runtime": 27.6161,
124
+ "eval_samples_per_second": 246.161,
125
+ "eval_steps_per_second": 0.978,
126
+ "step": 9582
127
+ },
128
+ {
129
+ "epoch": 6.261740763932373,
130
+ "grad_norm": 1.517451286315918,
131
+ "learning_rate": 2.076810686704237e-05,
132
+ "loss": 0.0357,
133
+ "step": 10000
134
+ },
135
+ {
136
+ "epoch": 6.88791484032561,
137
+ "grad_norm": 0.0043836492113769054,
138
+ "learning_rate": 1.7289361998191053e-05,
139
+ "loss": 0.0258,
140
+ "step": 11000
141
+ },
142
+ {
143
+ "epoch": 7.0,
144
+ "eval_accuracy": 0.9798470138275963,
145
+ "eval_loss": 0.13380160927772522,
146
+ "eval_runtime": 28.81,
147
+ "eval_samples_per_second": 235.96,
148
+ "eval_steps_per_second": 0.937,
149
+ "step": 11179
150
+ },
151
+ {
152
+ "epoch": 7.514088916718848,
153
+ "grad_norm": 0.0014675065176561475,
154
+ "learning_rate": 1.3810617129339735e-05,
155
+ "loss": 0.0231,
156
+ "step": 12000
157
+ },
158
+ {
159
+ "epoch": 8.0,
160
+ "eval_accuracy": 0.979552809649897,
161
+ "eval_loss": 0.14468975365161896,
162
+ "eval_runtime": 26.2171,
163
+ "eval_samples_per_second": 259.296,
164
+ "eval_steps_per_second": 1.03,
165
+ "step": 12776
166
+ },
167
+ {
168
+ "epoch": 8.140262993112085,
169
+ "grad_norm": 3.176596164703369,
170
+ "learning_rate": 1.0331872260488417e-05,
171
+ "loss": 0.021,
172
+ "step": 13000
173
+ },
174
+ {
175
+ "epoch": 8.766437069505322,
176
+ "grad_norm": 0.0021598339080810547,
177
+ "learning_rate": 6.853127391637097e-06,
178
+ "loss": 0.0171,
179
+ "step": 14000
180
+ },
181
+ {
182
+ "epoch": 9.0,
183
+ "eval_accuracy": 0.9796999117387467,
184
+ "eval_loss": 0.14350397884845734,
185
+ "eval_runtime": 25.9468,
186
+ "eval_samples_per_second": 261.997,
187
+ "eval_steps_per_second": 1.041,
188
+ "step": 14373
189
+ },
190
+ {
191
+ "epoch": 9.39261114589856,
192
+ "grad_norm": 0.0006765660946257412,
193
+ "learning_rate": 3.3743825227857787e-06,
194
+ "loss": 0.0142,
195
+ "step": 15000
196
+ },
197
+ {
198
+ "epoch": 10.0,
199
+ "eval_accuracy": 0.9801412180052956,
200
+ "eval_loss": 0.14339740574359894,
201
+ "eval_runtime": 26.2356,
202
+ "eval_samples_per_second": 259.114,
203
+ "eval_steps_per_second": 1.029,
204
+ "step": 15970
205
+ },
206
+ {
207
+ "epoch": 10.0,
208
+ "step": 15970,
209
+ "total_flos": 5.6903389435008e+18,
210
+ "train_loss": 0.09692646174708529,
211
+ "train_runtime": 3764.7682,
212
+ "train_samples_per_second": 135.716,
213
+ "train_steps_per_second": 4.242
214
+ }
215
+ ],
216
+ "logging_steps": 1000,
217
+ "max_steps": 15970,
218
+ "num_input_tokens_seen": 0,
219
+ "num_train_epochs": 10,
220
+ "save_steps": 500,
221
+ "total_flos": 5.6903389435008e+18,
222
+ "train_batch_size": 32,
223
+ "trial_name": null,
224
+ "trial_params": null
225
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:734b5fa74bca8a553154a8842516df32fbe29da497c3bb2c0098c57b2454afd7
3
+ size 5112