Hugging Face's logo

fydhfzh
/
sewd_classifier

Audio Classification
Transformers
TensorBoard
Safetensors
sew-d
Generated from Trainer
Model card Files Metrics Community
fydhfzh commited on
Commit
92308a5
·
verified ·
1 Parent(s): 391590a

Training in progress, step 500

Browse files
Files changed (13) hide show
  1. model.safetensors +1 -1
  2. runs/Jun12_19-13-54_LAPTOP-1GID9RGH/events.out.tfevents.1718194435.LAPTOP-1GID9RGH.5996.0 +3 -0
  3. runs/Jun12_19-24-20_LAPTOP-1GID9RGH/events.out.tfevents.1718195061.LAPTOP-1GID9RGH.17516.0 +3 -0
  4. runs/Jun12_21-53-48_LAPTOP-1GID9RGH/events.out.tfevents.1718204029.LAPTOP-1GID9RGH.18328.0 +3 -0
  5. tmp-checkpoint-500/config.json +274 -0
  6. tmp-checkpoint-500/model.safetensors +3 -0
  7. tmp-checkpoint-500/optimizer.pt +3 -0
  8. tmp-checkpoint-500/preprocessor_config.json +9 -0
  9. tmp-checkpoint-500/rng_state.pth +3 -0
  10. tmp-checkpoint-500/scheduler.pt +3 -0
  11. tmp-checkpoint-500/trainer_state.json +158 -0
  12. tmp-checkpoint-500/training_args.bin +3 -0
  13. training_args.bin +1 -1
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:79ff77b47cc02c6ad91798d5a75fa4087ad43a70d3b11f9716973cb542460603
3
  size 96968812
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e37ee6672e4f4d1f5c287f6af0f282db7f9c77aa178b01992cbe08423408a09f
3
  size 96968812
runs/Jun12_19-13-54_LAPTOP-1GID9RGH/events.out.tfevents.1718194435.LAPTOP-1GID9RGH.5996.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c4917b3f9e3e011dff28cd817e5f84ac2e97f27db3c2424304e2e9fdacc56929
3
+ size 10405
runs/Jun12_19-24-20_LAPTOP-1GID9RGH/events.out.tfevents.1718195061.LAPTOP-1GID9RGH.17516.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f9fb5f83dc2eb00058413e54e8608dafa9a3577cea11e47549d34a8e22c9d150
3
+ size 13221
runs/Jun12_21-53-48_LAPTOP-1GID9RGH/events.out.tfevents.1718204029.LAPTOP-1GID9RGH.18328.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bbb112ca2be44ec7a762c4e1428d68ac62b80c1383c51c957b530aea8f2aeafc
3
+ size 14783
tmp-checkpoint-500/config.json ADDED
@@ -0,0 +1,274 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "asapp/sew-d-tiny-100k-ft-ls100h",
3
+ "activation_dropout": 0.1,
4
+ "apply_spec_augment": true,
5
+ "architectures": [
6
+ "SEWDForSequenceClassification"
7
+ ],
8
+ "attention_dropout": 0.0,
9
+ "bos_token_id": 1,
10
+ "classifier_proj_size": 256,
11
+ "conv_bias": false,
12
+ "conv_dim": [
13
+ 64,
14
+ 128,
15
+ 128,
16
+ 128,
17
+ 128,
18
+ 256,
19
+ 256,
20
+ 256,
21
+ 256,
22
+ 512,
23
+ 512,
24
+ 512,
25
+ 512
26
+ ],
27
+ "conv_kernel": [
28
+ 10,
29
+ 3,
30
+ 1,
31
+ 3,
32
+ 1,
33
+ 3,
34
+ 1,
35
+ 3,
36
+ 1,
37
+ 2,
38
+ 1,
39
+ 2,
40
+ 1
41
+ ],
42
+ "conv_stride": [
43
+ 5,
44
+ 2,
45
+ 1,
46
+ 2,
47
+ 1,
48
+ 2,
49
+ 1,
50
+ 2,
51
+ 1,
52
+ 2,
53
+ 1,
54
+ 2,
55
+ 1
56
+ ],
57
+ "ctc_loss_reduction": "mean",
58
+ "ctc_zero_infinity": false,
59
+ "eos_token_id": 2,
60
+ "feat_extract_activation": "gelu",
61
+ "feat_extract_norm": "group",
62
+ "feat_proj_dropout": 0.0,
63
+ "feature_extractor_type": "Wav2Vec2FeatureExtractor",
64
+ "feature_layer_norm_eps": 1e-05,
65
+ "final_dropout": 0.0,
66
+ "hidden_act": "gelu_python",
67
+ "hidden_dropout": 0.0,
68
+ "hidden_size": 384,
69
+ "id2label": {
70
+ "0": "LABEL_0",
71
+ "1": "LABEL_1",
72
+ "2": "LABEL_2",
73
+ "3": "LABEL_3",
74
+ "4": "LABEL_4",
75
+ "5": "LABEL_5",
76
+ "6": "LABEL_6",
77
+ "7": "LABEL_7",
78
+ "8": "LABEL_8",
79
+ "9": "LABEL_9",
80
+ "10": "LABEL_10",
81
+ "11": "LABEL_11",
82
+ "12": "LABEL_12",
83
+ "13": "LABEL_13",
84
+ "14": "LABEL_14",
85
+ "15": "LABEL_15",
86
+ "16": "LABEL_16",
87
+ "17": "LABEL_17",
88
+ "18": "LABEL_18",
89
+ "19": "LABEL_19",
90
+ "20": "LABEL_20",
91
+ "21": "LABEL_21",
92
+ "22": "LABEL_22",
93
+ "23": "LABEL_23",
94
+ "24": "LABEL_24",
95
+ "25": "LABEL_25",
96
+ "26": "LABEL_26",
97
+ "27": "LABEL_27",
98
+ "28": "LABEL_28",
99
+ "29": "LABEL_29",
100
+ "30": "LABEL_30",
101
+ "31": "LABEL_31",
102
+ "32": "LABEL_32",
103
+ "33": "LABEL_33",
104
+ "34": "LABEL_34",
105
+ "35": "LABEL_35",
106
+ "36": "LABEL_36",
107
+ "37": "LABEL_37",
108
+ "38": "LABEL_38",
109
+ "39": "LABEL_39",
110
+ "40": "LABEL_40",
111
+ "41": "LABEL_41",
112
+ "42": "LABEL_42",
113
+ "43": "LABEL_43",
114
+ "44": "LABEL_44",
115
+ "45": "LABEL_45",
116
+ "46": "LABEL_46",
117
+ "47": "LABEL_47",
118
+ "48": "LABEL_48",
119
+ "49": "LABEL_49",
120
+ "50": "LABEL_50",
121
+ "51": "LABEL_51",
122
+ "52": "LABEL_52",
123
+ "53": "LABEL_53",
124
+ "54": "LABEL_54",
125
+ "55": "LABEL_55",
126
+ "56": "LABEL_56",
127
+ "57": "LABEL_57",
128
+ "58": "LABEL_58",
129
+ "59": "LABEL_59",
130
+ "60": "LABEL_60",
131
+ "61": "LABEL_61",
132
+ "62": "LABEL_62",
133
+ "63": "LABEL_63",
134
+ "64": "LABEL_64",
135
+ "65": "LABEL_65",
136
+ "66": "LABEL_66",
137
+ "67": "LABEL_67",
138
+ "68": "LABEL_68",
139
+ "69": "LABEL_69",
140
+ "70": "LABEL_70",
141
+ "71": "LABEL_71",
142
+ "72": "LABEL_72",
143
+ "73": "LABEL_73",
144
+ "74": "LABEL_74",
145
+ "75": "LABEL_75",
146
+ "76": "LABEL_76",
147
+ "77": "LABEL_77",
148
+ "78": "LABEL_78",
149
+ "79": "LABEL_79",
150
+ "80": "LABEL_80",
151
+ "81": "LABEL_81",
152
+ "82": "LABEL_82",
153
+ "83": "LABEL_83"
154
+ },
155
+ "initializer_range": 0.02,
156
+ "intermediate_size": 1536,
157
+ "label2id": {
158
+ "LABEL_0": 0,
159
+ "LABEL_1": 1,
160
+ "LABEL_10": 10,
161
+ "LABEL_11": 11,
162
+ "LABEL_12": 12,
163
+ "LABEL_13": 13,
164
+ "LABEL_14": 14,
165
+ "LABEL_15": 15,
166
+ "LABEL_16": 16,
167
+ "LABEL_17": 17,
168
+ "LABEL_18": 18,
169
+ "LABEL_19": 19,
170
+ "LABEL_2": 2,
171
+ "LABEL_20": 20,
172
+ "LABEL_21": 21,
173
+ "LABEL_22": 22,
174
+ "LABEL_23": 23,
175
+ "LABEL_24": 24,
176
+ "LABEL_25": 25,
177
+ "LABEL_26": 26,
178
+ "LABEL_27": 27,
179
+ "LABEL_28": 28,
180
+ "LABEL_29": 29,
181
+ "LABEL_3": 3,
182
+ "LABEL_30": 30,
183
+ "LABEL_31": 31,
184
+ "LABEL_32": 32,
185
+ "LABEL_33": 33,
186
+ "LABEL_34": 34,
187
+ "LABEL_35": 35,
188
+ "LABEL_36": 36,
189
+ "LABEL_37": 37,
190
+ "LABEL_38": 38,
191
+ "LABEL_39": 39,
192
+ "LABEL_4": 4,
193
+ "LABEL_40": 40,
194
+ "LABEL_41": 41,
195
+ "LABEL_42": 42,
196
+ "LABEL_43": 43,
197
+ "LABEL_44": 44,
198
+ "LABEL_45": 45,
199
+ "LABEL_46": 46,
200
+ "LABEL_47": 47,
201
+ "LABEL_48": 48,
202
+ "LABEL_49": 49,
203
+ "LABEL_5": 5,
204
+ "LABEL_50": 50,
205
+ "LABEL_51": 51,
206
+ "LABEL_52": 52,
207
+ "LABEL_53": 53,
208
+ "LABEL_54": 54,
209
+ "LABEL_55": 55,
210
+ "LABEL_56": 56,
211
+ "LABEL_57": 57,
212
+ "LABEL_58": 58,
213
+ "LABEL_59": 59,
214
+ "LABEL_6": 6,
215
+ "LABEL_60": 60,
216
+ "LABEL_61": 61,
217
+ "LABEL_62": 62,
218
+ "LABEL_63": 63,
219
+ "LABEL_64": 64,
220
+ "LABEL_65": 65,
221
+ "LABEL_66": 66,
222
+ "LABEL_67": 67,
223
+ "LABEL_68": 68,
224
+ "LABEL_69": 69,
225
+ "LABEL_7": 7,
226
+ "LABEL_70": 70,
227
+ "LABEL_71": 71,
228
+ "LABEL_72": 72,
229
+ "LABEL_73": 73,
230
+ "LABEL_74": 74,
231
+ "LABEL_75": 75,
232
+ "LABEL_76": 76,
233
+ "LABEL_77": 77,
234
+ "LABEL_78": 78,
235
+ "LABEL_79": 79,
236
+ "LABEL_8": 8,
237
+ "LABEL_80": 80,
238
+ "LABEL_81": 81,
239
+ "LABEL_82": 82,
240
+ "LABEL_83": 83,
241
+ "LABEL_9": 9
242
+ },
243
+ "layer_norm_eps": 1e-07,
244
+ "layerdrop": 0.1,
245
+ "mask_feature_length": 64,
246
+ "mask_feature_min_masks": 0,
247
+ "mask_feature_prob": 0.5,
248
+ "mask_time_length": 10,
249
+ "mask_time_min_masks": 2,
250
+ "mask_time_prob": 0.65,
251
+ "max_position_embeddings": 512,
252
+ "model_type": "sew-d",
253
+ "norm_rel_ebd": "layer_norm",
254
+ "num_attention_heads": 6,
255
+ "num_conv_pos_embedding_groups": 16,
256
+ "num_conv_pos_embeddings": 31,
257
+ "num_feat_extract_layers": 13,
258
+ "num_hidden_layers": 12,
259
+ "pad_token_id": 0,
260
+ "pos_att_type": [
261
+ "p2c",
262
+ "c2p"
263
+ ],
264
+ "position_biased_input": false,
265
+ "position_buckets": 256,
266
+ "relative_attention": true,
267
+ "share_att_key": true,
268
+ "squeeze_factor": 2,
269
+ "tokenizer_class": "Wav2Vec2CTCTokenizer",
270
+ "torch_dtype": "float32",
271
+ "transformers_version": "4.38.2",
272
+ "use_weighted_layer_sum": false,
273
+ "vocab_size": 32
274
+ }
tmp-checkpoint-500/model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e37ee6672e4f4d1f5c287f6af0f282db7f9c77aa178b01992cbe08423408a09f
3
+ size 96968812
tmp-checkpoint-500/optimizer.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4cddbb1e7794d50366ca34b7e053ebd5e194e7a0373289708e6d7fe50284e3e8
3
+ size 179304954
tmp-checkpoint-500/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,
7
+ "return_attention_mask": false,
8
+ "sampling_rate": 16000
9
+ }
tmp-checkpoint-500/rng_state.pth ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3fa45e781fcba09d74d64bd4d6f2c48b0c94dd65dd88ad8c386a62c6b2493a18
3
+ size 14308
tmp-checkpoint-500/scheduler.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:af99c05e5d9ca370b2f99bbd9515c31b42aa4bc8c7434310774a52ca256443d2
3
+ size 1064
tmp-checkpoint-500/trainer_state.json ADDED
@@ -0,0 +1,158 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 0.0355601739229453,
3
+ "best_model_checkpoint": "fydhfzh/sewd_classifier\\checkpoint-500",
4
+ "epoch": 1.9398642095053347,
5
+ "eval_steps": 50,
6
+ "global_step": 500,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.19,
13
+ "eval_accuracy": 0.015776699029126214,
14
+ "eval_binary": 0.12427184466019374,
15
+ "eval_f1": 0.0021836259473444657,
16
+ "eval_loss": 4.4300618171691895,
17
+ "eval_precision": 0.0012509378194052938,
18
+ "eval_recall": 0.015776699029126214,
19
+ "eval_runtime": 1.5569,
20
+ "eval_samples_per_second": 529.245,
21
+ "eval_steps_per_second": 16.699,
22
+ "step": 50
23
+ },
24
+ {
25
+ "epoch": 0.39,
26
+ "eval_accuracy": 0.014563106796116505,
27
+ "eval_binary": 0.17548543689320378,
28
+ "eval_f1": 0.0013947717644465854,
29
+ "eval_loss": 4.409471035003662,
30
+ "eval_precision": 0.0009291296564070783,
31
+ "eval_recall": 0.014563106796116505,
32
+ "eval_runtime": 1.3313,
33
+ "eval_samples_per_second": 618.964,
34
+ "eval_steps_per_second": 19.53,
35
+ "step": 100
36
+ },
37
+ {
38
+ "epoch": 0.58,
39
+ "eval_accuracy": 0.025485436893203883,
40
+ "eval_binary": 0.22245145631068086,
41
+ "eval_f1": 0.008549061468070154,
42
+ "eval_loss": 4.337551593780518,
43
+ "eval_precision": 0.011859165691412587,
44
+ "eval_recall": 0.025485436893203883,
45
+ "eval_runtime": 1.4169,
46
+ "eval_samples_per_second": 581.534,
47
+ "eval_steps_per_second": 18.349,
48
+ "step": 150
49
+ },
50
+ {
51
+ "epoch": 0.78,
52
+ "eval_accuracy": 0.03640776699029126,
53
+ "eval_binary": 0.2802184466019448,
54
+ "eval_f1": 0.014380764940975713,
55
+ "eval_loss": 4.227560520172119,
56
+ "eval_precision": 0.032203883955239945,
57
+ "eval_recall": 0.03640776699029126,
58
+ "eval_runtime": 1.5427,
59
+ "eval_samples_per_second": 534.121,
60
+ "eval_steps_per_second": 16.853,
61
+ "step": 200
62
+ },
63
+ {
64
+ "epoch": 0.97,
65
+ "eval_accuracy": 0.05339805825242718,
66
+ "eval_binary": 0.3122572815534021,
67
+ "eval_f1": 0.02055877629770541,
68
+ "eval_loss": 4.134052276611328,
69
+ "eval_precision": 0.021501409455781946,
70
+ "eval_recall": 0.05339805825242718,
71
+ "eval_runtime": 1.6496,
72
+ "eval_samples_per_second": 499.507,
73
+ "eval_steps_per_second": 15.761,
74
+ "step": 250
75
+ },
76
+ {
77
+ "epoch": 1.0,
78
+ "grad_norm": 1.8999079465866089,
79
+ "learning_rate": 9e-06,
80
+ "loss": 4.3244,
81
+ "step": 257
82
+ },
83
+ {
84
+ "epoch": 1.16,
85
+ "eval_accuracy": 0.050970873786407765,
86
+ "eval_binary": 0.3190533980582566,
87
+ "eval_f1": 0.017415580929307544,
88
+ "eval_loss": 4.047242164611816,
89
+ "eval_precision": 0.013874415351638724,
90
+ "eval_recall": 0.050970873786407765,
91
+ "eval_runtime": 1.6726,
92
+ "eval_samples_per_second": 492.646,
93
+ "eval_steps_per_second": 15.545,
94
+ "step": 300
95
+ },
96
+ {
97
+ "epoch": 1.36,
98
+ "eval_accuracy": 0.06674757281553398,
99
+ "eval_binary": 0.33737864077670343,
100
+ "eval_f1": 0.021109602049388235,
101
+ "eval_loss": 3.9619970321655273,
102
+ "eval_precision": 0.015144910708081936,
103
+ "eval_recall": 0.06674757281553398,
104
+ "eval_runtime": 1.544,
105
+ "eval_samples_per_second": 533.67,
106
+ "eval_steps_per_second": 16.839,
107
+ "step": 350
108
+ },
109
+ {
110
+ "epoch": 1.55,
111
+ "eval_accuracy": 0.08495145631067962,
112
+ "eval_binary": 0.3495145631068007,
113
+ "eval_f1": 0.0362232684065308,
114
+ "eval_loss": 3.8773157596588135,
115
+ "eval_precision": 0.043156601029494375,
116
+ "eval_recall": 0.08495145631067962,
117
+ "eval_runtime": 1.6284,
118
+ "eval_samples_per_second": 506.031,
119
+ "eval_steps_per_second": 15.967,
120
+ "step": 400
121
+ },
122
+ {
123
+ "epoch": 1.75,
124
+ "eval_accuracy": 0.08859223300970874,
125
+ "eval_binary": 0.35618932038835416,
126
+ "eval_f1": 0.032488157810960344,
127
+ "eval_loss": 3.807878017425537,
128
+ "eval_precision": 0.027817262527314878,
129
+ "eval_recall": 0.08859223300970874,
130
+ "eval_runtime": 1.5357,
131
+ "eval_samples_per_second": 536.564,
132
+ "eval_steps_per_second": 16.93,
133
+ "step": 450
134
+ },
135
+ {
136
+ "epoch": 1.94,
137
+ "eval_accuracy": 0.08859223300970874,
138
+ "eval_binary": 0.3560679611650534,
139
+ "eval_f1": 0.0355601739229453,
140
+ "eval_loss": 3.7442734241485596,
141
+ "eval_precision": 0.03374590733472966,
142
+ "eval_recall": 0.08859223300970874,
143
+ "eval_runtime": 1.5994,
144
+ "eval_samples_per_second": 515.209,
145
+ "eval_steps_per_second": 16.257,
146
+ "step": 500
147
+ }
148
+ ],
149
+ "logging_steps": 500,
150
+ "max_steps": 2570,
151
+ "num_input_tokens_seen": 0,
152
+ "num_train_epochs": 10,
153
+ "save_steps": 500,
154
+ "total_flos": 2.1424965481558547e+17,
155
+ "train_batch_size": 32,
156
+ "trial_name": null,
157
+ "trial_params": null
158
+ }
tmp-checkpoint-500/training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a32a66e2bead39540dae299d70a2c30caa64ca81395c7788a1cc7dee06f4706
3
+ size 4984
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9b67d26a553a2ef4eb806550922b7f0cc62cb2dd5edce049e08c38aa48b7179d
3
  size 4984
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0a32a66e2bead39540dae299d70a2c30caa64ca81395c7788a1cc7dee06f4706
3
  size 4984