prompteus commited on
Commit
842c8d4
1 Parent(s): 4551ebe

Upload WhisperForAudioCaptioning (#1)

Browse files

- Upload WhisperForAudioCaptioning (e5c65d22e32ae1389bac1a5dec13f3a67881ab9a)

Files changed (3) hide show
  1. config.json +153 -0
  2. generation_config.json +231 -0
  3. pytorch_model.bin +3 -0
config.json ADDED
@@ -0,0 +1,153 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "../checkpoints/fancy-galaxy-115/checkpoint-3900/",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
+ "architectures": [
7
+ "WhisperForAudioCaptioning"
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
+ "forced_decoder_ids": [
29
+ [
30
+ 1,
31
+ 50259
32
+ ],
33
+ [
34
+ 2,
35
+ 50359
36
+ ],
37
+ [
38
+ 3,
39
+ 50363
40
+ ]
41
+ ],
42
+ "init_std": 0.02,
43
+ "is_encoder_decoder": true,
44
+ "mask_feature_length": 10,
45
+ "mask_feature_min_masks": 0,
46
+ "mask_feature_prob": 0.0,
47
+ "mask_time_length": 10,
48
+ "mask_time_min_masks": 2,
49
+ "mask_time_prob": 0.05,
50
+ "max_length": 448,
51
+ "max_source_positions": 1500,
52
+ "max_target_positions": 448,
53
+ "model_type": "whisper",
54
+ "num_hidden_layers": 4,
55
+ "num_mel_bins": 80,
56
+ "pad_token_id": 50257,
57
+ "scale_embedding": false,
58
+ "suppress_tokens": [
59
+ 1,
60
+ 2,
61
+ 7,
62
+ 8,
63
+ 9,
64
+ 10,
65
+ 14,
66
+ 25,
67
+ 26,
68
+ 27,
69
+ 28,
70
+ 29,
71
+ 31,
72
+ 58,
73
+ 59,
74
+ 60,
75
+ 61,
76
+ 62,
77
+ 63,
78
+ 90,
79
+ 91,
80
+ 92,
81
+ 93,
82
+ 359,
83
+ 503,
84
+ 522,
85
+ 542,
86
+ 873,
87
+ 893,
88
+ 902,
89
+ 918,
90
+ 922,
91
+ 931,
92
+ 1350,
93
+ 1853,
94
+ 1982,
95
+ 2460,
96
+ 2627,
97
+ 3246,
98
+ 3253,
99
+ 3268,
100
+ 3536,
101
+ 3846,
102
+ 3961,
103
+ 4183,
104
+ 4667,
105
+ 6585,
106
+ 6647,
107
+ 7273,
108
+ 9061,
109
+ 9383,
110
+ 10428,
111
+ 10929,
112
+ 11938,
113
+ 12033,
114
+ 12331,
115
+ 12562,
116
+ 13793,
117
+ 14157,
118
+ 14635,
119
+ 15265,
120
+ 15618,
121
+ 16553,
122
+ 16604,
123
+ 18362,
124
+ 18956,
125
+ 20075,
126
+ 21675,
127
+ 22520,
128
+ 26130,
129
+ 26161,
130
+ 26435,
131
+ 28279,
132
+ 29464,
133
+ 31650,
134
+ 32302,
135
+ 32470,
136
+ 36865,
137
+ 42863,
138
+ 47425,
139
+ 49870,
140
+ 50254,
141
+ 50258,
142
+ 50358,
143
+ 50359,
144
+ 50360,
145
+ 50361,
146
+ 50362
147
+ ],
148
+ "torch_dtype": "float32",
149
+ "transformers_version": "4.27.1",
150
+ "use_cache": true,
151
+ "use_weighted_layer_sum": false,
152
+ "vocab_size": 51865
153
+ }
generation_config.json ADDED
@@ -0,0 +1,231 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "begin_suppress_tokens": [
3
+ 220,
4
+ 50257
5
+ ],
6
+ "bos_token_id": 50257,
7
+ "decoder_start_token_id": 50258,
8
+ "eos_token_id": 50257,
9
+ "forced_decoder_ids": [
10
+ [
11
+ 0,
12
+ 50258
13
+ ],
14
+ [
15
+ 1,
16
+ 50259
17
+ ],
18
+ [
19
+ 2,
20
+ 50359
21
+ ],
22
+ [
23
+ 3,
24
+ 50363
25
+ ]
26
+ ],
27
+ "is_multilingual": true,
28
+ "lang_to_id": {
29
+ "<|af|>": 50327,
30
+ "<|am|>": 50334,
31
+ "<|ar|>": 50272,
32
+ "<|as|>": 50350,
33
+ "<|az|>": 50304,
34
+ "<|ba|>": 50355,
35
+ "<|be|>": 50330,
36
+ "<|bg|>": 50292,
37
+ "<|bn|>": 50302,
38
+ "<|bo|>": 50347,
39
+ "<|br|>": 50309,
40
+ "<|bs|>": 50315,
41
+ "<|ca|>": 50270,
42
+ "<|cs|>": 50283,
43
+ "<|cy|>": 50297,
44
+ "<|da|>": 50285,
45
+ "<|de|>": 50261,
46
+ "<|el|>": 50281,
47
+ "<|en|>": 50259,
48
+ "<|es|>": 50262,
49
+ "<|et|>": 50307,
50
+ "<|eu|>": 50310,
51
+ "<|fa|>": 50300,
52
+ "<|fi|>": 50277,
53
+ "<|fo|>": 50338,
54
+ "<|fr|>": 50265,
55
+ "<|gl|>": 50319,
56
+ "<|gu|>": 50333,
57
+ "<|haw|>": 50352,
58
+ "<|ha|>": 50354,
59
+ "<|he|>": 50279,
60
+ "<|hi|>": 50276,
61
+ "<|hr|>": 50291,
62
+ "<|ht|>": 50339,
63
+ "<|hu|>": 50286,
64
+ "<|hy|>": 50312,
65
+ "<|id|>": 50275,
66
+ "<|is|>": 50311,
67
+ "<|it|>": 50274,
68
+ "<|ja|>": 50266,
69
+ "<|jw|>": 50356,
70
+ "<|ka|>": 50329,
71
+ "<|kk|>": 50316,
72
+ "<|km|>": 50323,
73
+ "<|kn|>": 50306,
74
+ "<|ko|>": 50264,
75
+ "<|la|>": 50294,
76
+ "<|lb|>": 50345,
77
+ "<|ln|>": 50353,
78
+ "<|lo|>": 50336,
79
+ "<|lt|>": 50293,
80
+ "<|lv|>": 50301,
81
+ "<|mg|>": 50349,
82
+ "<|mi|>": 50295,
83
+ "<|mk|>": 50308,
84
+ "<|ml|>": 50296,
85
+ "<|mn|>": 50314,
86
+ "<|mr|>": 50320,
87
+ "<|ms|>": 50282,
88
+ "<|mt|>": 50343,
89
+ "<|my|>": 50346,
90
+ "<|ne|>": 50313,
91
+ "<|nl|>": 50271,
92
+ "<|nn|>": 50342,
93
+ "<|no|>": 50288,
94
+ "<|oc|>": 50328,
95
+ "<|pa|>": 50321,
96
+ "<|pl|>": 50269,
97
+ "<|ps|>": 50340,
98
+ "<|pt|>": 50267,
99
+ "<|ro|>": 50284,
100
+ "<|ru|>": 50263,
101
+ "<|sa|>": 50344,
102
+ "<|sd|>": 50332,
103
+ "<|si|>": 50322,
104
+ "<|sk|>": 50298,
105
+ "<|sl|>": 50305,
106
+ "<|sn|>": 50324,
107
+ "<|so|>": 50326,
108
+ "<|sq|>": 50317,
109
+ "<|sr|>": 50303,
110
+ "<|su|>": 50357,
111
+ "<|sv|>": 50273,
112
+ "<|sw|>": 50318,
113
+ "<|ta|>": 50287,
114
+ "<|te|>": 50299,
115
+ "<|tg|>": 50331,
116
+ "<|th|>": 50289,
117
+ "<|tk|>": 50341,
118
+ "<|tl|>": 50348,
119
+ "<|tr|>": 50268,
120
+ "<|tt|>": 50351,
121
+ "<|uk|>": 50280,
122
+ "<|ur|>": 50290,
123
+ "<|uz|>": 50337,
124
+ "<|vi|>": 50278,
125
+ "<|yi|>": 50335,
126
+ "<|yo|>": 50325,
127
+ "<|zh|>": 50260
128
+ },
129
+ "language": "english",
130
+ "max_initial_timestamp_index": 1,
131
+ "max_length": 448,
132
+ "no_timestamps_token_id": 50363,
133
+ "pad_token_id": 50257,
134
+ "return_timestamps": false,
135
+ "suppress_tokens": [
136
+ 1,
137
+ 2,
138
+ 7,
139
+ 8,
140
+ 9,
141
+ 10,
142
+ 14,
143
+ 25,
144
+ 26,
145
+ 27,
146
+ 28,
147
+ 29,
148
+ 31,
149
+ 58,
150
+ 59,
151
+ 60,
152
+ 61,
153
+ 62,
154
+ 63,
155
+ 90,
156
+ 91,
157
+ 92,
158
+ 93,
159
+ 359,
160
+ 503,
161
+ 522,
162
+ 542,
163
+ 873,
164
+ 893,
165
+ 902,
166
+ 918,
167
+ 922,
168
+ 931,
169
+ 1350,
170
+ 1853,
171
+ 1982,
172
+ 2460,
173
+ 2627,
174
+ 3246,
175
+ 3253,
176
+ 3268,
177
+ 3536,
178
+ 3846,
179
+ 3961,
180
+ 4183,
181
+ 4667,
182
+ 6585,
183
+ 6647,
184
+ 7273,
185
+ 9061,
186
+ 9383,
187
+ 10428,
188
+ 10929,
189
+ 11938,
190
+ 12033,
191
+ 12331,
192
+ 12562,
193
+ 13793,
194
+ 14157,
195
+ 14635,
196
+ 15265,
197
+ 15618,
198
+ 16553,
199
+ 16604,
200
+ 18362,
201
+ 18956,
202
+ 20075,
203
+ 21675,
204
+ 22520,
205
+ 26130,
206
+ 26161,
207
+ 26435,
208
+ 28279,
209
+ 29464,
210
+ 31650,
211
+ 32302,
212
+ 32470,
213
+ 36865,
214
+ 42863,
215
+ 47425,
216
+ 49870,
217
+ 50254,
218
+ 50258,
219
+ 50358,
220
+ 50359,
221
+ 50360,
222
+ 50361,
223
+ 50362
224
+ ],
225
+ "task": "transcribe",
226
+ "task_to_id": {
227
+ "transcribe": 50359,
228
+ "translate": 50358
229
+ },
230
+ "transformers_version": "4.27.1"
231
+ }
pytorch_model.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f8c77ef91b396939fab93d1e4b78a7307ccfadabc9982adb487fb1e564088be6
3
+ size 151096617