Hugging Face's logo

patrickvonplaten
/
whisper-medium-24-2

Automatic Speech Recognition
Transformers
Safetensors
whisper
Inference Endpoints
Model card Files Community
patrickvonplaten commited on
Commit
e63407d
•
1 Parent(s): 6fd093a

Upload WhisperForConditionalGeneration

Browse files
Files changed (3) hide show
  1. config.json +148 -0
  2. generation_config.json +117 -0
  3. model.safetensors +3 -0
config.json ADDED
@@ -0,0 +1,148 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "sanchit-gandhi/medium-24-2-tpu-timestamped-prob-0.2",
3
+ "activation_dropout": 0.0,
4
+ "activation_function": "gelu",
5
+ "apply_spec_augment": false,
6
+ "architectures": [
7
+ "WhisperForConditionalGeneration"
8
+ ],
9
+ "attention_dropout": 0.0,
10
+ "begin_suppress_tokens": [
11
+ 220,
12
+ 50256
13
+ ],
14
+ "bos_token_id": 50257,
15
+ "classifier_proj_size": 256,
16
+ "d_model": 1024,
17
+ "decoder_attention_heads": 16,
18
+ "decoder_ffn_dim": 4096,
19
+ "decoder_layerdrop": 0.0,
20
+ "decoder_layers": 2,
21
+ "decoder_start_token_id": 50257,
22
+ "dropout": 0.0,
23
+ "encoder_attention_heads": 16,
24
+ "encoder_ffn_dim": 4096,
25
+ "encoder_layerdrop": 0.0,
26
+ "encoder_layers": 24,
27
+ "eos_token_id": 50256,
28
+ "forced_decoder_ids": [
29
+ [
30
+ 1,
31
+ 50362
32
+ ]
33
+ ],
34
+ "init_std": 0.02,
35
+ "is_encoder_decoder": true,
36
+ "mask_feature_length": 10,
37
+ "mask_feature_min_masks": 0,
38
+ "mask_feature_prob": 0.0,
39
+ "mask_time_length": 10,
40
+ "mask_time_min_masks": 2,
41
+ "mask_time_prob": 0.05,
42
+ "max_length": 448,
43
+ "max_source_positions": 1500,
44
+ "max_target_positions": 448,
45
+ "median_filter_width": 7,
46
+ "model_type": "whisper",
47
+ "num_hidden_layers": 24,
48
+ "num_mel_bins": 80,
49
+ "pad_token_id": 50256,
50
+ "scale_embedding": false,
51
+ "suppress_tokens": [
52
+ 1,
53
+ 2,
54
+ 7,
55
+ 8,
56
+ 9,
57
+ 10,
58
+ 14,
59
+ 25,
60
+ 26,
61
+ 27,
62
+ 28,
63
+ 29,
64
+ 31,
65
+ 58,
66
+ 59,
67
+ 60,
68
+ 61,
69
+ 62,
70
+ 63,
71
+ 90,
72
+ 91,
73
+ 92,
74
+ 93,
75
+ 357,
76
+ 366,
77
+ 438,
78
+ 532,
79
+ 685,
80
+ 705,
81
+ 796,
82
+ 930,
83
+ 1058,
84
+ 1220,
85
+ 1267,
86
+ 1279,
87
+ 1303,
88
+ 1343,
89
+ 1377,
90
+ 1391,
91
+ 1635,
92
+ 1782,
93
+ 1875,
94
+ 2162,
95
+ 2361,
96
+ 2488,
97
+ 3467,
98
+ 4008,
99
+ 4211,
100
+ 4600,
101
+ 4808,
102
+ 5299,
103
+ 5855,
104
+ 6329,
105
+ 7203,
106
+ 9609,
107
+ 9959,
108
+ 10563,
109
+ 10786,
110
+ 11420,
111
+ 11709,
112
+ 11907,
113
+ 13163,
114
+ 13697,
115
+ 13700,
116
+ 14808,
117
+ 15306,
118
+ 16410,
119
+ 16791,
120
+ 17992,
121
+ 19203,
122
+ 19510,
123
+ 20724,
124
+ 22305,
125
+ 22935,
126
+ 27007,
127
+ 30109,
128
+ 30420,
129
+ 33409,
130
+ 34949,
131
+ 40283,
132
+ 40493,
133
+ 40549,
134
+ 47282,
135
+ 49146,
136
+ 50257,
137
+ 50357,
138
+ 50358,
139
+ 50359,
140
+ 50360,
141
+ 50361
142
+ ],
143
+ "torch_dtype": "float32",
144
+ "transformers_version": "4.35.0.dev0",
145
+ "use_cache": true,
146
+ "use_weighted_layer_sum": false,
147
+ "vocab_size": 51864
148
+ }
generation_config.json ADDED
@@ -0,0 +1,117 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "begin_suppress_tokens": [
3
+ 220,
4
+ 50256
5
+ ],
6
+ "bos_token_id": 50257,
7
+ "decoder_start_token_id": 50257,
8
+ "eos_token_id": 50256,
9
+ "forced_decoder_ids": [
10
+ [
11
+ 1,
12
+ 50362
13
+ ]
14
+ ],
15
+ "is_multilingual": false,
16
+ "language": "<|en|>",
17
+ "max_initial_timestamp_index": 1,
18
+ "max_length": 448,
19
+ "no_timestamps_token_id": 50362,
20
+ "pad_token_id": 50256,
21
+ "return_timestamps": false,
22
+ "suppress_tokens": [
23
+ 1,
24
+ 2,
25
+ 7,
26
+ 8,
27
+ 9,
28
+ 10,
29
+ 14,
30
+ 25,
31
+ 26,
32
+ 27,
33
+ 28,
34
+ 29,
35
+ 31,
36
+ 58,
37
+ 59,
38
+ 60,
39
+ 61,
40
+ 62,
41
+ 63,
42
+ 90,
43
+ 91,
44
+ 92,
45
+ 93,
46
+ 357,
47
+ 366,
48
+ 438,
49
+ 532,
50
+ 685,
51
+ 705,
52
+ 796,
53
+ 930,
54
+ 1058,
55
+ 1220,
56
+ 1267,
57
+ 1279,
58
+ 1303,
59
+ 1343,
60
+ 1377,
61
+ 1391,
62
+ 1635,
63
+ 1782,
64
+ 1875,
65
+ 2162,
66
+ 2361,
67
+ 2488,
68
+ 3467,
69
+ 4008,
70
+ 4211,
71
+ 4600,
72
+ 4808,
73
+ 5299,
74
+ 5855,
75
+ 6329,
76
+ 7203,
77
+ 9609,
78
+ 9959,
79
+ 10563,
80
+ 10786,
81
+ 11420,
82
+ 11709,
83
+ 11907,
84
+ 13163,
85
+ 13697,
86
+ 13700,
87
+ 14808,
88
+ 15306,
89
+ 16410,
90
+ 16791,
91
+ 17992,
92
+ 19203,
93
+ 19510,
94
+ 20724,
95
+ 22305,
96
+ 22935,
97
+ 27007,
98
+ 30109,
99
+ 30420,
100
+ 33409,
101
+ 34949,
102
+ 40283,
103
+ 40493,
104
+ 40549,
105
+ 47282,
106
+ 49146,
107
+ 50257,
108
+ 50357,
109
+ 50358,
110
+ 50359,
111
+ 50360,
112
+ 50361
113
+ ],
114
+ "task": "transcribe",
115
+ "transformers_version": "4.35.0.dev0",
116
+ "use_scan": false
117
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c9c60a56f81e492449ffadb505fbf8b987b2dd14e8422f8858ddaab2c37c4650
3
+ size 1577549616