pluttodk commited on
Commit
dcdecb6
·
verified ·
1 Parent(s): a60788a

Upload config.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. config.json +224 -0
config.json ADDED
@@ -0,0 +1,224 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3ASRForConditionalGeneration"
4
+ ],
5
+ "dtype": "bfloat16",
6
+ "eos_token_id": 151645,
7
+ "model_type": "qwen3_asr",
8
+ "pad_token_id": 151643,
9
+ "support_languages": [
10
+ "Chinese",
11
+ "English",
12
+ "Cantonese",
13
+ "Arabic",
14
+ "German",
15
+ "French",
16
+ "Spanish",
17
+ "Portuguese",
18
+ "Indonesian",
19
+ "Italian",
20
+ "Korean",
21
+ "Russian",
22
+ "Thai",
23
+ "Vietnamese",
24
+ "Japanese",
25
+ "Turkish",
26
+ "Hindi",
27
+ "Malay",
28
+ "Dutch",
29
+ "Swedish",
30
+ "Danish",
31
+ "Finnish",
32
+ "Polish",
33
+ "Czech",
34
+ "Filipino",
35
+ "Persian",
36
+ "Greek",
37
+ "Romanian",
38
+ "Hungarian",
39
+ "Macedonian"
40
+ ],
41
+ "thinker_config": {
42
+ "architectures": [
43
+ "Qwen3ASRForConditionalGeneration"
44
+ ],
45
+ "audio_config": {
46
+ "_name_or_path": "",
47
+ "activation_dropout": 0,
48
+ "activation_function": "gelu",
49
+ "add_cross_attention": false,
50
+ "architectures": null,
51
+ "attention_dropout": 0,
52
+ "bad_words_ids": null,
53
+ "begin_suppress_tokens": null,
54
+ "bos_token_id": null,
55
+ "chunk_size_feed_forward": 0,
56
+ "conv_chunksize": 500,
57
+ "cross_attention_hidden_size": null,
58
+ "d_model": 1024,
59
+ "decoder_start_token_id": null,
60
+ "diversity_penalty": 0.0,
61
+ "do_sample": false,
62
+ "downsample_hidden_size": 480,
63
+ "dropout": 0,
64
+ "dtype": null,
65
+ "early_stopping": false,
66
+ "encoder_attention_heads": 16,
67
+ "encoder_ffn_dim": 4096,
68
+ "encoder_layers": 24,
69
+ "encoder_no_repeat_ngram_size": 0,
70
+ "eos_token_id": null,
71
+ "exponential_decay_length_penalty": null,
72
+ "finetuning_task": null,
73
+ "forced_bos_token_id": null,
74
+ "forced_eos_token_id": null,
75
+ "id2label": {
76
+ "0": "LABEL_0",
77
+ "1": "LABEL_1"
78
+ },
79
+ "initializer_range": 0.02,
80
+ "is_decoder": false,
81
+ "is_encoder_decoder": false,
82
+ "label2id": {
83
+ "LABEL_0": 0,
84
+ "LABEL_1": 1
85
+ },
86
+ "length_penalty": 1.0,
87
+ "max_length": 20,
88
+ "max_source_positions": 1500,
89
+ "min_length": 0,
90
+ "model_type": "qwen3_asr_audio_encoder",
91
+ "n_window": 50,
92
+ "n_window_infer": 800,
93
+ "no_repeat_ngram_size": 0,
94
+ "num_beam_groups": 1,
95
+ "num_beams": 1,
96
+ "num_hidden_layers": 24,
97
+ "num_mel_bins": 128,
98
+ "num_return_sequences": 1,
99
+ "output_attentions": false,
100
+ "output_dim": 2048,
101
+ "output_hidden_states": false,
102
+ "output_scores": false,
103
+ "pad_token_id": null,
104
+ "prefix": null,
105
+ "problem_type": null,
106
+ "pruned_heads": {},
107
+ "remove_invalid_values": false,
108
+ "repetition_penalty": 1.0,
109
+ "return_dict": true,
110
+ "return_dict_in_generate": false,
111
+ "scale_embedding": false,
112
+ "sep_token_id": null,
113
+ "suppress_tokens": null,
114
+ "task_specific_params": null,
115
+ "temperature": 1.0,
116
+ "tf_legacy_loss": false,
117
+ "tie_encoder_decoder": false,
118
+ "tie_word_embeddings": true,
119
+ "tokenizer_class": null,
120
+ "top_k": 50,
121
+ "top_p": 1.0,
122
+ "torchscript": false,
123
+ "typical_p": 1.0,
124
+ "use_bfloat16": false
125
+ },
126
+ "audio_end_token_id": 151670,
127
+ "audio_start_token_id": 151669,
128
+ "audio_token_id": 151676,
129
+ "dtype": "bfloat16",
130
+ "initializer_range": 0.02,
131
+ "model_type": "qwen3_asr_thinker",
132
+ "text_config": {
133
+ "_name_or_path": "",
134
+ "add_cross_attention": false,
135
+ "architectures": null,
136
+ "attention_bias": false,
137
+ "attention_dropout": 0.0,
138
+ "bad_words_ids": null,
139
+ "begin_suppress_tokens": null,
140
+ "bos_token_id": null,
141
+ "chunk_size_feed_forward": 0,
142
+ "cross_attention_hidden_size": null,
143
+ "decoder_start_token_id": null,
144
+ "diversity_penalty": 0.0,
145
+ "do_sample": false,
146
+ "dtype": null,
147
+ "early_stopping": false,
148
+ "encoder_no_repeat_ngram_size": 0,
149
+ "eos_token_id": null,
150
+ "exponential_decay_length_penalty": null,
151
+ "finetuning_task": null,
152
+ "forced_bos_token_id": null,
153
+ "forced_eos_token_id": null,
154
+ "head_dim": 128,
155
+ "hidden_act": "silu",
156
+ "hidden_size": 2048,
157
+ "id2label": {
158
+ "0": "LABEL_0",
159
+ "1": "LABEL_1"
160
+ },
161
+ "initializer_range": 0.02,
162
+ "intermediate_size": 6144,
163
+ "is_decoder": false,
164
+ "is_encoder_decoder": false,
165
+ "label2id": {
166
+ "LABEL_0": 0,
167
+ "LABEL_1": 1
168
+ },
169
+ "length_penalty": 1.0,
170
+ "max_length": 20,
171
+ "max_position_embeddings": 65536,
172
+ "min_length": 0,
173
+ "model_type": "qwen3_asr_text",
174
+ "no_repeat_ngram_size": 0,
175
+ "num_attention_heads": 16,
176
+ "num_beam_groups": 1,
177
+ "num_beams": 1,
178
+ "num_hidden_layers": 28,
179
+ "num_key_value_heads": 8,
180
+ "num_return_sequences": 1,
181
+ "output_attentions": false,
182
+ "output_hidden_states": false,
183
+ "output_scores": false,
184
+ "pad_token_id": null,
185
+ "prefix": null,
186
+ "problem_type": null,
187
+ "pruned_heads": {},
188
+ "remove_invalid_values": false,
189
+ "repetition_penalty": 1.0,
190
+ "return_dict": true,
191
+ "return_dict_in_generate": false,
192
+ "rms_norm_eps": 1e-06,
193
+ "rope_scaling": {
194
+ "interleaved": true,
195
+ "mrope_interleaved": true,
196
+ "mrope_section": [
197
+ 24,
198
+ 20,
199
+ 20
200
+ ],
201
+ "rope_type": "default",
202
+ "type": "default"
203
+ },
204
+ "rope_theta": 1000000,
205
+ "sep_token_id": null,
206
+ "suppress_tokens": null,
207
+ "task_specific_params": null,
208
+ "temperature": 1.0,
209
+ "tf_legacy_loss": false,
210
+ "tie_encoder_decoder": false,
211
+ "tie_word_embeddings": true,
212
+ "tokenizer_class": null,
213
+ "top_k": 50,
214
+ "top_p": 1.0,
215
+ "torchscript": false,
216
+ "typical_p": 1.0,
217
+ "use_bfloat16": false,
218
+ "use_cache": true,
219
+ "vocab_size": 151936
220
+ },
221
+ "user_token_id": 872
222
+ },
223
+ "transformers_version": "4.57.6"
224
+ }