Hugging Face's logo

voidful
/
QAQ_4b_exp6

Safetensors
desta25
Model card Files
xet
Community
voidful commited on
Commit
7b752a9
·
verified ·
1 Parent(s): 477ac77

Upload 2 files

Browse files
Files changed (2) hide show
  1. config.json +150 -0
  2. model.safetensors +3 -0
config.json ADDED
@@ -0,0 +1,150 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "DeSTA25AudioModel"
4
+ ],
5
+ "audio_locator": "<|AUDIO|>",
6
+ "connector_mode": "orca_hybrid",
7
+ "dtype": "bfloat16",
8
+ "encoder_config": {
9
+ "_name_or_path": "openai/whisper-large-v3",
10
+ "activation_dropout": 0.0,
11
+ "activation_function": "gelu",
12
+ "apply_spec_augment": false,
13
+ "architectures": [
14
+ "WhisperForConditionalGeneration"
15
+ ],
16
+ "attention_dropout": 0.0,
17
+ "begin_suppress_tokens": [
18
+ 220,
19
+ 50257
20
+ ],
21
+ "bos_token_id": 50257,
22
+ "classifier_proj_size": 256,
23
+ "d_model": 1280,
24
+ "decoder_attention_heads": 20,
25
+ "decoder_ffn_dim": 5120,
26
+ "decoder_layerdrop": 0.0,
27
+ "decoder_layers": 32,
28
+ "decoder_start_token_id": 50258,
29
+ "dropout": 0.0,
30
+ "dtype": "float16",
31
+ "encoder_attention_heads": 20,
32
+ "encoder_ffn_dim": 5120,
33
+ "encoder_layerdrop": 0.0,
34
+ "encoder_layers": 32,
35
+ "eos_token_id": 50257,
36
+ "init_std": 0.02,
37
+ "mask_feature_length": 10,
38
+ "mask_feature_min_masks": 0,
39
+ "mask_feature_prob": 0.0,
40
+ "mask_time_length": 10,
41
+ "mask_time_min_masks": 2,
42
+ "mask_time_prob": 0.05,
43
+ "max_length": 448,
44
+ "max_source_positions": 1500,
45
+ "max_target_positions": 448,
46
+ "median_filter_width": 7,
47
+ "model_type": "whisper",
48
+ "num_hidden_layers": 32,
49
+ "num_mel_bins": 128,
50
+ "scale_embedding": false,
51
+ "use_cache": true,
52
+ "use_weighted_layer_sum": false,
53
+ "vocab_size": 51866
54
+ },
55
+ "encoder_model_id": "openai/whisper-large-v3",
56
+ "eos_token_id": 151645,
57
+ "info": "\uff24\uff45\uff53\uff54\uff41\uff12\u3002\uff15 \uff21\uff55\uff44\uff49\uff4f",
58
+ "llm_config": {
59
+ "_name_or_path": "Qwen/Qwen3-4B-Instruct-2507",
60
+ "architectures": [
61
+ "Qwen3ForCausalLM"
62
+ ],
63
+ "attention_bias": false,
64
+ "attention_dropout": 0.0,
65
+ "bos_token_id": 151643,
66
+ "dtype": "bfloat16",
67
+ "eos_token_id": 151645,
68
+ "head_dim": 128,
69
+ "hidden_act": "silu",
70
+ "hidden_size": 2560,
71
+ "initializer_range": 0.02,
72
+ "intermediate_size": 9728,
73
+ "layer_types": [
74
+ "full_attention",
75
+ "full_attention",
76
+ "full_attention",
77
+ "full_attention",
78
+ "full_attention",
79
+ "full_attention",
80
+ "full_attention",
81
+ "full_attention",
82
+ "full_attention",
83
+ "full_attention",
84
+ "full_attention",
85
+ "full_attention",
86
+ "full_attention",
87
+ "full_attention",
88
+ "full_attention",
89
+ "full_attention",
90
+ "full_attention",
91
+ "full_attention",
92
+ "full_attention",
93
+ "full_attention",
94
+ "full_attention",
95
+ "full_attention",
96
+ "full_attention",
97
+ "full_attention",
98
+ "full_attention",
99
+ "full_attention",
100
+ "full_attention",
101
+ "full_attention",
102
+ "full_attention",
103
+ "full_attention",
104
+ "full_attention",
105
+ "full_attention",
106
+ "full_attention",
107
+ "full_attention",
108
+ "full_attention",
109
+ "full_attention"
110
+ ],
111
+ "max_position_embeddings": 262144,
112
+ "max_window_layers": 36,
113
+ "model_type": "qwen3",
114
+ "num_attention_heads": 32,
115
+ "num_hidden_layers": 36,
116
+ "num_key_value_heads": 8,
117
+ "rms_norm_eps": 1e-06,
118
+ "rope_scaling": null,
119
+ "rope_theta": 5000000,
120
+ "sliding_window": null,
121
+ "tie_word_embeddings": true,
122
+ "use_cache": true,
123
+ "use_sliding_window": false,
124
+ "vocab_size": 151936
125
+ },
126
+ "llm_model_id": "Qwen/Qwen3-4B-Instruct-2507",
127
+ "model_type": "desta25",
128
+ "orca_align_weight_local": 0.05,
129
+ "orca_audio_position_scale": 2.5,
130
+ "orca_deep_injection_enabled": true,
131
+ "orca_deep_injection_stride": 1,
132
+ "orca_enabled": true,
133
+ "orca_gate_init": 0.1,
134
+ "orca_global_cross_attn": true,
135
+ "orca_global_num_tokens": 64,
136
+ "orca_local_downsample": 4,
137
+ "orca_local_enabled": true,
138
+ "orca_local_kernel_size": 5,
139
+ "orca_ortho_diversity_weight": 0.05,
140
+ "orca_ortho_weight_global": 0.05,
141
+ "orca_ortho_weight_qformer_local": 0.05,
142
+ "orca_use_all_layers": false,
143
+ "pad_token_id": 151645,
144
+ "placeholder_token": "<|video_pad|>",
145
+ "prompt_size": 64,
146
+ "qformer_num_hidden_layers": 6,
147
+ "train_id": 30678,
148
+ "transformers_version": "4.57.3",
149
+ "use_lora": false
150
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:012feb9adb43cb7d72cc3456bbe5dc8733d4e72216f548e93ea40994903cc5ac
3
+ size 2337920272