Hugging Face's logo

RedHatAI
/
Voxtral-Mini-3B-2507-FP8-dynamic

Automatic Speech Recognition
Safetensors
voxtral
fp8
quantized
multimodal
conversational
text-generation-inference
automatic-speech-translation
audio-text-to-text
video-text-to-text
compressed-tensors
Model card Files
xet
Community
4
Voxtral-Mini-3B-2507-FP8-dynamic
File size: 10,961 Bytes 
ec72c57
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
 
{
  "architectures": [
    "VoxtralForConditionalGeneration"
  ],
  "audio_config": {
    "activation_dropout": 0.0,
    "activation_function": "gelu",
    "attention_dropout": 0.0,
    "dropout": 0.0,
    "head_dim": 64,
    "hidden_size": 1280,
    "initializer_range": 0.02,
    "intermediate_size": 5120,
    "layerdrop": 0.0,
    "max_source_positions": 1500,
    "model_type": "voxtral_encoder",
    "num_attention_heads": 20,
    "num_hidden_layers": 32,
    "num_key_value_heads": 20,
    "num_mel_bins": 128,
    "scale_embedding": false,
    "torch_dtype": "bfloat16",
    "vocab_size": 51866
  },
  "audio_token_id": 24,
  "hidden_size": 3072,
  "model_type": "voxtral",
  "projector_hidden_act": "gelu",
  "quantization_config": {
    "config_groups": {
      "group_0": {
        "input_activations": {
          "actorder": null,
          "block_structure": null,
          "dynamic": true,
          "group_size": null,
          "num_bits": 8,
          "observer": null,
          "observer_kwargs": {},
          "strategy": "token",
          "symmetric": true,
          "type": "float"
        },
        "output_activations": null,
        "targets": [
          "Linear"
        ],
        "weights": {
          "actorder": null,
          "block_structure": null,
          "dynamic": false,
          "group_size": null,
          "num_bits": 8,
          "observer": "minmax",
          "observer_kwargs": {},
          "strategy": "channel",
          "symmetric": true,
          "type": "float"
        }
      }
    },
    "format": "float-quantized",
    "global_compression_ratio": null,
    "ignore": [
      "audio_tower.layers.0.self_attn.k_proj",
      "audio_tower.layers.0.self_attn.v_proj",
      "audio_tower.layers.0.self_attn.q_proj",
      "audio_tower.layers.0.self_attn.out_proj",
      "audio_tower.layers.0.fc1",
      "audio_tower.layers.0.fc2",
      "audio_tower.layers.1.self_attn.k_proj",
      "audio_tower.layers.1.self_attn.v_proj",
      "audio_tower.layers.1.self_attn.q_proj",
      "audio_tower.layers.1.self_attn.out_proj",
      "audio_tower.layers.1.fc1",
      "audio_tower.layers.1.fc2",
      "audio_tower.layers.2.self_attn.k_proj",
      "audio_tower.layers.2.self_attn.v_proj",
      "audio_tower.layers.2.self_attn.q_proj",
      "audio_tower.layers.2.self_attn.out_proj",
      "audio_tower.layers.2.fc1",
      "audio_tower.layers.2.fc2",
      "audio_tower.layers.3.self_attn.k_proj",
      "audio_tower.layers.3.self_attn.v_proj",
      "audio_tower.layers.3.self_attn.q_proj",
      "audio_tower.layers.3.self_attn.out_proj",
      "audio_tower.layers.3.fc1",
      "audio_tower.layers.3.fc2",
      "audio_tower.layers.4.self_attn.k_proj",
      "audio_tower.layers.4.self_attn.v_proj",
      "audio_tower.layers.4.self_attn.q_proj",
      "audio_tower.layers.4.self_attn.out_proj",
      "audio_tower.layers.4.fc1",
      "audio_tower.layers.4.fc2",
      "audio_tower.layers.5.self_attn.k_proj",
      "audio_tower.layers.5.self_attn.v_proj",
      "audio_tower.layers.5.self_attn.q_proj",
      "audio_tower.layers.5.self_attn.out_proj",
      "audio_tower.layers.5.fc1",
      "audio_tower.layers.5.fc2",
      "audio_tower.layers.6.self_attn.k_proj",
      "audio_tower.layers.6.self_attn.v_proj",
      "audio_tower.layers.6.self_attn.q_proj",
      "audio_tower.layers.6.self_attn.out_proj",
      "audio_tower.layers.6.fc1",
      "audio_tower.layers.6.fc2",
      "audio_tower.layers.7.self_attn.k_proj",
      "audio_tower.layers.7.self_attn.v_proj",
      "audio_tower.layers.7.self_attn.q_proj",
      "audio_tower.layers.7.self_attn.out_proj",
      "audio_tower.layers.7.fc1",
      "audio_tower.layers.7.fc2",
      "audio_tower.layers.8.self_attn.k_proj",
      "audio_tower.layers.8.self_attn.v_proj",
      "audio_tower.layers.8.self_attn.q_proj",
      "audio_tower.layers.8.self_attn.out_proj",
      "audio_tower.layers.8.fc1",
      "audio_tower.layers.8.fc2",
      "audio_tower.layers.9.self_attn.k_proj",
      "audio_tower.layers.9.self_attn.v_proj",
      "audio_tower.layers.9.self_attn.q_proj",
      "audio_tower.layers.9.self_attn.out_proj",
      "audio_tower.layers.9.fc1",
      "audio_tower.layers.9.fc2",
      "audio_tower.layers.10.self_attn.k_proj",
      "audio_tower.layers.10.self_attn.v_proj",
      "audio_tower.layers.10.self_attn.q_proj",
      "audio_tower.layers.10.self_attn.out_proj",
      "audio_tower.layers.10.fc1",
      "audio_tower.layers.10.fc2",
      "audio_tower.layers.11.self_attn.k_proj",
      "audio_tower.layers.11.self_attn.v_proj",
      "audio_tower.layers.11.self_attn.q_proj",
      "audio_tower.layers.11.self_attn.out_proj",
      "audio_tower.layers.11.fc1",
      "audio_tower.layers.11.fc2",
      "audio_tower.layers.12.self_attn.k_proj",
      "audio_tower.layers.12.self_attn.v_proj",
      "audio_tower.layers.12.self_attn.q_proj",
      "audio_tower.layers.12.self_attn.out_proj",
      "audio_tower.layers.12.fc1",
      "audio_tower.layers.12.fc2",
      "audio_tower.layers.13.self_attn.k_proj",
      "audio_tower.layers.13.self_attn.v_proj",
      "audio_tower.layers.13.self_attn.q_proj",
      "audio_tower.layers.13.self_attn.out_proj",
      "audio_tower.layers.13.fc1",
      "audio_tower.layers.13.fc2",
      "audio_tower.layers.14.self_attn.k_proj",
      "audio_tower.layers.14.self_attn.v_proj",
      "audio_tower.layers.14.self_attn.q_proj",
      "audio_tower.layers.14.self_attn.out_proj",
      "audio_tower.layers.14.fc1",
      "audio_tower.layers.14.fc2",
      "audio_tower.layers.15.self_attn.k_proj",
      "audio_tower.layers.15.self_attn.v_proj",
      "audio_tower.layers.15.self_attn.q_proj",
      "audio_tower.layers.15.self_attn.out_proj",
      "audio_tower.layers.15.fc1",
      "audio_tower.layers.15.fc2",
      "audio_tower.layers.16.self_attn.k_proj",
      "audio_tower.layers.16.self_attn.v_proj",
      "audio_tower.layers.16.self_attn.q_proj",
      "audio_tower.layers.16.self_attn.out_proj",
      "audio_tower.layers.16.fc1",
      "audio_tower.layers.16.fc2",
      "audio_tower.layers.17.self_attn.k_proj",
      "audio_tower.layers.17.self_attn.v_proj",
      "audio_tower.layers.17.self_attn.q_proj",
      "audio_tower.layers.17.self_attn.out_proj",
      "audio_tower.layers.17.fc1",
      "audio_tower.layers.17.fc2",
      "audio_tower.layers.18.self_attn.k_proj",
      "audio_tower.layers.18.self_attn.v_proj",
      "audio_tower.layers.18.self_attn.q_proj",
      "audio_tower.layers.18.self_attn.out_proj",
      "audio_tower.layers.18.fc1",
      "audio_tower.layers.18.fc2",
      "audio_tower.layers.19.self_attn.k_proj",
      "audio_tower.layers.19.self_attn.v_proj",
      "audio_tower.layers.19.self_attn.q_proj",
      "audio_tower.layers.19.self_attn.out_proj",
      "audio_tower.layers.19.fc1",
      "audio_tower.layers.19.fc2",
      "audio_tower.layers.20.self_attn.k_proj",
      "audio_tower.layers.20.self_attn.v_proj",
      "audio_tower.layers.20.self_attn.q_proj",
      "audio_tower.layers.20.self_attn.out_proj",
      "audio_tower.layers.20.fc1",
      "audio_tower.layers.20.fc2",
      "audio_tower.layers.21.self_attn.k_proj",
      "audio_tower.layers.21.self_attn.v_proj",
      "audio_tower.layers.21.self_attn.q_proj",
      "audio_tower.layers.21.self_attn.out_proj",
      "audio_tower.layers.21.fc1",
      "audio_tower.layers.21.fc2",
      "audio_tower.layers.22.self_attn.k_proj",
      "audio_tower.layers.22.self_attn.v_proj",
      "audio_tower.layers.22.self_attn.q_proj",
      "audio_tower.layers.22.self_attn.out_proj",
      "audio_tower.layers.22.fc1",
      "audio_tower.layers.22.fc2",
      "audio_tower.layers.23.self_attn.k_proj",
      "audio_tower.layers.23.self_attn.v_proj",
      "audio_tower.layers.23.self_attn.q_proj",
      "audio_tower.layers.23.self_attn.out_proj",
      "audio_tower.layers.23.fc1",
      "audio_tower.layers.23.fc2",
      "audio_tower.layers.24.self_attn.k_proj",
      "audio_tower.layers.24.self_attn.v_proj",
      "audio_tower.layers.24.self_attn.q_proj",
      "audio_tower.layers.24.self_attn.out_proj",
      "audio_tower.layers.24.fc1",
      "audio_tower.layers.24.fc2",
      "audio_tower.layers.25.self_attn.k_proj",
      "audio_tower.layers.25.self_attn.v_proj",
      "audio_tower.layers.25.self_attn.q_proj",
      "audio_tower.layers.25.self_attn.out_proj",
      "audio_tower.layers.25.fc1",
      "audio_tower.layers.25.fc2",
      "audio_tower.layers.26.self_attn.k_proj",
      "audio_tower.layers.26.self_attn.v_proj",
      "audio_tower.layers.26.self_attn.q_proj",
      "audio_tower.layers.26.self_attn.out_proj",
      "audio_tower.layers.26.fc1",
      "audio_tower.layers.26.fc2",
      "audio_tower.layers.27.self_attn.k_proj",
      "audio_tower.layers.27.self_attn.v_proj",
      "audio_tower.layers.27.self_attn.q_proj",
      "audio_tower.layers.27.self_attn.out_proj",
      "audio_tower.layers.27.fc1",
      "audio_tower.layers.27.fc2",
      "audio_tower.layers.28.self_attn.k_proj",
      "audio_tower.layers.28.self_attn.v_proj",
      "audio_tower.layers.28.self_attn.q_proj",
      "audio_tower.layers.28.self_attn.out_proj",
      "audio_tower.layers.28.fc1",
      "audio_tower.layers.28.fc2",
      "audio_tower.layers.29.self_attn.k_proj",
      "audio_tower.layers.29.self_attn.v_proj",
      "audio_tower.layers.29.self_attn.q_proj",
      "audio_tower.layers.29.self_attn.out_proj",
      "audio_tower.layers.29.fc1",
      "audio_tower.layers.29.fc2",
      "audio_tower.layers.30.self_attn.k_proj",
      "audio_tower.layers.30.self_attn.v_proj",
      "audio_tower.layers.30.self_attn.q_proj",
      "audio_tower.layers.30.self_attn.out_proj",
      "audio_tower.layers.30.fc1",
      "audio_tower.layers.30.fc2",
      "audio_tower.layers.31.self_attn.k_proj",
      "audio_tower.layers.31.self_attn.v_proj",
      "audio_tower.layers.31.self_attn.q_proj",
      "audio_tower.layers.31.self_attn.out_proj",
      "audio_tower.layers.31.fc1",
      "audio_tower.layers.31.fc2",
      "language_model.lm_head",
      "multi_modal_projector.linear_1",
      "multi_modal_projector.linear_2"
    ],
    "kv_cache_scheme": null,
    "quant_method": "compressed-tensors",
    "quantization_status": "compressed"
  },
  "text_config": {
    "attention_bias": false,
    "attention_dropout": 0.0,
    "head_dim": 128,
    "hidden_act": "silu",
    "hidden_size": 3072,
    "initializer_range": 0.02,
    "intermediate_size": 8192,
    "max_position_embeddings": 131072,
    "mlp_bias": false,
    "model_type": "llama",
    "num_attention_heads": 32,
    "num_hidden_layers": 30,
    "num_key_value_heads": 8,
    "pretraining_tp": 1,
    "rms_norm_eps": 1e-05,
    "rope_scaling": null,
    "rope_theta": 100000000.0,
    "sliding_window": null,
    "torch_dtype": "bfloat16",
    "use_cache": true,
    "vocab_size": 131072
  },
  "torch_dtype": "bfloat16",
  "transformers_version": "4.54.0",
  "vocab_size": 131072
}