Upload folder using huggingface_hub
Browse files- .gitattributes +3 -0
- config.json +305 -0
- generation_config.json +13 -0
- onnx/decoder_model.onnx +3 -0
- onnx/decoder_model.onnx_data +3 -0
- onnx/decoder_model_merged.onnx +3 -0
- onnx/decoder_model_merged.onnx_data +3 -0
- onnx/decoder_with_past_model.onnx +3 -0
- onnx/decoder_with_past_model.onnx_data +3 -0
- onnx/encoder_model.onnx +3 -0
- onnx/encoder_model_bnb4.onnx +3 -0
- onnx/encoder_model_fp16.onnx +3 -0
- onnx/encoder_model_int8.onnx +3 -0
- onnx/encoder_model_q4.onnx +3 -0
- onnx/encoder_model_q4f16.onnx +3 -0
- onnx/encoder_model_quantized.onnx +3 -0
- onnx/encoder_model_uint8.onnx +3 -0
.gitattributes
CHANGED
@@ -33,3 +33,6 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
|
|
33 |
*.zip filter=lfs diff=lfs merge=lfs -text
|
34 |
*.zst filter=lfs diff=lfs merge=lfs -text
|
35 |
*tfevents* filter=lfs diff=lfs merge=lfs -text
|
36 |
+
onnx/decoder_model.onnx_data filter=lfs diff=lfs merge=lfs -text
|
37 |
+
onnx/decoder_model_merged.onnx_data filter=lfs diff=lfs merge=lfs -text
|
38 |
+
onnx/decoder_with_past_model.onnx_data filter=lfs diff=lfs merge=lfs -text
|
config.json
ADDED
@@ -0,0 +1,305 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_attn_implementation_autoset": true,
|
3 |
+
"_name_or_path": "efficient-speech/lite-whisper-large-v3-fast",
|
4 |
+
"activation_dropout": 0.0,
|
5 |
+
"activation_function": "gelu",
|
6 |
+
"apply_spec_augment": false,
|
7 |
+
"architectures": [
|
8 |
+
"LiteWhisperForConditionalGeneration"
|
9 |
+
],
|
10 |
+
"attention_dropout": 0.0,
|
11 |
+
"auto_map": {
|
12 |
+
"AutoConfig": "efficient-speech/lite-whisper-large-v3-fast--configuration_lite_whisper.LiteWhisperConfig",
|
13 |
+
"AutoModel": "efficient-speech/lite-whisper-large-v3-fast--modeling_lite_whisper.LiteWhisperForConditionalGeneration"
|
14 |
+
},
|
15 |
+
"begin_suppress_tokens": null,
|
16 |
+
"bos_token_id": 50257,
|
17 |
+
"classifier_proj_size": 256,
|
18 |
+
"d_model": 1280,
|
19 |
+
"decoder_attention_heads": 20,
|
20 |
+
"decoder_ffn_dim": 5120,
|
21 |
+
"decoder_layerdrop": 0.0,
|
22 |
+
"decoder_layers": 32,
|
23 |
+
"decoder_start_token_id": 50258,
|
24 |
+
"dropout": 0.0,
|
25 |
+
"encoder_attention_heads": 20,
|
26 |
+
"encoder_ffn_dim": 5120,
|
27 |
+
"encoder_layerdrop": 0.0,
|
28 |
+
"encoder_layers": 32,
|
29 |
+
"eos_token_id": 50257,
|
30 |
+
"init_std": 0.02,
|
31 |
+
"is_encoder_decoder": true,
|
32 |
+
"low_rank_config": [
|
33 |
+
{
|
34 |
+
"fc1": 320,
|
35 |
+
"fc2": 272,
|
36 |
+
"k_proj": 32,
|
37 |
+
"out_proj": 32,
|
38 |
+
"q_proj": 32,
|
39 |
+
"v_proj": 48
|
40 |
+
},
|
41 |
+
{
|
42 |
+
"fc1": 192,
|
43 |
+
"fc2": 224,
|
44 |
+
"k_proj": 64,
|
45 |
+
"out_proj": 112,
|
46 |
+
"q_proj": 64,
|
47 |
+
"v_proj": 128
|
48 |
+
},
|
49 |
+
{
|
50 |
+
"fc1": 128,
|
51 |
+
"fc2": 112,
|
52 |
+
"k_proj": 48,
|
53 |
+
"out_proj": 160,
|
54 |
+
"q_proj": 48,
|
55 |
+
"v_proj": 160
|
56 |
+
},
|
57 |
+
{
|
58 |
+
"fc1": 192,
|
59 |
+
"fc2": 96,
|
60 |
+
"k_proj": 48,
|
61 |
+
"out_proj": 160,
|
62 |
+
"q_proj": 48,
|
63 |
+
"v_proj": 192
|
64 |
+
},
|
65 |
+
{
|
66 |
+
"fc1": 192,
|
67 |
+
"fc2": 160,
|
68 |
+
"k_proj": 48,
|
69 |
+
"out_proj": 144,
|
70 |
+
"q_proj": 64,
|
71 |
+
"v_proj": 208
|
72 |
+
},
|
73 |
+
{
|
74 |
+
"fc1": 192,
|
75 |
+
"fc2": 160,
|
76 |
+
"k_proj": 48,
|
77 |
+
"out_proj": 144,
|
78 |
+
"q_proj": 48,
|
79 |
+
"v_proj": 192
|
80 |
+
},
|
81 |
+
{
|
82 |
+
"fc1": 224,
|
83 |
+
"fc2": 224,
|
84 |
+
"k_proj": 64,
|
85 |
+
"out_proj": 208,
|
86 |
+
"q_proj": 80,
|
87 |
+
"v_proj": 304
|
88 |
+
},
|
89 |
+
{
|
90 |
+
"fc1": 240,
|
91 |
+
"fc2": 432,
|
92 |
+
"k_proj": 80,
|
93 |
+
"out_proj": 208,
|
94 |
+
"q_proj": 96,
|
95 |
+
"v_proj": 256
|
96 |
+
},
|
97 |
+
{
|
98 |
+
"fc1": 288,
|
99 |
+
"fc2": 512,
|
100 |
+
"k_proj": 80,
|
101 |
+
"out_proj": 224,
|
102 |
+
"q_proj": 96,
|
103 |
+
"v_proj": 288
|
104 |
+
},
|
105 |
+
{
|
106 |
+
"fc1": 288,
|
107 |
+
"fc2": 512,
|
108 |
+
"k_proj": 80,
|
109 |
+
"out_proj": 224,
|
110 |
+
"q_proj": 96,
|
111 |
+
"v_proj": 288
|
112 |
+
},
|
113 |
+
{
|
114 |
+
"fc1": 320,
|
115 |
+
"fc2": 528,
|
116 |
+
"k_proj": 48,
|
117 |
+
"out_proj": 240,
|
118 |
+
"q_proj": 64,
|
119 |
+
"v_proj": 288
|
120 |
+
},
|
121 |
+
{
|
122 |
+
"fc1": 320,
|
123 |
+
"fc2": 528,
|
124 |
+
"k_proj": 64,
|
125 |
+
"out_proj": 224,
|
126 |
+
"q_proj": 80,
|
127 |
+
"v_proj": 320
|
128 |
+
},
|
129 |
+
{
|
130 |
+
"fc1": 352,
|
131 |
+
"fc2": 592,
|
132 |
+
"k_proj": 64,
|
133 |
+
"out_proj": 224,
|
134 |
+
"q_proj": 64,
|
135 |
+
"v_proj": 352
|
136 |
+
},
|
137 |
+
{
|
138 |
+
"fc1": 400,
|
139 |
+
"fc2": 688,
|
140 |
+
"k_proj": 48,
|
141 |
+
"out_proj": 240,
|
142 |
+
"q_proj": 64,
|
143 |
+
"v_proj": 288
|
144 |
+
},
|
145 |
+
{
|
146 |
+
"fc1": 480,
|
147 |
+
"fc2": 656,
|
148 |
+
"k_proj": 64,
|
149 |
+
"out_proj": 256,
|
150 |
+
"q_proj": 80,
|
151 |
+
"v_proj": 448
|
152 |
+
},
|
153 |
+
{
|
154 |
+
"fc1": 384,
|
155 |
+
"fc2": 768,
|
156 |
+
"k_proj": 48,
|
157 |
+
"out_proj": 256,
|
158 |
+
"q_proj": 48,
|
159 |
+
"v_proj": 432
|
160 |
+
},
|
161 |
+
{
|
162 |
+
"fc1": 400,
|
163 |
+
"fc2": 800,
|
164 |
+
"k_proj": 48,
|
165 |
+
"out_proj": 224,
|
166 |
+
"q_proj": 64,
|
167 |
+
"v_proj": 272
|
168 |
+
},
|
169 |
+
{
|
170 |
+
"fc1": 464,
|
171 |
+
"fc2": 784,
|
172 |
+
"k_proj": 80,
|
173 |
+
"out_proj": 288,
|
174 |
+
"q_proj": 96,
|
175 |
+
"v_proj": 368
|
176 |
+
},
|
177 |
+
{
|
178 |
+
"fc1": 512,
|
179 |
+
"fc2": 848,
|
180 |
+
"k_proj": 64,
|
181 |
+
"out_proj": 256,
|
182 |
+
"q_proj": 80,
|
183 |
+
"v_proj": 304
|
184 |
+
},
|
185 |
+
{
|
186 |
+
"fc1": 528,
|
187 |
+
"fc2": 752,
|
188 |
+
"k_proj": 80,
|
189 |
+
"out_proj": 272,
|
190 |
+
"q_proj": 80,
|
191 |
+
"v_proj": 416
|
192 |
+
},
|
193 |
+
{
|
194 |
+
"fc1": 400,
|
195 |
+
"fc2": 16,
|
196 |
+
"k_proj": 80,
|
197 |
+
"out_proj": 352,
|
198 |
+
"q_proj": 96,
|
199 |
+
"v_proj": 432
|
200 |
+
},
|
201 |
+
{
|
202 |
+
"fc1": 672,
|
203 |
+
"fc2": 960,
|
204 |
+
"k_proj": 96,
|
205 |
+
"out_proj": 336,
|
206 |
+
"q_proj": 112,
|
207 |
+
"v_proj": 448
|
208 |
+
},
|
209 |
+
{
|
210 |
+
"fc1": 640,
|
211 |
+
"fc2": 976,
|
212 |
+
"k_proj": 112,
|
213 |
+
"out_proj": 448,
|
214 |
+
"q_proj": 144,
|
215 |
+
"v_proj": 544
|
216 |
+
},
|
217 |
+
{
|
218 |
+
"fc1": 720,
|
219 |
+
"k_proj": 96,
|
220 |
+
"out_proj": 416,
|
221 |
+
"q_proj": 128,
|
222 |
+
"v_proj": 464
|
223 |
+
},
|
224 |
+
{
|
225 |
+
"fc1": 752,
|
226 |
+
"k_proj": 128,
|
227 |
+
"out_proj": 448,
|
228 |
+
"q_proj": 176,
|
229 |
+
"v_proj": 560
|
230 |
+
},
|
231 |
+
{
|
232 |
+
"fc1": 752,
|
233 |
+
"k_proj": 128,
|
234 |
+
"out_proj": 416,
|
235 |
+
"q_proj": 176,
|
236 |
+
"v_proj": 592
|
237 |
+
},
|
238 |
+
{
|
239 |
+
"fc1": 848,
|
240 |
+
"k_proj": 144,
|
241 |
+
"out_proj": 496,
|
242 |
+
"q_proj": 192,
|
243 |
+
"v_proj": 624
|
244 |
+
},
|
245 |
+
{
|
246 |
+
"fc1": 912,
|
247 |
+
"fc2": 1024,
|
248 |
+
"k_proj": 128,
|
249 |
+
"out_proj": 400,
|
250 |
+
"q_proj": 192,
|
251 |
+
"v_proj": 608
|
252 |
+
},
|
253 |
+
{
|
254 |
+
"fc1": 944,
|
255 |
+
"fc2": 944,
|
256 |
+
"k_proj": 144,
|
257 |
+
"out_proj": 448,
|
258 |
+
"q_proj": 208,
|
259 |
+
"v_proj": 592
|
260 |
+
},
|
261 |
+
{
|
262 |
+
"fc1": 960,
|
263 |
+
"fc2": 976,
|
264 |
+
"k_proj": 176,
|
265 |
+
"out_proj": 496,
|
266 |
+
"q_proj": 272,
|
267 |
+
"v_proj": 560
|
268 |
+
},
|
269 |
+
{
|
270 |
+
"fc1": 896,
|
271 |
+
"fc2": 992,
|
272 |
+
"k_proj": 192,
|
273 |
+
"out_proj": 464,
|
274 |
+
"q_proj": 272,
|
275 |
+
"v_proj": 560
|
276 |
+
},
|
277 |
+
{
|
278 |
+
"fc1": 800,
|
279 |
+
"fc2": 736,
|
280 |
+
"k_proj": 176,
|
281 |
+
"out_proj": 464,
|
282 |
+
"q_proj": 272,
|
283 |
+
"v_proj": 432
|
284 |
+
}
|
285 |
+
],
|
286 |
+
"mask_feature_length": 10,
|
287 |
+
"mask_feature_min_masks": 0,
|
288 |
+
"mask_feature_prob": 0.0,
|
289 |
+
"mask_time_length": 10,
|
290 |
+
"mask_time_min_masks": 2,
|
291 |
+
"mask_time_prob": 0.05,
|
292 |
+
"max_length": null,
|
293 |
+
"max_source_positions": 1500,
|
294 |
+
"max_target_positions": 448,
|
295 |
+
"median_filter_width": 7,
|
296 |
+
"model_type": "lite-whisper",
|
297 |
+
"num_hidden_layers": 32,
|
298 |
+
"num_mel_bins": 128,
|
299 |
+
"pad_token_id": 50256,
|
300 |
+
"scale_embedding": false,
|
301 |
+
"transformers_version": "4.48.3",
|
302 |
+
"use_cache": true,
|
303 |
+
"use_weighted_layer_sum": false,
|
304 |
+
"vocab_size": 51866
|
305 |
+
}
|
generation_config.json
ADDED
@@ -0,0 +1,13 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"_from_model_config": true,
|
3 |
+
"begin_suppress_tokens": [
|
4 |
+
220,
|
5 |
+
50257
|
6 |
+
],
|
7 |
+
"bos_token_id": 50257,
|
8 |
+
"decoder_start_token_id": 50258,
|
9 |
+
"eos_token_id": 50257,
|
10 |
+
"max_length": 448,
|
11 |
+
"pad_token_id": 50256,
|
12 |
+
"transformers_version": "4.48.3"
|
13 |
+
}
|
onnx/decoder_model.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d0f62c74af0f48b8e6e17487c49b4f462ee76ea66418f79d2390e89342886c7e
|
3 |
+
size 1164232
|
onnx/decoder_model.onnx_data
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:7d72c229cb198f4e397b8081fbc31a3528b64f338fb41a934301237043267ca9
|
3 |
+
size 3626086400
|
onnx/decoder_model_merged.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d33d49464fd7c4dc040e924c29461b772227103f2115a699fd11ca95e2c8101b
|
3 |
+
size 2569563
|
onnx/decoder_model_merged.onnx_data
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:69700fe5f05cae25e834e9f226e678aebbc18af13c5526ad81829489c357515d
|
3 |
+
size 3626086400
|
onnx/decoder_with_past_model.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:53d4e685a26a9ef50eef77099cd5cbb93735bc864da83ac8c6cac817b4c7f687
|
3 |
+
size 903746
|
onnx/decoder_with_past_model.onnx_data
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ae383dec2237699bebf91bd1b75c9d357b3da020a4a54a0f77041ce86f7ed74d
|
3 |
+
size 3206492160
|
onnx/encoder_model.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:53eb8ec41c9723526b1f503878953b187c306aabfe021e578b7f8fd34a0d6d74
|
3 |
+
size 1238999916
|
onnx/encoder_model_bnb4.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:03e9853d551fa90f5f5df8923c33d0c5aa3b92f63bc62191ff30df9993186d60
|
3 |
+
size 201724891
|
onnx/encoder_model_fp16.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:426b5a5ec84f7dd630bb273f8787ae9dae68d9c69271e19bc7b459b07864c97d
|
3 |
+
size 619725689
|
onnx/encoder_model_int8.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:8331e00bd603234cb4593e7752b57900f339dbd50e81699fd9b68384bc1e9984
|
3 |
+
size 317890354
|
onnx/encoder_model_q4.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:0c09ed8b9fad1f824cb897f1c087b5d5385bc791a1c14be9d96331c994f9dc33
|
3 |
+
size 222234014
|
onnx/encoder_model_q4f16.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:40bc53802318f98e836f78528dc5dee6e68439be8cee96356f7b5063c1e79c0e
|
3 |
+
size 187479509
|
onnx/encoder_model_quantized.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bd65cb12f994ec6d854bbb650a9605febbbcec5efbd84670fded09b390facd6f
|
3 |
+
size 317890535
|
onnx/encoder_model_uint8.onnx
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bd65cb12f994ec6d854bbb650a9605febbbcec5efbd84670fded09b390facd6f
|
3 |
+
size 317890535
|