Isaak Carter Augustus
commited on
Commit
•
22daade
1
Parent(s):
a8c7fe9
Delete model_architecture.txt
Browse files- model_architecture.txt +0 -1977
model_architecture.txt
DELETED
@@ -1,1977 +0,0 @@
|
|
1 |
-
Currently not using the OG trained model for easy and fast loading, ...
|
2 |
-
Used LLM Qwen2 0.5B
|
3 |
-
|
4 |
-
JOSIE(
|
5 |
-
(imagebind_encoder): ImageBindModel(
|
6 |
-
(modality_preprocessors): ModuleDict(
|
7 |
-
(vision): RGBDTPreprocessor(
|
8 |
-
(cls_token): tensor((1, 1, 1280), requires_grad=False)
|
9 |
-
|
10 |
-
(rgbt_stem): PatchEmbedGeneric(
|
11 |
-
(proj): Sequential(
|
12 |
-
(0): PadIm2Video()
|
13 |
-
(1): Conv3d(3, 1280, kernel_size=(2, 14, 14), stride=(2, 14, 14), bias=False)
|
14 |
-
)
|
15 |
-
)
|
16 |
-
(pos_embedding_helper): SpatioTemporalPosEmbeddingHelper(
|
17 |
-
(pos_embed): tensor((1, 257, 1280), requires_grad=False)
|
18 |
-
|
19 |
-
)
|
20 |
-
)
|
21 |
-
(text): TextPreprocessor(
|
22 |
-
(pos_embed): tensor((1, 77, 1024), requires_grad=False)
|
23 |
-
(mask): tensor((77, 77), requires_grad=False)
|
24 |
-
|
25 |
-
(token_embedding): Embedding(49408, 1024)
|
26 |
-
)
|
27 |
-
(audio): AudioPreprocessor(
|
28 |
-
(cls_token): tensor((1, 1, 768), requires_grad=False)
|
29 |
-
|
30 |
-
(rgbt_stem): PatchEmbedGeneric(
|
31 |
-
(proj): Conv2d(1, 768, kernel_size=(16, 16), stride=(10, 10), bias=False)
|
32 |
-
(norm_layer): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
|
33 |
-
)
|
34 |
-
(pos_embedding_helper): SpatioTemporalPosEmbeddingHelper(
|
35 |
-
(pos_embed): tensor((1, 229, 768), requires_grad=False)
|
36 |
-
|
37 |
-
)
|
38 |
-
)
|
39 |
-
(depth): RGBDTPreprocessor(
|
40 |
-
(cls_token): tensor((1, 1, 384), requires_grad=False)
|
41 |
-
|
42 |
-
(depth_stem): PatchEmbedGeneric(
|
43 |
-
(proj): Conv2d(1, 384, kernel_size=(16, 16), stride=(16, 16), bias=False)
|
44 |
-
(norm_layer): LayerNorm((384,), eps=1e-05, elementwise_affine=True)
|
45 |
-
)
|
46 |
-
(pos_embedding_helper): SpatioTemporalPosEmbeddingHelper(
|
47 |
-
(pos_embed): tensor((1, 197, 384), requires_grad=False)
|
48 |
-
|
49 |
-
)
|
50 |
-
)
|
51 |
-
(thermal): ThermalPreprocessor(
|
52 |
-
(cls_token): tensor((1, 1, 768), requires_grad=False)
|
53 |
-
|
54 |
-
(rgbt_stem): PatchEmbedGeneric(
|
55 |
-
(proj): Conv2d(1, 768, kernel_size=(16, 16), stride=(16, 16), bias=False)
|
56 |
-
(norm_layer): LayerNorm((768,), eps=1e-05, elementwise_affine=True)
|
57 |
-
)
|
58 |
-
(pos_embedding_helper): SpatioTemporalPosEmbeddingHelper(
|
59 |
-
(pos_embed): tensor((1, 197, 768), requires_grad=False)
|
60 |
-
|
61 |
-
)
|
62 |
-
)
|
63 |
-
(imu): IMUPreprocessor(
|
64 |
-
(pos_embed): tensor((1, 251, 512), requires_grad=False)
|
65 |
-
(cls_token): tensor((1, 1, 512), requires_grad=False)
|
66 |
-
|
67 |
-
(imu_stem): PatchEmbedGeneric(
|
68 |
-
(proj): Linear(in_features=48, out_features=512, bias=False)
|
69 |
-
(norm_layer): LayerNorm((512,), eps=1e-05, elementwise_affine=True)
|
70 |
-
)
|
71 |
-
)
|
72 |
-
)
|
73 |
-
(modality_trunks): ModuleDict(
|
74 |
-
(vision): SimpleTransformer(
|
75 |
-
(pre_transformer_layer): Sequential(
|
76 |
-
(0): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
77 |
-
(1): EinOpsRearrange()
|
78 |
-
)
|
79 |
-
(blocks): Sequential(
|
80 |
-
(0): BlockWithMasking(
|
81 |
-
(attn): MultiheadAttention(
|
82 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
83 |
-
)
|
84 |
-
(drop_path): Identity()
|
85 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
86 |
-
(mlp): Mlp(
|
87 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
88 |
-
(act): GELU(approximate='none')
|
89 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
90 |
-
(drop): Dropout(p=0.0, inplace=False)
|
91 |
-
)
|
92 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
93 |
-
)
|
94 |
-
(1): BlockWithMasking(
|
95 |
-
(attn): MultiheadAttention(
|
96 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
97 |
-
)
|
98 |
-
(drop_path): Identity()
|
99 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
100 |
-
(mlp): Mlp(
|
101 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
102 |
-
(act): GELU(approximate='none')
|
103 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
104 |
-
(drop): Dropout(p=0.0, inplace=False)
|
105 |
-
)
|
106 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
107 |
-
)
|
108 |
-
(2): BlockWithMasking(
|
109 |
-
(attn): MultiheadAttention(
|
110 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
111 |
-
)
|
112 |
-
(drop_path): Identity()
|
113 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
114 |
-
(mlp): Mlp(
|
115 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
116 |
-
(act): GELU(approximate='none')
|
117 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
118 |
-
(drop): Dropout(p=0.0, inplace=False)
|
119 |
-
)
|
120 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
121 |
-
)
|
122 |
-
(3): BlockWithMasking(
|
123 |
-
(attn): MultiheadAttention(
|
124 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
125 |
-
)
|
126 |
-
(drop_path): Identity()
|
127 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
128 |
-
(mlp): Mlp(
|
129 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
130 |
-
(act): GELU(approximate='none')
|
131 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
132 |
-
(drop): Dropout(p=0.0, inplace=False)
|
133 |
-
)
|
134 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
135 |
-
)
|
136 |
-
(4): BlockWithMasking(
|
137 |
-
(attn): MultiheadAttention(
|
138 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
139 |
-
)
|
140 |
-
(drop_path): Identity()
|
141 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
142 |
-
(mlp): Mlp(
|
143 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
144 |
-
(act): GELU(approximate='none')
|
145 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
146 |
-
(drop): Dropout(p=0.0, inplace=False)
|
147 |
-
)
|
148 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
149 |
-
)
|
150 |
-
(5): BlockWithMasking(
|
151 |
-
(attn): MultiheadAttention(
|
152 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
153 |
-
)
|
154 |
-
(drop_path): Identity()
|
155 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
156 |
-
(mlp): Mlp(
|
157 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
158 |
-
(act): GELU(approximate='none')
|
159 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
160 |
-
(drop): Dropout(p=0.0, inplace=False)
|
161 |
-
)
|
162 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
163 |
-
)
|
164 |
-
(6): BlockWithMasking(
|
165 |
-
(attn): MultiheadAttention(
|
166 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
167 |
-
)
|
168 |
-
(drop_path): Identity()
|
169 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
170 |
-
(mlp): Mlp(
|
171 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
172 |
-
(act): GELU(approximate='none')
|
173 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
174 |
-
(drop): Dropout(p=0.0, inplace=False)
|
175 |
-
)
|
176 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
177 |
-
)
|
178 |
-
(7): BlockWithMasking(
|
179 |
-
(attn): MultiheadAttention(
|
180 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
181 |
-
)
|
182 |
-
(drop_path): Identity()
|
183 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
184 |
-
(mlp): Mlp(
|
185 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
186 |
-
(act): GELU(approximate='none')
|
187 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
188 |
-
(drop): Dropout(p=0.0, inplace=False)
|
189 |
-
)
|
190 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
191 |
-
)
|
192 |
-
(8): BlockWithMasking(
|
193 |
-
(attn): MultiheadAttention(
|
194 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
195 |
-
)
|
196 |
-
(drop_path): Identity()
|
197 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
198 |
-
(mlp): Mlp(
|
199 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
200 |
-
(act): GELU(approximate='none')
|
201 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
202 |
-
(drop): Dropout(p=0.0, inplace=False)
|
203 |
-
)
|
204 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
205 |
-
)
|
206 |
-
(9): BlockWithMasking(
|
207 |
-
(attn): MultiheadAttention(
|
208 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
209 |
-
)
|
210 |
-
(drop_path): Identity()
|
211 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
212 |
-
(mlp): Mlp(
|
213 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
214 |
-
(act): GELU(approximate='none')
|
215 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
216 |
-
(drop): Dropout(p=0.0, inplace=False)
|
217 |
-
)
|
218 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
219 |
-
)
|
220 |
-
(10): BlockWithMasking(
|
221 |
-
(attn): MultiheadAttention(
|
222 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
223 |
-
)
|
224 |
-
(drop_path): Identity()
|
225 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
226 |
-
(mlp): Mlp(
|
227 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
228 |
-
(act): GELU(approximate='none')
|
229 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
230 |
-
(drop): Dropout(p=0.0, inplace=False)
|
231 |
-
)
|
232 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
233 |
-
)
|
234 |
-
(11): BlockWithMasking(
|
235 |
-
(attn): MultiheadAttention(
|
236 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
237 |
-
)
|
238 |
-
(drop_path): Identity()
|
239 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
240 |
-
(mlp): Mlp(
|
241 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
242 |
-
(act): GELU(approximate='none')
|
243 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
244 |
-
(drop): Dropout(p=0.0, inplace=False)
|
245 |
-
)
|
246 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
247 |
-
)
|
248 |
-
(12): BlockWithMasking(
|
249 |
-
(attn): MultiheadAttention(
|
250 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
251 |
-
)
|
252 |
-
(drop_path): Identity()
|
253 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
254 |
-
(mlp): Mlp(
|
255 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
256 |
-
(act): GELU(approximate='none')
|
257 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
258 |
-
(drop): Dropout(p=0.0, inplace=False)
|
259 |
-
)
|
260 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
261 |
-
)
|
262 |
-
(13): BlockWithMasking(
|
263 |
-
(attn): MultiheadAttention(
|
264 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
265 |
-
)
|
266 |
-
(drop_path): Identity()
|
267 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
268 |
-
(mlp): Mlp(
|
269 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
270 |
-
(act): GELU(approximate='none')
|
271 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
272 |
-
(drop): Dropout(p=0.0, inplace=False)
|
273 |
-
)
|
274 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
275 |
-
)
|
276 |
-
(14): BlockWithMasking(
|
277 |
-
(attn): MultiheadAttention(
|
278 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
279 |
-
)
|
280 |
-
(drop_path): Identity()
|
281 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
282 |
-
(mlp): Mlp(
|
283 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
284 |
-
(act): GELU(approximate='none')
|
285 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
286 |
-
(drop): Dropout(p=0.0, inplace=False)
|
287 |
-
)
|
288 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
289 |
-
)
|
290 |
-
(15): BlockWithMasking(
|
291 |
-
(attn): MultiheadAttention(
|
292 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
293 |
-
)
|
294 |
-
(drop_path): Identity()
|
295 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
296 |
-
(mlp): Mlp(
|
297 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
298 |
-
(act): GELU(approximate='none')
|
299 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
300 |
-
(drop): Dropout(p=0.0, inplace=False)
|
301 |
-
)
|
302 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
303 |
-
)
|
304 |
-
(16): BlockWithMasking(
|
305 |
-
(attn): MultiheadAttention(
|
306 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
307 |
-
)
|
308 |
-
(drop_path): Identity()
|
309 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
310 |
-
(mlp): Mlp(
|
311 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
312 |
-
(act): GELU(approximate='none')
|
313 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
314 |
-
(drop): Dropout(p=0.0, inplace=False)
|
315 |
-
)
|
316 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
317 |
-
)
|
318 |
-
(17): BlockWithMasking(
|
319 |
-
(attn): MultiheadAttention(
|
320 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
321 |
-
)
|
322 |
-
(drop_path): Identity()
|
323 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
324 |
-
(mlp): Mlp(
|
325 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
326 |
-
(act): GELU(approximate='none')
|
327 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
328 |
-
(drop): Dropout(p=0.0, inplace=False)
|
329 |
-
)
|
330 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
331 |
-
)
|
332 |
-
(18): BlockWithMasking(
|
333 |
-
(attn): MultiheadAttention(
|
334 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
335 |
-
)
|
336 |
-
(drop_path): Identity()
|
337 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
338 |
-
(mlp): Mlp(
|
339 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
340 |
-
(act): GELU(approximate='none')
|
341 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
342 |
-
(drop): Dropout(p=0.0, inplace=False)
|
343 |
-
)
|
344 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
345 |
-
)
|
346 |
-
(19): BlockWithMasking(
|
347 |
-
(attn): MultiheadAttention(
|
348 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
349 |
-
)
|
350 |
-
(drop_path): Identity()
|
351 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
352 |
-
(mlp): Mlp(
|
353 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
354 |
-
(act): GELU(approximate='none')
|
355 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
356 |
-
(drop): Dropout(p=0.0, inplace=False)
|
357 |
-
)
|
358 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
359 |
-
)
|
360 |
-
(20): BlockWithMasking(
|
361 |
-
(attn): MultiheadAttention(
|
362 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
363 |
-
)
|
364 |
-
(drop_path): Identity()
|
365 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
366 |
-
(mlp): Mlp(
|
367 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
368 |
-
(act): GELU(approximate='none')
|
369 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
370 |
-
(drop): Dropout(p=0.0, inplace=False)
|
371 |
-
)
|
372 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
373 |
-
)
|
374 |
-
(21): BlockWithMasking(
|
375 |
-
(attn): MultiheadAttention(
|
376 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
377 |
-
)
|
378 |
-
(drop_path): Identity()
|
379 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
380 |
-
(mlp): Mlp(
|
381 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
382 |
-
(act): GELU(approximate='none')
|
383 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
384 |
-
(drop): Dropout(p=0.0, inplace=False)
|
385 |
-
)
|
386 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
387 |
-
)
|
388 |
-
(22): BlockWithMasking(
|
389 |
-
(attn): MultiheadAttention(
|
390 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
391 |
-
)
|
392 |
-
(drop_path): Identity()
|
393 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
394 |
-
(mlp): Mlp(
|
395 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
396 |
-
(act): GELU(approximate='none')
|
397 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
398 |
-
(drop): Dropout(p=0.0, inplace=False)
|
399 |
-
)
|
400 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
401 |
-
)
|
402 |
-
(23): BlockWithMasking(
|
403 |
-
(attn): MultiheadAttention(
|
404 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
405 |
-
)
|
406 |
-
(drop_path): Identity()
|
407 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
408 |
-
(mlp): Mlp(
|
409 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
410 |
-
(act): GELU(approximate='none')
|
411 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
412 |
-
(drop): Dropout(p=0.0, inplace=False)
|
413 |
-
)
|
414 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
415 |
-
)
|
416 |
-
(24): BlockWithMasking(
|
417 |
-
(attn): MultiheadAttention(
|
418 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
419 |
-
)
|
420 |
-
(drop_path): Identity()
|
421 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
422 |
-
(mlp): Mlp(
|
423 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
424 |
-
(act): GELU(approximate='none')
|
425 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
426 |
-
(drop): Dropout(p=0.0, inplace=False)
|
427 |
-
)
|
428 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
429 |
-
)
|
430 |
-
(25): BlockWithMasking(
|
431 |
-
(attn): MultiheadAttention(
|
432 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
433 |
-
)
|
434 |
-
(drop_path): Identity()
|
435 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
436 |
-
(mlp): Mlp(
|
437 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
438 |
-
(act): GELU(approximate='none')
|
439 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
440 |
-
(drop): Dropout(p=0.0, inplace=False)
|
441 |
-
)
|
442 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
443 |
-
)
|
444 |
-
(26): BlockWithMasking(
|
445 |
-
(attn): MultiheadAttention(
|
446 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
447 |
-
)
|
448 |
-
(drop_path): Identity()
|
449 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
450 |
-
(mlp): Mlp(
|
451 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
452 |
-
(act): GELU(approximate='none')
|
453 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
454 |
-
(drop): Dropout(p=0.0, inplace=False)
|
455 |
-
)
|
456 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
457 |
-
)
|
458 |
-
(27): BlockWithMasking(
|
459 |
-
(attn): MultiheadAttention(
|
460 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
461 |
-
)
|
462 |
-
(drop_path): Identity()
|
463 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
464 |
-
(mlp): Mlp(
|
465 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
466 |
-
(act): GELU(approximate='none')
|
467 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
468 |
-
(drop): Dropout(p=0.0, inplace=False)
|
469 |
-
)
|
470 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
471 |
-
)
|
472 |
-
(28): BlockWithMasking(
|
473 |
-
(attn): MultiheadAttention(
|
474 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
475 |
-
)
|
476 |
-
(drop_path): Identity()
|
477 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
478 |
-
(mlp): Mlp(
|
479 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
480 |
-
(act): GELU(approximate='none')
|
481 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
482 |
-
(drop): Dropout(p=0.0, inplace=False)
|
483 |
-
)
|
484 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
485 |
-
)
|
486 |
-
(29): BlockWithMasking(
|
487 |
-
(attn): MultiheadAttention(
|
488 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
489 |
-
)
|
490 |
-
(drop_path): Identity()
|
491 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
492 |
-
(mlp): Mlp(
|
493 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
494 |
-
(act): GELU(approximate='none')
|
495 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
496 |
-
(drop): Dropout(p=0.0, inplace=False)
|
497 |
-
)
|
498 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
499 |
-
)
|
500 |
-
(30): BlockWithMasking(
|
501 |
-
(attn): MultiheadAttention(
|
502 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
503 |
-
)
|
504 |
-
(drop_path): Identity()
|
505 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
506 |
-
(mlp): Mlp(
|
507 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
508 |
-
(act): GELU(approximate='none')
|
509 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
510 |
-
(drop): Dropout(p=0.0, inplace=False)
|
511 |
-
)
|
512 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
513 |
-
)
|
514 |
-
(31): BlockWithMasking(
|
515 |
-
(attn): MultiheadAttention(
|
516 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1280, out_features=1280, bias=True)
|
517 |
-
)
|
518 |
-
(drop_path): Identity()
|
519 |
-
(norm_1): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
520 |
-
(mlp): Mlp(
|
521 |
-
(fc1): Linear(in_features=1280, out_features=5120, bias=True)
|
522 |
-
(act): GELU(approximate='none')
|
523 |
-
(fc2): Linear(in_features=5120, out_features=1280, bias=True)
|
524 |
-
(drop): Dropout(p=0.0, inplace=False)
|
525 |
-
)
|
526 |
-
(norm_2): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
527 |
-
)
|
528 |
-
)
|
529 |
-
(post_transformer_layer): EinOpsRearrange()
|
530 |
-
)
|
531 |
-
(text): SimpleTransformer(
|
532 |
-
(pre_transformer_layer): Sequential(
|
533 |
-
(0): Identity()
|
534 |
-
(1): EinOpsRearrange()
|
535 |
-
)
|
536 |
-
(blocks): Sequential(
|
537 |
-
(0): BlockWithMasking(
|
538 |
-
(attn): MultiheadAttention(
|
539 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
540 |
-
)
|
541 |
-
(drop_path): Identity()
|
542 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
543 |
-
(mlp): Mlp(
|
544 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
545 |
-
(act): GELU(approximate='none')
|
546 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
547 |
-
(drop): Dropout(p=0.0, inplace=False)
|
548 |
-
)
|
549 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
550 |
-
)
|
551 |
-
(1): BlockWithMasking(
|
552 |
-
(attn): MultiheadAttention(
|
553 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
554 |
-
)
|
555 |
-
(drop_path): Identity()
|
556 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
557 |
-
(mlp): Mlp(
|
558 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
559 |
-
(act): GELU(approximate='none')
|
560 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
561 |
-
(drop): Dropout(p=0.0, inplace=False)
|
562 |
-
)
|
563 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
564 |
-
)
|
565 |
-
(2): BlockWithMasking(
|
566 |
-
(attn): MultiheadAttention(
|
567 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
568 |
-
)
|
569 |
-
(drop_path): Identity()
|
570 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
571 |
-
(mlp): Mlp(
|
572 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
573 |
-
(act): GELU(approximate='none')
|
574 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
575 |
-
(drop): Dropout(p=0.0, inplace=False)
|
576 |
-
)
|
577 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
578 |
-
)
|
579 |
-
(3): BlockWithMasking(
|
580 |
-
(attn): MultiheadAttention(
|
581 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
582 |
-
)
|
583 |
-
(drop_path): Identity()
|
584 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
585 |
-
(mlp): Mlp(
|
586 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
587 |
-
(act): GELU(approximate='none')
|
588 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
589 |
-
(drop): Dropout(p=0.0, inplace=False)
|
590 |
-
)
|
591 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
592 |
-
)
|
593 |
-
(4): BlockWithMasking(
|
594 |
-
(attn): MultiheadAttention(
|
595 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
596 |
-
)
|
597 |
-
(drop_path): Identity()
|
598 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
599 |
-
(mlp): Mlp(
|
600 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
601 |
-
(act): GELU(approximate='none')
|
602 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
603 |
-
(drop): Dropout(p=0.0, inplace=False)
|
604 |
-
)
|
605 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
606 |
-
)
|
607 |
-
(5): BlockWithMasking(
|
608 |
-
(attn): MultiheadAttention(
|
609 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
610 |
-
)
|
611 |
-
(drop_path): Identity()
|
612 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
613 |
-
(mlp): Mlp(
|
614 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
615 |
-
(act): GELU(approximate='none')
|
616 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
617 |
-
(drop): Dropout(p=0.0, inplace=False)
|
618 |
-
)
|
619 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
620 |
-
)
|
621 |
-
(6): BlockWithMasking(
|
622 |
-
(attn): MultiheadAttention(
|
623 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
624 |
-
)
|
625 |
-
(drop_path): Identity()
|
626 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
627 |
-
(mlp): Mlp(
|
628 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
629 |
-
(act): GELU(approximate='none')
|
630 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
631 |
-
(drop): Dropout(p=0.0, inplace=False)
|
632 |
-
)
|
633 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
634 |
-
)
|
635 |
-
(7): BlockWithMasking(
|
636 |
-
(attn): MultiheadAttention(
|
637 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
638 |
-
)
|
639 |
-
(drop_path): Identity()
|
640 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
641 |
-
(mlp): Mlp(
|
642 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
643 |
-
(act): GELU(approximate='none')
|
644 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
645 |
-
(drop): Dropout(p=0.0, inplace=False)
|
646 |
-
)
|
647 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
648 |
-
)
|
649 |
-
(8): BlockWithMasking(
|
650 |
-
(attn): MultiheadAttention(
|
651 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
652 |
-
)
|
653 |
-
(drop_path): Identity()
|
654 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
655 |
-
(mlp): Mlp(
|
656 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
657 |
-
(act): GELU(approximate='none')
|
658 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
659 |
-
(drop): Dropout(p=0.0, inplace=False)
|
660 |
-
)
|
661 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
662 |
-
)
|
663 |
-
(9): BlockWithMasking(
|
664 |
-
(attn): MultiheadAttention(
|
665 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
666 |
-
)
|
667 |
-
(drop_path): Identity()
|
668 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
669 |
-
(mlp): Mlp(
|
670 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
671 |
-
(act): GELU(approximate='none')
|
672 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
673 |
-
(drop): Dropout(p=0.0, inplace=False)
|
674 |
-
)
|
675 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
676 |
-
)
|
677 |
-
(10): BlockWithMasking(
|
678 |
-
(attn): MultiheadAttention(
|
679 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
680 |
-
)
|
681 |
-
(drop_path): Identity()
|
682 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
683 |
-
(mlp): Mlp(
|
684 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
685 |
-
(act): GELU(approximate='none')
|
686 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
687 |
-
(drop): Dropout(p=0.0, inplace=False)
|
688 |
-
)
|
689 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
690 |
-
)
|
691 |
-
(11): BlockWithMasking(
|
692 |
-
(attn): MultiheadAttention(
|
693 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
694 |
-
)
|
695 |
-
(drop_path): Identity()
|
696 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
697 |
-
(mlp): Mlp(
|
698 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
699 |
-
(act): GELU(approximate='none')
|
700 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
701 |
-
(drop): Dropout(p=0.0, inplace=False)
|
702 |
-
)
|
703 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
704 |
-
)
|
705 |
-
(12): BlockWithMasking(
|
706 |
-
(attn): MultiheadAttention(
|
707 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
708 |
-
)
|
709 |
-
(drop_path): Identity()
|
710 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
711 |
-
(mlp): Mlp(
|
712 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
713 |
-
(act): GELU(approximate='none')
|
714 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
715 |
-
(drop): Dropout(p=0.0, inplace=False)
|
716 |
-
)
|
717 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
718 |
-
)
|
719 |
-
(13): BlockWithMasking(
|
720 |
-
(attn): MultiheadAttention(
|
721 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
722 |
-
)
|
723 |
-
(drop_path): Identity()
|
724 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
725 |
-
(mlp): Mlp(
|
726 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
727 |
-
(act): GELU(approximate='none')
|
728 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
729 |
-
(drop): Dropout(p=0.0, inplace=False)
|
730 |
-
)
|
731 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
732 |
-
)
|
733 |
-
(14): BlockWithMasking(
|
734 |
-
(attn): MultiheadAttention(
|
735 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
736 |
-
)
|
737 |
-
(drop_path): Identity()
|
738 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
739 |
-
(mlp): Mlp(
|
740 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
741 |
-
(act): GELU(approximate='none')
|
742 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
743 |
-
(drop): Dropout(p=0.0, inplace=False)
|
744 |
-
)
|
745 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
746 |
-
)
|
747 |
-
(15): BlockWithMasking(
|
748 |
-
(attn): MultiheadAttention(
|
749 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
750 |
-
)
|
751 |
-
(drop_path): Identity()
|
752 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
753 |
-
(mlp): Mlp(
|
754 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
755 |
-
(act): GELU(approximate='none')
|
756 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
757 |
-
(drop): Dropout(p=0.0, inplace=False)
|
758 |
-
)
|
759 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
760 |
-
)
|
761 |
-
(16): BlockWithMasking(
|
762 |
-
(attn): MultiheadAttention(
|
763 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
764 |
-
)
|
765 |
-
(drop_path): Identity()
|
766 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
767 |
-
(mlp): Mlp(
|
768 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
769 |
-
(act): GELU(approximate='none')
|
770 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
771 |
-
(drop): Dropout(p=0.0, inplace=False)
|
772 |
-
)
|
773 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
774 |
-
)
|
775 |
-
(17): BlockWithMasking(
|
776 |
-
(attn): MultiheadAttention(
|
777 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
778 |
-
)
|
779 |
-
(drop_path): Identity()
|
780 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
781 |
-
(mlp): Mlp(
|
782 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
783 |
-
(act): GELU(approximate='none')
|
784 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
785 |
-
(drop): Dropout(p=0.0, inplace=False)
|
786 |
-
)
|
787 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
788 |
-
)
|
789 |
-
(18): BlockWithMasking(
|
790 |
-
(attn): MultiheadAttention(
|
791 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
792 |
-
)
|
793 |
-
(drop_path): Identity()
|
794 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
795 |
-
(mlp): Mlp(
|
796 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
797 |
-
(act): GELU(approximate='none')
|
798 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
799 |
-
(drop): Dropout(p=0.0, inplace=False)
|
800 |
-
)
|
801 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
802 |
-
)
|
803 |
-
(19): BlockWithMasking(
|
804 |
-
(attn): MultiheadAttention(
|
805 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
806 |
-
)
|
807 |
-
(drop_path): Identity()
|
808 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
809 |
-
(mlp): Mlp(
|
810 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
811 |
-
(act): GELU(approximate='none')
|
812 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
813 |
-
(drop): Dropout(p=0.0, inplace=False)
|
814 |
-
)
|
815 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
816 |
-
)
|
817 |
-
(20): BlockWithMasking(
|
818 |
-
(attn): MultiheadAttention(
|
819 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
820 |
-
)
|
821 |
-
(drop_path): Identity()
|
822 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
823 |
-
(mlp): Mlp(
|
824 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
825 |
-
(act): GELU(approximate='none')
|
826 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
827 |
-
(drop): Dropout(p=0.0, inplace=False)
|
828 |
-
)
|
829 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
830 |
-
)
|
831 |
-
(21): BlockWithMasking(
|
832 |
-
(attn): MultiheadAttention(
|
833 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
834 |
-
)
|
835 |
-
(drop_path): Identity()
|
836 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
837 |
-
(mlp): Mlp(
|
838 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
839 |
-
(act): GELU(approximate='none')
|
840 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
841 |
-
(drop): Dropout(p=0.0, inplace=False)
|
842 |
-
)
|
843 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
844 |
-
)
|
845 |
-
(22): BlockWithMasking(
|
846 |
-
(attn): MultiheadAttention(
|
847 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
848 |
-
)
|
849 |
-
(drop_path): Identity()
|
850 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
851 |
-
(mlp): Mlp(
|
852 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
853 |
-
(act): GELU(approximate='none')
|
854 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
855 |
-
(drop): Dropout(p=0.0, inplace=False)
|
856 |
-
)
|
857 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
858 |
-
)
|
859 |
-
(23): BlockWithMasking(
|
860 |
-
(attn): MultiheadAttention(
|
861 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=1024, out_features=1024, bias=True)
|
862 |
-
)
|
863 |
-
(drop_path): Identity()
|
864 |
-
(norm_1): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
865 |
-
(mlp): Mlp(
|
866 |
-
(fc1): Linear(in_features=1024, out_features=4096, bias=True)
|
867 |
-
(act): GELU(approximate='none')
|
868 |
-
(fc2): Linear(in_features=4096, out_features=1024, bias=True)
|
869 |
-
(drop): Dropout(p=0.0, inplace=False)
|
870 |
-
)
|
871 |
-
(norm_2): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
872 |
-
)
|
873 |
-
)
|
874 |
-
(post_transformer_layer): EinOpsRearrange()
|
875 |
-
)
|
876 |
-
(audio): SimpleTransformer(
|
877 |
-
(pre_transformer_layer): Sequential(
|
878 |
-
(0): Identity()
|
879 |
-
(1): EinOpsRearrange()
|
880 |
-
)
|
881 |
-
(blocks): Sequential(
|
882 |
-
(0): BlockWithMasking(
|
883 |
-
(attn): MultiheadAttention(
|
884 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
885 |
-
)
|
886 |
-
(drop_path): Identity()
|
887 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
888 |
-
(mlp): Mlp(
|
889 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
890 |
-
(act): GELU(approximate='none')
|
891 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
892 |
-
(drop): Dropout(p=0.0, inplace=False)
|
893 |
-
)
|
894 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
895 |
-
)
|
896 |
-
(1): BlockWithMasking(
|
897 |
-
(attn): MultiheadAttention(
|
898 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
899 |
-
)
|
900 |
-
(drop_path): DropPath(drop_prob=0.009)
|
901 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
902 |
-
(mlp): Mlp(
|
903 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
904 |
-
(act): GELU(approximate='none')
|
905 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
906 |
-
(drop): Dropout(p=0.0, inplace=False)
|
907 |
-
)
|
908 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
909 |
-
)
|
910 |
-
(2): BlockWithMasking(
|
911 |
-
(attn): MultiheadAttention(
|
912 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
913 |
-
)
|
914 |
-
(drop_path): DropPath(drop_prob=0.018)
|
915 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
916 |
-
(mlp): Mlp(
|
917 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
918 |
-
(act): GELU(approximate='none')
|
919 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
920 |
-
(drop): Dropout(p=0.0, inplace=False)
|
921 |
-
)
|
922 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
923 |
-
)
|
924 |
-
(3): BlockWithMasking(
|
925 |
-
(attn): MultiheadAttention(
|
926 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
927 |
-
)
|
928 |
-
(drop_path): DropPath(drop_prob=0.027)
|
929 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
930 |
-
(mlp): Mlp(
|
931 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
932 |
-
(act): GELU(approximate='none')
|
933 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
934 |
-
(drop): Dropout(p=0.0, inplace=False)
|
935 |
-
)
|
936 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
937 |
-
)
|
938 |
-
(4): BlockWithMasking(
|
939 |
-
(attn): MultiheadAttention(
|
940 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
941 |
-
)
|
942 |
-
(drop_path): DropPath(drop_prob=0.036)
|
943 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
944 |
-
(mlp): Mlp(
|
945 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
946 |
-
(act): GELU(approximate='none')
|
947 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
948 |
-
(drop): Dropout(p=0.0, inplace=False)
|
949 |
-
)
|
950 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
951 |
-
)
|
952 |
-
(5): BlockWithMasking(
|
953 |
-
(attn): MultiheadAttention(
|
954 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
955 |
-
)
|
956 |
-
(drop_path): DropPath(drop_prob=0.045)
|
957 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
958 |
-
(mlp): Mlp(
|
959 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
960 |
-
(act): GELU(approximate='none')
|
961 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
962 |
-
(drop): Dropout(p=0.0, inplace=False)
|
963 |
-
)
|
964 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
965 |
-
)
|
966 |
-
(6): BlockWithMasking(
|
967 |
-
(attn): MultiheadAttention(
|
968 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
969 |
-
)
|
970 |
-
(drop_path): DropPath(drop_prob=0.055)
|
971 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
972 |
-
(mlp): Mlp(
|
973 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
974 |
-
(act): GELU(approximate='none')
|
975 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
976 |
-
(drop): Dropout(p=0.0, inplace=False)
|
977 |
-
)
|
978 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
979 |
-
)
|
980 |
-
(7): BlockWithMasking(
|
981 |
-
(attn): MultiheadAttention(
|
982 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
983 |
-
)
|
984 |
-
(drop_path): DropPath(drop_prob=0.064)
|
985 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
986 |
-
(mlp): Mlp(
|
987 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
988 |
-
(act): GELU(approximate='none')
|
989 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
990 |
-
(drop): Dropout(p=0.0, inplace=False)
|
991 |
-
)
|
992 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
993 |
-
)
|
994 |
-
(8): BlockWithMasking(
|
995 |
-
(attn): MultiheadAttention(
|
996 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
997 |
-
)
|
998 |
-
(drop_path): DropPath(drop_prob=0.073)
|
999 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1000 |
-
(mlp): Mlp(
|
1001 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1002 |
-
(act): GELU(approximate='none')
|
1003 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1004 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1005 |
-
)
|
1006 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1007 |
-
)
|
1008 |
-
(9): BlockWithMasking(
|
1009 |
-
(attn): MultiheadAttention(
|
1010 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
1011 |
-
)
|
1012 |
-
(drop_path): DropPath(drop_prob=0.082)
|
1013 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1014 |
-
(mlp): Mlp(
|
1015 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1016 |
-
(act): GELU(approximate='none')
|
1017 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1018 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1019 |
-
)
|
1020 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1021 |
-
)
|
1022 |
-
(10): BlockWithMasking(
|
1023 |
-
(attn): MultiheadAttention(
|
1024 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
1025 |
-
)
|
1026 |
-
(drop_path): DropPath(drop_prob=0.091)
|
1027 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1028 |
-
(mlp): Mlp(
|
1029 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1030 |
-
(act): GELU(approximate='none')
|
1031 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1032 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1033 |
-
)
|
1034 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1035 |
-
)
|
1036 |
-
(11): BlockWithMasking(
|
1037 |
-
(attn): MultiheadAttention(
|
1038 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
1039 |
-
)
|
1040 |
-
(drop_path): DropPath(drop_prob=0.100)
|
1041 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1042 |
-
(mlp): Mlp(
|
1043 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1044 |
-
(act): GELU(approximate='none')
|
1045 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1046 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1047 |
-
)
|
1048 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1049 |
-
)
|
1050 |
-
)
|
1051 |
-
(post_transformer_layer): EinOpsRearrange()
|
1052 |
-
)
|
1053 |
-
(depth): SimpleTransformer(
|
1054 |
-
(pre_transformer_layer): Sequential(
|
1055 |
-
(0): Identity()
|
1056 |
-
(1): EinOpsRearrange()
|
1057 |
-
)
|
1058 |
-
(blocks): Sequential(
|
1059 |
-
(0): BlockWithMasking(
|
1060 |
-
(attn): MultiheadAttention(
|
1061 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
1062 |
-
)
|
1063 |
-
(drop_path): Identity()
|
1064 |
-
(norm_1): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1065 |
-
(mlp): Mlp(
|
1066 |
-
(fc1): Linear(in_features=384, out_features=1536, bias=True)
|
1067 |
-
(act): GELU(approximate='none')
|
1068 |
-
(fc2): Linear(in_features=1536, out_features=384, bias=True)
|
1069 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1070 |
-
)
|
1071 |
-
(norm_2): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1072 |
-
)
|
1073 |
-
(1): BlockWithMasking(
|
1074 |
-
(attn): MultiheadAttention(
|
1075 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
1076 |
-
)
|
1077 |
-
(drop_path): Identity()
|
1078 |
-
(norm_1): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1079 |
-
(mlp): Mlp(
|
1080 |
-
(fc1): Linear(in_features=384, out_features=1536, bias=True)
|
1081 |
-
(act): GELU(approximate='none')
|
1082 |
-
(fc2): Linear(in_features=1536, out_features=384, bias=True)
|
1083 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1084 |
-
)
|
1085 |
-
(norm_2): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1086 |
-
)
|
1087 |
-
(2): BlockWithMasking(
|
1088 |
-
(attn): MultiheadAttention(
|
1089 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
1090 |
-
)
|
1091 |
-
(drop_path): Identity()
|
1092 |
-
(norm_1): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1093 |
-
(mlp): Mlp(
|
1094 |
-
(fc1): Linear(in_features=384, out_features=1536, bias=True)
|
1095 |
-
(act): GELU(approximate='none')
|
1096 |
-
(fc2): Linear(in_features=1536, out_features=384, bias=True)
|
1097 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1098 |
-
)
|
1099 |
-
(norm_2): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1100 |
-
)
|
1101 |
-
(3): BlockWithMasking(
|
1102 |
-
(attn): MultiheadAttention(
|
1103 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
1104 |
-
)
|
1105 |
-
(drop_path): Identity()
|
1106 |
-
(norm_1): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1107 |
-
(mlp): Mlp(
|
1108 |
-
(fc1): Linear(in_features=384, out_features=1536, bias=True)
|
1109 |
-
(act): GELU(approximate='none')
|
1110 |
-
(fc2): Linear(in_features=1536, out_features=384, bias=True)
|
1111 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1112 |
-
)
|
1113 |
-
(norm_2): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1114 |
-
)
|
1115 |
-
(4): BlockWithMasking(
|
1116 |
-
(attn): MultiheadAttention(
|
1117 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
1118 |
-
)
|
1119 |
-
(drop_path): Identity()
|
1120 |
-
(norm_1): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1121 |
-
(mlp): Mlp(
|
1122 |
-
(fc1): Linear(in_features=384, out_features=1536, bias=True)
|
1123 |
-
(act): GELU(approximate='none')
|
1124 |
-
(fc2): Linear(in_features=1536, out_features=384, bias=True)
|
1125 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1126 |
-
)
|
1127 |
-
(norm_2): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1128 |
-
)
|
1129 |
-
(5): BlockWithMasking(
|
1130 |
-
(attn): MultiheadAttention(
|
1131 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
1132 |
-
)
|
1133 |
-
(drop_path): Identity()
|
1134 |
-
(norm_1): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1135 |
-
(mlp): Mlp(
|
1136 |
-
(fc1): Linear(in_features=384, out_features=1536, bias=True)
|
1137 |
-
(act): GELU(approximate='none')
|
1138 |
-
(fc2): Linear(in_features=1536, out_features=384, bias=True)
|
1139 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1140 |
-
)
|
1141 |
-
(norm_2): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1142 |
-
)
|
1143 |
-
(6): BlockWithMasking(
|
1144 |
-
(attn): MultiheadAttention(
|
1145 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
1146 |
-
)
|
1147 |
-
(drop_path): Identity()
|
1148 |
-
(norm_1): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1149 |
-
(mlp): Mlp(
|
1150 |
-
(fc1): Linear(in_features=384, out_features=1536, bias=True)
|
1151 |
-
(act): GELU(approximate='none')
|
1152 |
-
(fc2): Linear(in_features=1536, out_features=384, bias=True)
|
1153 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1154 |
-
)
|
1155 |
-
(norm_2): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1156 |
-
)
|
1157 |
-
(7): BlockWithMasking(
|
1158 |
-
(attn): MultiheadAttention(
|
1159 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
1160 |
-
)
|
1161 |
-
(drop_path): Identity()
|
1162 |
-
(norm_1): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1163 |
-
(mlp): Mlp(
|
1164 |
-
(fc1): Linear(in_features=384, out_features=1536, bias=True)
|
1165 |
-
(act): GELU(approximate='none')
|
1166 |
-
(fc2): Linear(in_features=1536, out_features=384, bias=True)
|
1167 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1168 |
-
)
|
1169 |
-
(norm_2): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1170 |
-
)
|
1171 |
-
(8): BlockWithMasking(
|
1172 |
-
(attn): MultiheadAttention(
|
1173 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
1174 |
-
)
|
1175 |
-
(drop_path): Identity()
|
1176 |
-
(norm_1): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1177 |
-
(mlp): Mlp(
|
1178 |
-
(fc1): Linear(in_features=384, out_features=1536, bias=True)
|
1179 |
-
(act): GELU(approximate='none')
|
1180 |
-
(fc2): Linear(in_features=1536, out_features=384, bias=True)
|
1181 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1182 |
-
)
|
1183 |
-
(norm_2): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1184 |
-
)
|
1185 |
-
(9): BlockWithMasking(
|
1186 |
-
(attn): MultiheadAttention(
|
1187 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
1188 |
-
)
|
1189 |
-
(drop_path): Identity()
|
1190 |
-
(norm_1): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1191 |
-
(mlp): Mlp(
|
1192 |
-
(fc1): Linear(in_features=384, out_features=1536, bias=True)
|
1193 |
-
(act): GELU(approximate='none')
|
1194 |
-
(fc2): Linear(in_features=1536, out_features=384, bias=True)
|
1195 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1196 |
-
)
|
1197 |
-
(norm_2): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1198 |
-
)
|
1199 |
-
(10): BlockWithMasking(
|
1200 |
-
(attn): MultiheadAttention(
|
1201 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
1202 |
-
)
|
1203 |
-
(drop_path): Identity()
|
1204 |
-
(norm_1): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1205 |
-
(mlp): Mlp(
|
1206 |
-
(fc1): Linear(in_features=384, out_features=1536, bias=True)
|
1207 |
-
(act): GELU(approximate='none')
|
1208 |
-
(fc2): Linear(in_features=1536, out_features=384, bias=True)
|
1209 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1210 |
-
)
|
1211 |
-
(norm_2): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1212 |
-
)
|
1213 |
-
(11): BlockWithMasking(
|
1214 |
-
(attn): MultiheadAttention(
|
1215 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
1216 |
-
)
|
1217 |
-
(drop_path): Identity()
|
1218 |
-
(norm_1): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1219 |
-
(mlp): Mlp(
|
1220 |
-
(fc1): Linear(in_features=384, out_features=1536, bias=True)
|
1221 |
-
(act): GELU(approximate='none')
|
1222 |
-
(fc2): Linear(in_features=1536, out_features=384, bias=True)
|
1223 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1224 |
-
)
|
1225 |
-
(norm_2): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1226 |
-
)
|
1227 |
-
)
|
1228 |
-
(post_transformer_layer): EinOpsRearrange()
|
1229 |
-
)
|
1230 |
-
(thermal): SimpleTransformer(
|
1231 |
-
(pre_transformer_layer): Sequential(
|
1232 |
-
(0): Identity()
|
1233 |
-
(1): EinOpsRearrange()
|
1234 |
-
)
|
1235 |
-
(blocks): Sequential(
|
1236 |
-
(0): BlockWithMasking(
|
1237 |
-
(attn): MultiheadAttention(
|
1238 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
1239 |
-
)
|
1240 |
-
(drop_path): Identity()
|
1241 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1242 |
-
(mlp): Mlp(
|
1243 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1244 |
-
(act): GELU(approximate='none')
|
1245 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1246 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1247 |
-
)
|
1248 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1249 |
-
)
|
1250 |
-
(1): BlockWithMasking(
|
1251 |
-
(attn): MultiheadAttention(
|
1252 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
1253 |
-
)
|
1254 |
-
(drop_path): Identity()
|
1255 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1256 |
-
(mlp): Mlp(
|
1257 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1258 |
-
(act): GELU(approximate='none')
|
1259 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1260 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1261 |
-
)
|
1262 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1263 |
-
)
|
1264 |
-
(2): BlockWithMasking(
|
1265 |
-
(attn): MultiheadAttention(
|
1266 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
1267 |
-
)
|
1268 |
-
(drop_path): Identity()
|
1269 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1270 |
-
(mlp): Mlp(
|
1271 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1272 |
-
(act): GELU(approximate='none')
|
1273 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1274 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1275 |
-
)
|
1276 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1277 |
-
)
|
1278 |
-
(3): BlockWithMasking(
|
1279 |
-
(attn): MultiheadAttention(
|
1280 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
1281 |
-
)
|
1282 |
-
(drop_path): Identity()
|
1283 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1284 |
-
(mlp): Mlp(
|
1285 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1286 |
-
(act): GELU(approximate='none')
|
1287 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1288 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1289 |
-
)
|
1290 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1291 |
-
)
|
1292 |
-
(4): BlockWithMasking(
|
1293 |
-
(attn): MultiheadAttention(
|
1294 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
1295 |
-
)
|
1296 |
-
(drop_path): Identity()
|
1297 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1298 |
-
(mlp): Mlp(
|
1299 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1300 |
-
(act): GELU(approximate='none')
|
1301 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1302 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1303 |
-
)
|
1304 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1305 |
-
)
|
1306 |
-
(5): BlockWithMasking(
|
1307 |
-
(attn): MultiheadAttention(
|
1308 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
1309 |
-
)
|
1310 |
-
(drop_path): Identity()
|
1311 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1312 |
-
(mlp): Mlp(
|
1313 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1314 |
-
(act): GELU(approximate='none')
|
1315 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1316 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1317 |
-
)
|
1318 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1319 |
-
)
|
1320 |
-
(6): BlockWithMasking(
|
1321 |
-
(attn): MultiheadAttention(
|
1322 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
1323 |
-
)
|
1324 |
-
(drop_path): Identity()
|
1325 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1326 |
-
(mlp): Mlp(
|
1327 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1328 |
-
(act): GELU(approximate='none')
|
1329 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1330 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1331 |
-
)
|
1332 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1333 |
-
)
|
1334 |
-
(7): BlockWithMasking(
|
1335 |
-
(attn): MultiheadAttention(
|
1336 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
1337 |
-
)
|
1338 |
-
(drop_path): Identity()
|
1339 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1340 |
-
(mlp): Mlp(
|
1341 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1342 |
-
(act): GELU(approximate='none')
|
1343 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1344 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1345 |
-
)
|
1346 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1347 |
-
)
|
1348 |
-
(8): BlockWithMasking(
|
1349 |
-
(attn): MultiheadAttention(
|
1350 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
1351 |
-
)
|
1352 |
-
(drop_path): Identity()
|
1353 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1354 |
-
(mlp): Mlp(
|
1355 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1356 |
-
(act): GELU(approximate='none')
|
1357 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1358 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1359 |
-
)
|
1360 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1361 |
-
)
|
1362 |
-
(9): BlockWithMasking(
|
1363 |
-
(attn): MultiheadAttention(
|
1364 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
1365 |
-
)
|
1366 |
-
(drop_path): Identity()
|
1367 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1368 |
-
(mlp): Mlp(
|
1369 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1370 |
-
(act): GELU(approximate='none')
|
1371 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1372 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1373 |
-
)
|
1374 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1375 |
-
)
|
1376 |
-
(10): BlockWithMasking(
|
1377 |
-
(attn): MultiheadAttention(
|
1378 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
1379 |
-
)
|
1380 |
-
(drop_path): Identity()
|
1381 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1382 |
-
(mlp): Mlp(
|
1383 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1384 |
-
(act): GELU(approximate='none')
|
1385 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1386 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1387 |
-
)
|
1388 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1389 |
-
)
|
1390 |
-
(11): BlockWithMasking(
|
1391 |
-
(attn): MultiheadAttention(
|
1392 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
1393 |
-
)
|
1394 |
-
(drop_path): Identity()
|
1395 |
-
(norm_1): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1396 |
-
(mlp): Mlp(
|
1397 |
-
(fc1): Linear(in_features=768, out_features=3072, bias=True)
|
1398 |
-
(act): GELU(approximate='none')
|
1399 |
-
(fc2): Linear(in_features=3072, out_features=768, bias=True)
|
1400 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1401 |
-
)
|
1402 |
-
(norm_2): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1403 |
-
)
|
1404 |
-
)
|
1405 |
-
(post_transformer_layer): EinOpsRearrange()
|
1406 |
-
)
|
1407 |
-
(imu): SimpleTransformer(
|
1408 |
-
(pre_transformer_layer): Sequential(
|
1409 |
-
(0): Identity()
|
1410 |
-
(1): EinOpsRearrange()
|
1411 |
-
)
|
1412 |
-
(blocks): Sequential(
|
1413 |
-
(0): BlockWithMasking(
|
1414 |
-
(attn): MultiheadAttention(
|
1415 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
|
1416 |
-
)
|
1417 |
-
(drop_path): Identity()
|
1418 |
-
(norm_1): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
|
1419 |
-
(mlp): Mlp(
|
1420 |
-
(fc1): Linear(in_features=512, out_features=2048, bias=True)
|
1421 |
-
(act): GELU(approximate='none')
|
1422 |
-
(fc2): Linear(in_features=2048, out_features=512, bias=True)
|
1423 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1424 |
-
)
|
1425 |
-
(norm_2): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
|
1426 |
-
)
|
1427 |
-
(1): BlockWithMasking(
|
1428 |
-
(attn): MultiheadAttention(
|
1429 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
|
1430 |
-
)
|
1431 |
-
(drop_path): DropPath(drop_prob=0.140)
|
1432 |
-
(norm_1): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
|
1433 |
-
(mlp): Mlp(
|
1434 |
-
(fc1): Linear(in_features=512, out_features=2048, bias=True)
|
1435 |
-
(act): GELU(approximate='none')
|
1436 |
-
(fc2): Linear(in_features=2048, out_features=512, bias=True)
|
1437 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1438 |
-
)
|
1439 |
-
(norm_2): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
|
1440 |
-
)
|
1441 |
-
(2): BlockWithMasking(
|
1442 |
-
(attn): MultiheadAttention(
|
1443 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
|
1444 |
-
)
|
1445 |
-
(drop_path): DropPath(drop_prob=0.280)
|
1446 |
-
(norm_1): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
|
1447 |
-
(mlp): Mlp(
|
1448 |
-
(fc1): Linear(in_features=512, out_features=2048, bias=True)
|
1449 |
-
(act): GELU(approximate='none')
|
1450 |
-
(fc2): Linear(in_features=2048, out_features=512, bias=True)
|
1451 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1452 |
-
)
|
1453 |
-
(norm_2): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
|
1454 |
-
)
|
1455 |
-
(3): BlockWithMasking(
|
1456 |
-
(attn): MultiheadAttention(
|
1457 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
|
1458 |
-
)
|
1459 |
-
(drop_path): DropPath(drop_prob=0.420)
|
1460 |
-
(norm_1): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
|
1461 |
-
(mlp): Mlp(
|
1462 |
-
(fc1): Linear(in_features=512, out_features=2048, bias=True)
|
1463 |
-
(act): GELU(approximate='none')
|
1464 |
-
(fc2): Linear(in_features=2048, out_features=512, bias=True)
|
1465 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1466 |
-
)
|
1467 |
-
(norm_2): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
|
1468 |
-
)
|
1469 |
-
(4): BlockWithMasking(
|
1470 |
-
(attn): MultiheadAttention(
|
1471 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
|
1472 |
-
)
|
1473 |
-
(drop_path): DropPath(drop_prob=0.560)
|
1474 |
-
(norm_1): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
|
1475 |
-
(mlp): Mlp(
|
1476 |
-
(fc1): Linear(in_features=512, out_features=2048, bias=True)
|
1477 |
-
(act): GELU(approximate='none')
|
1478 |
-
(fc2): Linear(in_features=2048, out_features=512, bias=True)
|
1479 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1480 |
-
)
|
1481 |
-
(norm_2): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
|
1482 |
-
)
|
1483 |
-
(5): BlockWithMasking(
|
1484 |
-
(attn): MultiheadAttention(
|
1485 |
-
(out_proj): NonDynamicallyQuantizableLinear(in_features=512, out_features=512, bias=True)
|
1486 |
-
)
|
1487 |
-
(drop_path): DropPath(drop_prob=0.700)
|
1488 |
-
(norm_1): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
|
1489 |
-
(mlp): Mlp(
|
1490 |
-
(fc1): Linear(in_features=512, out_features=2048, bias=True)
|
1491 |
-
(act): GELU(approximate='none')
|
1492 |
-
(fc2): Linear(in_features=2048, out_features=512, bias=True)
|
1493 |
-
(drop): Dropout(p=0.0, inplace=False)
|
1494 |
-
)
|
1495 |
-
(norm_2): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
|
1496 |
-
)
|
1497 |
-
)
|
1498 |
-
(post_transformer_layer): EinOpsRearrange()
|
1499 |
-
)
|
1500 |
-
)
|
1501 |
-
(modality_heads): ModuleDict(
|
1502 |
-
(vision): Sequential(
|
1503 |
-
(0): LayerNorm((1280,), eps=1e-06, elementwise_affine=True)
|
1504 |
-
(1): SelectElement()
|
1505 |
-
(2): Linear(in_features=1280, out_features=1024, bias=False)
|
1506 |
-
)
|
1507 |
-
(text): SelectEOSAndProject(
|
1508 |
-
(proj): Sequential(
|
1509 |
-
(0): LayerNorm((1024,), eps=1e-06, elementwise_affine=True)
|
1510 |
-
(1): Linear(in_features=1024, out_features=1024, bias=False)
|
1511 |
-
)
|
1512 |
-
)
|
1513 |
-
(audio): Sequential(
|
1514 |
-
(0): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1515 |
-
(1): SelectElement()
|
1516 |
-
(2): Linear(in_features=768, out_features=1024, bias=False)
|
1517 |
-
)
|
1518 |
-
(depth): Sequential(
|
1519 |
-
(0): LayerNorm((384,), eps=1e-06, elementwise_affine=True)
|
1520 |
-
(1): SelectElement()
|
1521 |
-
(2): Linear(in_features=384, out_features=1024, bias=False)
|
1522 |
-
)
|
1523 |
-
(thermal): Sequential(
|
1524 |
-
(0): LayerNorm((768,), eps=1e-06, elementwise_affine=True)
|
1525 |
-
(1): SelectElement()
|
1526 |
-
(2): Linear(in_features=768, out_features=1024, bias=False)
|
1527 |
-
)
|
1528 |
-
(imu): Sequential(
|
1529 |
-
(0): LayerNorm((512,), eps=1e-06, elementwise_affine=True)
|
1530 |
-
(1): SelectElement()
|
1531 |
-
(2): Dropout(p=0.5, inplace=False)
|
1532 |
-
(3): Linear(in_features=512, out_features=1024, bias=False)
|
1533 |
-
)
|
1534 |
-
)
|
1535 |
-
(modality_postprocessors): ModuleDict(
|
1536 |
-
(vision): Normalize()
|
1537 |
-
(text): Sequential(
|
1538 |
-
(0): Normalize()
|
1539 |
-
(1): LearnableLogitScaling(logit_scale_init=14.285714285714285,learnable=True, max_logit_scale=100)
|
1540 |
-
)
|
1541 |
-
(audio): Sequential(
|
1542 |
-
(0): Normalize()
|
1543 |
-
(1): LearnableLogitScaling(logit_scale_init=20.0,learnable=False, max_logit_scale=100)
|
1544 |
-
)
|
1545 |
-
(depth): Sequential(
|
1546 |
-
(0): Normalize()
|
1547 |
-
(1): LearnableLogitScaling(logit_scale_init=5.0,learnable=False, max_logit_scale=100)
|
1548 |
-
)
|
1549 |
-
(thermal): Sequential(
|
1550 |
-
(0): Normalize()
|
1551 |
-
(1): LearnableLogitScaling(logit_scale_init=10.0,learnable=False, max_logit_scale=100)
|
1552 |
-
)
|
1553 |
-
(imu): Sequential(
|
1554 |
-
(0): Normalize()
|
1555 |
-
(1): LearnableLogitScaling(logit_scale_init=5.0,learnable=False, max_logit_scale=100)
|
1556 |
-
)
|
1557 |
-
)
|
1558 |
-
)
|
1559 |
-
(reasoner): Qwen2ForCausalLM(
|
1560 |
-
(model): Qwen2Model(
|
1561 |
-
(embed_tokens): Embedding(151936, 896)
|
1562 |
-
(layers): ModuleList(
|
1563 |
-
(0): Qwen2DecoderLayer(
|
1564 |
-
(self_attn): Qwen2Attention(
|
1565 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1566 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1567 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1568 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1569 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1570 |
-
)
|
1571 |
-
(mlp): Qwen2MLP(
|
1572 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1573 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1574 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1575 |
-
(act_fn): SiLU()
|
1576 |
-
)
|
1577 |
-
(input_layernorm): Qwen2RMSNorm()
|
1578 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1579 |
-
)
|
1580 |
-
(1): Qwen2DecoderLayer(
|
1581 |
-
(self_attn): Qwen2Attention(
|
1582 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1583 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1584 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1585 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1586 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1587 |
-
)
|
1588 |
-
(mlp): Qwen2MLP(
|
1589 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1590 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1591 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1592 |
-
(act_fn): SiLU()
|
1593 |
-
)
|
1594 |
-
(input_layernorm): Qwen2RMSNorm()
|
1595 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1596 |
-
)
|
1597 |
-
(2): Qwen2DecoderLayer(
|
1598 |
-
(self_attn): Qwen2Attention(
|
1599 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1600 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1601 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1602 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1603 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1604 |
-
)
|
1605 |
-
(mlp): Qwen2MLP(
|
1606 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1607 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1608 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1609 |
-
(act_fn): SiLU()
|
1610 |
-
)
|
1611 |
-
(input_layernorm): Qwen2RMSNorm()
|
1612 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1613 |
-
)
|
1614 |
-
(3): Qwen2DecoderLayer(
|
1615 |
-
(self_attn): Qwen2Attention(
|
1616 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1617 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1618 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1619 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1620 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1621 |
-
)
|
1622 |
-
(mlp): Qwen2MLP(
|
1623 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1624 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1625 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1626 |
-
(act_fn): SiLU()
|
1627 |
-
)
|
1628 |
-
(input_layernorm): Qwen2RMSNorm()
|
1629 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1630 |
-
)
|
1631 |
-
(4): Qwen2DecoderLayer(
|
1632 |
-
(self_attn): Qwen2Attention(
|
1633 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1634 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1635 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1636 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1637 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1638 |
-
)
|
1639 |
-
(mlp): Qwen2MLP(
|
1640 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1641 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1642 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1643 |
-
(act_fn): SiLU()
|
1644 |
-
)
|
1645 |
-
(input_layernorm): Qwen2RMSNorm()
|
1646 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1647 |
-
)
|
1648 |
-
(5): Qwen2DecoderLayer(
|
1649 |
-
(self_attn): Qwen2Attention(
|
1650 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1651 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1652 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1653 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1654 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1655 |
-
)
|
1656 |
-
(mlp): Qwen2MLP(
|
1657 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1658 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1659 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1660 |
-
(act_fn): SiLU()
|
1661 |
-
)
|
1662 |
-
(input_layernorm): Qwen2RMSNorm()
|
1663 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1664 |
-
)
|
1665 |
-
(6): Qwen2DecoderLayer(
|
1666 |
-
(self_attn): Qwen2Attention(
|
1667 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1668 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1669 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1670 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1671 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1672 |
-
)
|
1673 |
-
(mlp): Qwen2MLP(
|
1674 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1675 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1676 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1677 |
-
(act_fn): SiLU()
|
1678 |
-
)
|
1679 |
-
(input_layernorm): Qwen2RMSNorm()
|
1680 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1681 |
-
)
|
1682 |
-
(7): Qwen2DecoderLayer(
|
1683 |
-
(self_attn): Qwen2Attention(
|
1684 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1685 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1686 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1687 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1688 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1689 |
-
)
|
1690 |
-
(mlp): Qwen2MLP(
|
1691 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1692 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1693 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1694 |
-
(act_fn): SiLU()
|
1695 |
-
)
|
1696 |
-
(input_layernorm): Qwen2RMSNorm()
|
1697 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1698 |
-
)
|
1699 |
-
(8): Qwen2DecoderLayer(
|
1700 |
-
(self_attn): Qwen2Attention(
|
1701 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1702 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1703 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1704 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1705 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1706 |
-
)
|
1707 |
-
(mlp): Qwen2MLP(
|
1708 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1709 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1710 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1711 |
-
(act_fn): SiLU()
|
1712 |
-
)
|
1713 |
-
(input_layernorm): Qwen2RMSNorm()
|
1714 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1715 |
-
)
|
1716 |
-
(9): Qwen2DecoderLayer(
|
1717 |
-
(self_attn): Qwen2Attention(
|
1718 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1719 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1720 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1721 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1722 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1723 |
-
)
|
1724 |
-
(mlp): Qwen2MLP(
|
1725 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1726 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1727 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1728 |
-
(act_fn): SiLU()
|
1729 |
-
)
|
1730 |
-
(input_layernorm): Qwen2RMSNorm()
|
1731 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1732 |
-
)
|
1733 |
-
(10): Qwen2DecoderLayer(
|
1734 |
-
(self_attn): Qwen2Attention(
|
1735 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1736 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1737 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1738 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1739 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1740 |
-
)
|
1741 |
-
(mlp): Qwen2MLP(
|
1742 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1743 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1744 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1745 |
-
(act_fn): SiLU()
|
1746 |
-
)
|
1747 |
-
(input_layernorm): Qwen2RMSNorm()
|
1748 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1749 |
-
)
|
1750 |
-
(11): Qwen2DecoderLayer(
|
1751 |
-
(self_attn): Qwen2Attention(
|
1752 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1753 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1754 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1755 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1756 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1757 |
-
)
|
1758 |
-
(mlp): Qwen2MLP(
|
1759 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1760 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1761 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1762 |
-
(act_fn): SiLU()
|
1763 |
-
)
|
1764 |
-
(input_layernorm): Qwen2RMSNorm()
|
1765 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1766 |
-
)
|
1767 |
-
(12): Qwen2DecoderLayer(
|
1768 |
-
(self_attn): Qwen2Attention(
|
1769 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1770 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1771 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1772 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1773 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1774 |
-
)
|
1775 |
-
(mlp): Qwen2MLP(
|
1776 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1777 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1778 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1779 |
-
(act_fn): SiLU()
|
1780 |
-
)
|
1781 |
-
(input_layernorm): Qwen2RMSNorm()
|
1782 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1783 |
-
)
|
1784 |
-
(13): Qwen2DecoderLayer(
|
1785 |
-
(self_attn): Qwen2Attention(
|
1786 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1787 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1788 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1789 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1790 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1791 |
-
)
|
1792 |
-
(mlp): Qwen2MLP(
|
1793 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1794 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1795 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1796 |
-
(act_fn): SiLU()
|
1797 |
-
)
|
1798 |
-
(input_layernorm): Qwen2RMSNorm()
|
1799 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1800 |
-
)
|
1801 |
-
(14): Qwen2DecoderLayer(
|
1802 |
-
(self_attn): Qwen2Attention(
|
1803 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1804 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1805 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1806 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1807 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1808 |
-
)
|
1809 |
-
(mlp): Qwen2MLP(
|
1810 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1811 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1812 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1813 |
-
(act_fn): SiLU()
|
1814 |
-
)
|
1815 |
-
(input_layernorm): Qwen2RMSNorm()
|
1816 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1817 |
-
)
|
1818 |
-
(15): Qwen2DecoderLayer(
|
1819 |
-
(self_attn): Qwen2Attention(
|
1820 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1821 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1822 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1823 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1824 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1825 |
-
)
|
1826 |
-
(mlp): Qwen2MLP(
|
1827 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1828 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1829 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1830 |
-
(act_fn): SiLU()
|
1831 |
-
)
|
1832 |
-
(input_layernorm): Qwen2RMSNorm()
|
1833 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1834 |
-
)
|
1835 |
-
(16): Qwen2DecoderLayer(
|
1836 |
-
(self_attn): Qwen2Attention(
|
1837 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1838 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1839 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1840 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1841 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1842 |
-
)
|
1843 |
-
(mlp): Qwen2MLP(
|
1844 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1845 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1846 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1847 |
-
(act_fn): SiLU()
|
1848 |
-
)
|
1849 |
-
(input_layernorm): Qwen2RMSNorm()
|
1850 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1851 |
-
)
|
1852 |
-
(17): Qwen2DecoderLayer(
|
1853 |
-
(self_attn): Qwen2Attention(
|
1854 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1855 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1856 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1857 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1858 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1859 |
-
)
|
1860 |
-
(mlp): Qwen2MLP(
|
1861 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1862 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1863 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1864 |
-
(act_fn): SiLU()
|
1865 |
-
)
|
1866 |
-
(input_layernorm): Qwen2RMSNorm()
|
1867 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1868 |
-
)
|
1869 |
-
(18): Qwen2DecoderLayer(
|
1870 |
-
(self_attn): Qwen2Attention(
|
1871 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1872 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1873 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1874 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1875 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1876 |
-
)
|
1877 |
-
(mlp): Qwen2MLP(
|
1878 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1879 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1880 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1881 |
-
(act_fn): SiLU()
|
1882 |
-
)
|
1883 |
-
(input_layernorm): Qwen2RMSNorm()
|
1884 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1885 |
-
)
|
1886 |
-
(19): Qwen2DecoderLayer(
|
1887 |
-
(self_attn): Qwen2Attention(
|
1888 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1889 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1890 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1891 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1892 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1893 |
-
)
|
1894 |
-
(mlp): Qwen2MLP(
|
1895 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1896 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1897 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1898 |
-
(act_fn): SiLU()
|
1899 |
-
)
|
1900 |
-
(input_layernorm): Qwen2RMSNorm()
|
1901 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1902 |
-
)
|
1903 |
-
(20): Qwen2DecoderLayer(
|
1904 |
-
(self_attn): Qwen2Attention(
|
1905 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1906 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1907 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1908 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1909 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1910 |
-
)
|
1911 |
-
(mlp): Qwen2MLP(
|
1912 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1913 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1914 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1915 |
-
(act_fn): SiLU()
|
1916 |
-
)
|
1917 |
-
(input_layernorm): Qwen2RMSNorm()
|
1918 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1919 |
-
)
|
1920 |
-
(21): Qwen2DecoderLayer(
|
1921 |
-
(self_attn): Qwen2Attention(
|
1922 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1923 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1924 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1925 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1926 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1927 |
-
)
|
1928 |
-
(mlp): Qwen2MLP(
|
1929 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1930 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1931 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1932 |
-
(act_fn): SiLU()
|
1933 |
-
)
|
1934 |
-
(input_layernorm): Qwen2RMSNorm()
|
1935 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1936 |
-
)
|
1937 |
-
(22): Qwen2DecoderLayer(
|
1938 |
-
(self_attn): Qwen2Attention(
|
1939 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1940 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1941 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1942 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1943 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1944 |
-
)
|
1945 |
-
(mlp): Qwen2MLP(
|
1946 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1947 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1948 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1949 |
-
(act_fn): SiLU()
|
1950 |
-
)
|
1951 |
-
(input_layernorm): Qwen2RMSNorm()
|
1952 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1953 |
-
)
|
1954 |
-
(23): Qwen2DecoderLayer(
|
1955 |
-
(self_attn): Qwen2Attention(
|
1956 |
-
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
1957 |
-
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
1958 |
-
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
1959 |
-
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
1960 |
-
(rotary_emb): Qwen2RotaryEmbedding()
|
1961 |
-
)
|
1962 |
-
(mlp): Qwen2MLP(
|
1963 |
-
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1964 |
-
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
1965 |
-
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
1966 |
-
(act_fn): SiLU()
|
1967 |
-
)
|
1968 |
-
(input_layernorm): Qwen2RMSNorm()
|
1969 |
-
(post_attention_layernorm): Qwen2RMSNorm()
|
1970 |
-
)
|
1971 |
-
)
|
1972 |
-
(norm): Qwen2RMSNorm()
|
1973 |
-
)
|
1974 |
-
(lm_head): Linear(in_features=896, out_features=151936, bias=False)
|
1975 |
-
)
|
1976 |
-
(input_projetor): Linear(in_features=1024, out_features=896, bias=True)
|
1977 |
-
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|