Isaak Carter Augustus
commited on
Commit
•
f2da02c
1
Parent(s):
22daade
Upload 3 files
Browse files- first_working_creation_with_custom_encoders.pth +3 -0
- josie_architecture.txt +1002 -0
- josie_dict.txt +693 -0
first_working_creation_with_custom_encoders.pth
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:9a018435767944a31ac38c17dfadd98f681af86227195db88c1763dcd3786ee9
|
3 |
+
size 2468592399
|
josie_architecture.txt
ADDED
@@ -0,0 +1,1002 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
JOSIE(
|
2 |
+
(encoder): Encoder(
|
3 |
+
(modality_preprocessors): ModuleDict(
|
4 |
+
(vision): RGBDTPreprocessor(
|
5 |
+
(cls_token): tensor((1, 1, 768), requires_grad=False)
|
6 |
+
|
7 |
+
(rgbt_stem): PatchEmbedGeneric(
|
8 |
+
(proj): Sequential(
|
9 |
+
(0): PadIm2Video()
|
10 |
+
(1): Conv3d(3, 768, kernel_size=(2, 14, 14), stride=(2, 14, 14), bias=False)
|
11 |
+
)
|
12 |
+
)
|
13 |
+
(pos_embedding_helper): SpatioTemporalPosEmbeddingHelper(
|
14 |
+
(pos_embed): tensor((1, 7681, 768), requires_grad=False)
|
15 |
+
|
16 |
+
)
|
17 |
+
)
|
18 |
+
(audio): AudioPreprocessor(
|
19 |
+
(cls_token): tensor((1, 1, 768), requires_grad=False)
|
20 |
+
|
21 |
+
(rgbt_stem): PatchEmbedGeneric(
|
22 |
+
(proj): Conv2d(1, 768, kernel_size=(16, 16), stride=(10, 10), bias=False)
|
23 |
+
(norm_layer): RMSNorm()
|
24 |
+
)
|
25 |
+
(pos_embedding_helper): SpatioTemporalPosEmbeddingHelper(
|
26 |
+
(pos_embed): tensor((1, 229, 768), requires_grad=False)
|
27 |
+
|
28 |
+
)
|
29 |
+
)
|
30 |
+
(depth): RGBDTPreprocessor(
|
31 |
+
(cls_token): tensor((1, 1, 384), requires_grad=False)
|
32 |
+
|
33 |
+
(depth_stem): PatchEmbedGeneric(
|
34 |
+
(proj): Conv2d(1, 384, kernel_size=(16, 16), stride=(16, 16), bias=False)
|
35 |
+
(norm_layer): RMSNorm()
|
36 |
+
)
|
37 |
+
(pos_embedding_helper): SpatioTemporalPosEmbeddingHelper(
|
38 |
+
(pos_embed): tensor((1, 197, 384), requires_grad=False)
|
39 |
+
|
40 |
+
)
|
41 |
+
)
|
42 |
+
(thermal): ThermalPreprocessor(
|
43 |
+
(cls_token): tensor((1, 1, 768), requires_grad=False)
|
44 |
+
|
45 |
+
(rgbt_stem): PatchEmbedGeneric(
|
46 |
+
(proj): Conv2d(1, 768, kernel_size=(16, 16), stride=(16, 16), bias=False)
|
47 |
+
(norm_layer): RMSNorm()
|
48 |
+
)
|
49 |
+
(pos_embedding_helper): SpatioTemporalPosEmbeddingHelper(
|
50 |
+
(pos_embed): tensor((1, 197, 768), requires_grad=False)
|
51 |
+
|
52 |
+
)
|
53 |
+
)
|
54 |
+
)
|
55 |
+
(modality_transformers): ModuleDict(
|
56 |
+
(vision): EncoderTransformer(
|
57 |
+
(pre_transformer_layer): Sequential(
|
58 |
+
(0): RMSNorm()
|
59 |
+
(1): EinOpsRearrange()
|
60 |
+
)
|
61 |
+
(post_transformer_layer): EinOpsRearrange()
|
62 |
+
(blocks): ModuleList(
|
63 |
+
(0): EncoderTransformerBlock(
|
64 |
+
(attn): MultiheadAttention(
|
65 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
66 |
+
)
|
67 |
+
(drop_path): Identity()
|
68 |
+
(norm1): RMSNorm()
|
69 |
+
(norm2): RMSNorm()
|
70 |
+
(mlp): MLP(
|
71 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
72 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
73 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
74 |
+
)
|
75 |
+
)
|
76 |
+
(1): EncoderTransformerBlock(
|
77 |
+
(attn): MultiheadAttention(
|
78 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
79 |
+
)
|
80 |
+
(drop_path): Identity()
|
81 |
+
(norm1): RMSNorm()
|
82 |
+
(norm2): RMSNorm()
|
83 |
+
(mlp): MLP(
|
84 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
85 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
86 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
87 |
+
)
|
88 |
+
)
|
89 |
+
(2): EncoderTransformerBlock(
|
90 |
+
(attn): MultiheadAttention(
|
91 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
92 |
+
)
|
93 |
+
(drop_path): Identity()
|
94 |
+
(norm1): RMSNorm()
|
95 |
+
(norm2): RMSNorm()
|
96 |
+
(mlp): MLP(
|
97 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
98 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
99 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
100 |
+
)
|
101 |
+
)
|
102 |
+
(3): EncoderTransformerBlock(
|
103 |
+
(attn): MultiheadAttention(
|
104 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
105 |
+
)
|
106 |
+
(drop_path): Identity()
|
107 |
+
(norm1): RMSNorm()
|
108 |
+
(norm2): RMSNorm()
|
109 |
+
(mlp): MLP(
|
110 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
111 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
112 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
113 |
+
)
|
114 |
+
)
|
115 |
+
(4): EncoderTransformerBlock(
|
116 |
+
(attn): MultiheadAttention(
|
117 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
118 |
+
)
|
119 |
+
(drop_path): Identity()
|
120 |
+
(norm1): RMSNorm()
|
121 |
+
(norm2): RMSNorm()
|
122 |
+
(mlp): MLP(
|
123 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
124 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
125 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
126 |
+
)
|
127 |
+
)
|
128 |
+
(5): EncoderTransformerBlock(
|
129 |
+
(attn): MultiheadAttention(
|
130 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
131 |
+
)
|
132 |
+
(drop_path): Identity()
|
133 |
+
(norm1): RMSNorm()
|
134 |
+
(norm2): RMSNorm()
|
135 |
+
(mlp): MLP(
|
136 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
137 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
138 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
139 |
+
)
|
140 |
+
)
|
141 |
+
(6): EncoderTransformerBlock(
|
142 |
+
(attn): MultiheadAttention(
|
143 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
144 |
+
)
|
145 |
+
(drop_path): Identity()
|
146 |
+
(norm1): RMSNorm()
|
147 |
+
(norm2): RMSNorm()
|
148 |
+
(mlp): MLP(
|
149 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
150 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
151 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
152 |
+
)
|
153 |
+
)
|
154 |
+
(7): EncoderTransformerBlock(
|
155 |
+
(attn): MultiheadAttention(
|
156 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
157 |
+
)
|
158 |
+
(drop_path): Identity()
|
159 |
+
(norm1): RMSNorm()
|
160 |
+
(norm2): RMSNorm()
|
161 |
+
(mlp): MLP(
|
162 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
163 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
164 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
165 |
+
)
|
166 |
+
)
|
167 |
+
(8): EncoderTransformerBlock(
|
168 |
+
(attn): MultiheadAttention(
|
169 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
170 |
+
)
|
171 |
+
(drop_path): Identity()
|
172 |
+
(norm1): RMSNorm()
|
173 |
+
(norm2): RMSNorm()
|
174 |
+
(mlp): MLP(
|
175 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
176 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
177 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
178 |
+
)
|
179 |
+
)
|
180 |
+
(9): EncoderTransformerBlock(
|
181 |
+
(attn): MultiheadAttention(
|
182 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
183 |
+
)
|
184 |
+
(drop_path): Identity()
|
185 |
+
(norm1): RMSNorm()
|
186 |
+
(norm2): RMSNorm()
|
187 |
+
(mlp): MLP(
|
188 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
189 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
190 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
191 |
+
)
|
192 |
+
)
|
193 |
+
(10): EncoderTransformerBlock(
|
194 |
+
(attn): MultiheadAttention(
|
195 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
196 |
+
)
|
197 |
+
(drop_path): Identity()
|
198 |
+
(norm1): RMSNorm()
|
199 |
+
(norm2): RMSNorm()
|
200 |
+
(mlp): MLP(
|
201 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
202 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
203 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
204 |
+
)
|
205 |
+
)
|
206 |
+
(11): EncoderTransformerBlock(
|
207 |
+
(attn): MultiheadAttention(
|
208 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
209 |
+
)
|
210 |
+
(drop_path): Identity()
|
211 |
+
(norm1): RMSNorm()
|
212 |
+
(norm2): RMSNorm()
|
213 |
+
(mlp): MLP(
|
214 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
215 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
216 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
217 |
+
)
|
218 |
+
)
|
219 |
+
)
|
220 |
+
)
|
221 |
+
(audio): EncoderTransformer(
|
222 |
+
(pre_transformer_layer): Sequential(
|
223 |
+
(0): RMSNorm()
|
224 |
+
(1): EinOpsRearrange()
|
225 |
+
)
|
226 |
+
(post_transformer_layer): EinOpsRearrange()
|
227 |
+
(blocks): ModuleList(
|
228 |
+
(0): EncoderTransformerBlock(
|
229 |
+
(attn): MultiheadAttention(
|
230 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
231 |
+
)
|
232 |
+
(drop_path): Identity()
|
233 |
+
(norm1): RMSNorm()
|
234 |
+
(norm2): RMSNorm()
|
235 |
+
(mlp): MLP(
|
236 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
237 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
238 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
239 |
+
)
|
240 |
+
)
|
241 |
+
(1): EncoderTransformerBlock(
|
242 |
+
(attn): MultiheadAttention(
|
243 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
244 |
+
)
|
245 |
+
(drop_path): DropPath(drop_prob=0.009)
|
246 |
+
(norm1): RMSNorm()
|
247 |
+
(norm2): RMSNorm()
|
248 |
+
(mlp): MLP(
|
249 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
250 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
251 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
252 |
+
)
|
253 |
+
)
|
254 |
+
(2): EncoderTransformerBlock(
|
255 |
+
(attn): MultiheadAttention(
|
256 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
257 |
+
)
|
258 |
+
(drop_path): DropPath(drop_prob=0.018)
|
259 |
+
(norm1): RMSNorm()
|
260 |
+
(norm2): RMSNorm()
|
261 |
+
(mlp): MLP(
|
262 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
263 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
264 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
265 |
+
)
|
266 |
+
)
|
267 |
+
(3): EncoderTransformerBlock(
|
268 |
+
(attn): MultiheadAttention(
|
269 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
270 |
+
)
|
271 |
+
(drop_path): DropPath(drop_prob=0.027)
|
272 |
+
(norm1): RMSNorm()
|
273 |
+
(norm2): RMSNorm()
|
274 |
+
(mlp): MLP(
|
275 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
276 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
277 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
278 |
+
)
|
279 |
+
)
|
280 |
+
(4): EncoderTransformerBlock(
|
281 |
+
(attn): MultiheadAttention(
|
282 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
283 |
+
)
|
284 |
+
(drop_path): DropPath(drop_prob=0.036)
|
285 |
+
(norm1): RMSNorm()
|
286 |
+
(norm2): RMSNorm()
|
287 |
+
(mlp): MLP(
|
288 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
289 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
290 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
291 |
+
)
|
292 |
+
)
|
293 |
+
(5): EncoderTransformerBlock(
|
294 |
+
(attn): MultiheadAttention(
|
295 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
296 |
+
)
|
297 |
+
(drop_path): DropPath(drop_prob=0.045)
|
298 |
+
(norm1): RMSNorm()
|
299 |
+
(norm2): RMSNorm()
|
300 |
+
(mlp): MLP(
|
301 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
302 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
303 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
304 |
+
)
|
305 |
+
)
|
306 |
+
(6): EncoderTransformerBlock(
|
307 |
+
(attn): MultiheadAttention(
|
308 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
309 |
+
)
|
310 |
+
(drop_path): DropPath(drop_prob=0.055)
|
311 |
+
(norm1): RMSNorm()
|
312 |
+
(norm2): RMSNorm()
|
313 |
+
(mlp): MLP(
|
314 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
315 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
316 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
317 |
+
)
|
318 |
+
)
|
319 |
+
(7): EncoderTransformerBlock(
|
320 |
+
(attn): MultiheadAttention(
|
321 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
322 |
+
)
|
323 |
+
(drop_path): DropPath(drop_prob=0.064)
|
324 |
+
(norm1): RMSNorm()
|
325 |
+
(norm2): RMSNorm()
|
326 |
+
(mlp): MLP(
|
327 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
328 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
329 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
330 |
+
)
|
331 |
+
)
|
332 |
+
(8): EncoderTransformerBlock(
|
333 |
+
(attn): MultiheadAttention(
|
334 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
335 |
+
)
|
336 |
+
(drop_path): DropPath(drop_prob=0.073)
|
337 |
+
(norm1): RMSNorm()
|
338 |
+
(norm2): RMSNorm()
|
339 |
+
(mlp): MLP(
|
340 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
341 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
342 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
343 |
+
)
|
344 |
+
)
|
345 |
+
(9): EncoderTransformerBlock(
|
346 |
+
(attn): MultiheadAttention(
|
347 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
348 |
+
)
|
349 |
+
(drop_path): DropPath(drop_prob=0.082)
|
350 |
+
(norm1): RMSNorm()
|
351 |
+
(norm2): RMSNorm()
|
352 |
+
(mlp): MLP(
|
353 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
354 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
355 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
356 |
+
)
|
357 |
+
)
|
358 |
+
(10): EncoderTransformerBlock(
|
359 |
+
(attn): MultiheadAttention(
|
360 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
361 |
+
)
|
362 |
+
(drop_path): DropPath(drop_prob=0.091)
|
363 |
+
(norm1): RMSNorm()
|
364 |
+
(norm2): RMSNorm()
|
365 |
+
(mlp): MLP(
|
366 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
367 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
368 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
369 |
+
)
|
370 |
+
)
|
371 |
+
(11): EncoderTransformerBlock(
|
372 |
+
(attn): MultiheadAttention(
|
373 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
374 |
+
)
|
375 |
+
(drop_path): DropPath(drop_prob=0.100)
|
376 |
+
(norm1): RMSNorm()
|
377 |
+
(norm2): RMSNorm()
|
378 |
+
(mlp): MLP(
|
379 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
380 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
381 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
382 |
+
)
|
383 |
+
)
|
384 |
+
)
|
385 |
+
)
|
386 |
+
(depth): EncoderTransformer(
|
387 |
+
(pre_transformer_layer): Sequential(
|
388 |
+
(0): RMSNorm()
|
389 |
+
(1): EinOpsRearrange()
|
390 |
+
)
|
391 |
+
(post_transformer_layer): EinOpsRearrange()
|
392 |
+
(blocks): ModuleList(
|
393 |
+
(0): EncoderTransformerBlock(
|
394 |
+
(attn): MultiheadAttention(
|
395 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
396 |
+
)
|
397 |
+
(drop_path): Identity()
|
398 |
+
(norm1): RMSNorm()
|
399 |
+
(norm2): RMSNorm()
|
400 |
+
(mlp): MLP(
|
401 |
+
(w1): Linear(in_features=384, out_features=256, bias=False)
|
402 |
+
(w2): Linear(in_features=256, out_features=384, bias=False)
|
403 |
+
(w3): Linear(in_features=384, out_features=256, bias=False)
|
404 |
+
)
|
405 |
+
)
|
406 |
+
(1): EncoderTransformerBlock(
|
407 |
+
(attn): MultiheadAttention(
|
408 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
409 |
+
)
|
410 |
+
(drop_path): Identity()
|
411 |
+
(norm1): RMSNorm()
|
412 |
+
(norm2): RMSNorm()
|
413 |
+
(mlp): MLP(
|
414 |
+
(w1): Linear(in_features=384, out_features=256, bias=False)
|
415 |
+
(w2): Linear(in_features=256, out_features=384, bias=False)
|
416 |
+
(w3): Linear(in_features=384, out_features=256, bias=False)
|
417 |
+
)
|
418 |
+
)
|
419 |
+
(2): EncoderTransformerBlock(
|
420 |
+
(attn): MultiheadAttention(
|
421 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
422 |
+
)
|
423 |
+
(drop_path): Identity()
|
424 |
+
(norm1): RMSNorm()
|
425 |
+
(norm2): RMSNorm()
|
426 |
+
(mlp): MLP(
|
427 |
+
(w1): Linear(in_features=384, out_features=256, bias=False)
|
428 |
+
(w2): Linear(in_features=256, out_features=384, bias=False)
|
429 |
+
(w3): Linear(in_features=384, out_features=256, bias=False)
|
430 |
+
)
|
431 |
+
)
|
432 |
+
(3): EncoderTransformerBlock(
|
433 |
+
(attn): MultiheadAttention(
|
434 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
435 |
+
)
|
436 |
+
(drop_path): Identity()
|
437 |
+
(norm1): RMSNorm()
|
438 |
+
(norm2): RMSNorm()
|
439 |
+
(mlp): MLP(
|
440 |
+
(w1): Linear(in_features=384, out_features=256, bias=False)
|
441 |
+
(w2): Linear(in_features=256, out_features=384, bias=False)
|
442 |
+
(w3): Linear(in_features=384, out_features=256, bias=False)
|
443 |
+
)
|
444 |
+
)
|
445 |
+
(4): EncoderTransformerBlock(
|
446 |
+
(attn): MultiheadAttention(
|
447 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
448 |
+
)
|
449 |
+
(drop_path): Identity()
|
450 |
+
(norm1): RMSNorm()
|
451 |
+
(norm2): RMSNorm()
|
452 |
+
(mlp): MLP(
|
453 |
+
(w1): Linear(in_features=384, out_features=256, bias=False)
|
454 |
+
(w2): Linear(in_features=256, out_features=384, bias=False)
|
455 |
+
(w3): Linear(in_features=384, out_features=256, bias=False)
|
456 |
+
)
|
457 |
+
)
|
458 |
+
(5): EncoderTransformerBlock(
|
459 |
+
(attn): MultiheadAttention(
|
460 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=384, out_features=384, bias=True)
|
461 |
+
)
|
462 |
+
(drop_path): Identity()
|
463 |
+
(norm1): RMSNorm()
|
464 |
+
(norm2): RMSNorm()
|
465 |
+
(mlp): MLP(
|
466 |
+
(w1): Linear(in_features=384, out_features=256, bias=False)
|
467 |
+
(w2): Linear(in_features=256, out_features=384, bias=False)
|
468 |
+
(w3): Linear(in_features=384, out_features=256, bias=False)
|
469 |
+
)
|
470 |
+
)
|
471 |
+
)
|
472 |
+
)
|
473 |
+
(thermal): EncoderTransformer(
|
474 |
+
(pre_transformer_layer): Sequential(
|
475 |
+
(0): RMSNorm()
|
476 |
+
(1): EinOpsRearrange()
|
477 |
+
)
|
478 |
+
(post_transformer_layer): EinOpsRearrange()
|
479 |
+
(blocks): ModuleList(
|
480 |
+
(0): EncoderTransformerBlock(
|
481 |
+
(attn): MultiheadAttention(
|
482 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
483 |
+
)
|
484 |
+
(drop_path): Identity()
|
485 |
+
(norm1): RMSNorm()
|
486 |
+
(norm2): RMSNorm()
|
487 |
+
(mlp): MLP(
|
488 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
489 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
490 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
491 |
+
)
|
492 |
+
)
|
493 |
+
(1): EncoderTransformerBlock(
|
494 |
+
(attn): MultiheadAttention(
|
495 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
496 |
+
)
|
497 |
+
(drop_path): Identity()
|
498 |
+
(norm1): RMSNorm()
|
499 |
+
(norm2): RMSNorm()
|
500 |
+
(mlp): MLP(
|
501 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
502 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
503 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
504 |
+
)
|
505 |
+
)
|
506 |
+
(2): EncoderTransformerBlock(
|
507 |
+
(attn): MultiheadAttention(
|
508 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
509 |
+
)
|
510 |
+
(drop_path): Identity()
|
511 |
+
(norm1): RMSNorm()
|
512 |
+
(norm2): RMSNorm()
|
513 |
+
(mlp): MLP(
|
514 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
515 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
516 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
517 |
+
)
|
518 |
+
)
|
519 |
+
(3): EncoderTransformerBlock(
|
520 |
+
(attn): MultiheadAttention(
|
521 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
522 |
+
)
|
523 |
+
(drop_path): Identity()
|
524 |
+
(norm1): RMSNorm()
|
525 |
+
(norm2): RMSNorm()
|
526 |
+
(mlp): MLP(
|
527 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
528 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
529 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
530 |
+
)
|
531 |
+
)
|
532 |
+
(4): EncoderTransformerBlock(
|
533 |
+
(attn): MultiheadAttention(
|
534 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
535 |
+
)
|
536 |
+
(drop_path): Identity()
|
537 |
+
(norm1): RMSNorm()
|
538 |
+
(norm2): RMSNorm()
|
539 |
+
(mlp): MLP(
|
540 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
541 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
542 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
543 |
+
)
|
544 |
+
)
|
545 |
+
(5): EncoderTransformerBlock(
|
546 |
+
(attn): MultiheadAttention(
|
547 |
+
(out_proj): NonDynamicallyQuantizableLinear(in_features=768, out_features=768, bias=True)
|
548 |
+
)
|
549 |
+
(drop_path): Identity()
|
550 |
+
(norm1): RMSNorm()
|
551 |
+
(norm2): RMSNorm()
|
552 |
+
(mlp): MLP(
|
553 |
+
(w1): Linear(in_features=768, out_features=512, bias=False)
|
554 |
+
(w2): Linear(in_features=512, out_features=768, bias=False)
|
555 |
+
(w3): Linear(in_features=768, out_features=512, bias=False)
|
556 |
+
)
|
557 |
+
)
|
558 |
+
)
|
559 |
+
)
|
560 |
+
)
|
561 |
+
(modality_heads): ModuleDict(
|
562 |
+
(vision): Sequential(
|
563 |
+
(0): RMSNorm()
|
564 |
+
(1): SelectElement()
|
565 |
+
(2): Linear(in_features=768, out_features=1024, bias=False)
|
566 |
+
)
|
567 |
+
(audio): Sequential(
|
568 |
+
(0): RMSNorm()
|
569 |
+
(1): SelectElement()
|
570 |
+
(2): Linear(in_features=768, out_features=1024, bias=False)
|
571 |
+
)
|
572 |
+
(depth): Sequential(
|
573 |
+
(0): RMSNorm()
|
574 |
+
(1): SelectElement()
|
575 |
+
(2): Linear(in_features=384, out_features=1024, bias=False)
|
576 |
+
)
|
577 |
+
(thermal): Sequential(
|
578 |
+
(0): RMSNorm()
|
579 |
+
(1): SelectElement()
|
580 |
+
(2): Linear(in_features=768, out_features=1024, bias=False)
|
581 |
+
)
|
582 |
+
)
|
583 |
+
)
|
584 |
+
(reasoner): Qwen2ForCausalLM(
|
585 |
+
(model): Qwen2Model(
|
586 |
+
(embed_tokens): Embedding(151936, 896)
|
587 |
+
(layers): ModuleList(
|
588 |
+
(0): Qwen2DecoderLayer(
|
589 |
+
(self_attn): Qwen2Attention(
|
590 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
591 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
592 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
593 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
594 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
595 |
+
)
|
596 |
+
(mlp): Qwen2MLP(
|
597 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
598 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
599 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
600 |
+
(act_fn): SiLU()
|
601 |
+
)
|
602 |
+
(input_layernorm): Qwen2RMSNorm()
|
603 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
604 |
+
)
|
605 |
+
(1): Qwen2DecoderLayer(
|
606 |
+
(self_attn): Qwen2Attention(
|
607 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
608 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
609 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
610 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
611 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
612 |
+
)
|
613 |
+
(mlp): Qwen2MLP(
|
614 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
615 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
616 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
617 |
+
(act_fn): SiLU()
|
618 |
+
)
|
619 |
+
(input_layernorm): Qwen2RMSNorm()
|
620 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
621 |
+
)
|
622 |
+
(2): Qwen2DecoderLayer(
|
623 |
+
(self_attn): Qwen2Attention(
|
624 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
625 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
626 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
627 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
628 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
629 |
+
)
|
630 |
+
(mlp): Qwen2MLP(
|
631 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
632 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
633 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
634 |
+
(act_fn): SiLU()
|
635 |
+
)
|
636 |
+
(input_layernorm): Qwen2RMSNorm()
|
637 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
638 |
+
)
|
639 |
+
(3): Qwen2DecoderLayer(
|
640 |
+
(self_attn): Qwen2Attention(
|
641 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
642 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
643 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
644 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
645 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
646 |
+
)
|
647 |
+
(mlp): Qwen2MLP(
|
648 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
649 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
650 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
651 |
+
(act_fn): SiLU()
|
652 |
+
)
|
653 |
+
(input_layernorm): Qwen2RMSNorm()
|
654 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
655 |
+
)
|
656 |
+
(4): Qwen2DecoderLayer(
|
657 |
+
(self_attn): Qwen2Attention(
|
658 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
659 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
660 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
661 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
662 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
663 |
+
)
|
664 |
+
(mlp): Qwen2MLP(
|
665 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
666 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
667 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
668 |
+
(act_fn): SiLU()
|
669 |
+
)
|
670 |
+
(input_layernorm): Qwen2RMSNorm()
|
671 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
672 |
+
)
|
673 |
+
(5): Qwen2DecoderLayer(
|
674 |
+
(self_attn): Qwen2Attention(
|
675 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
676 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
677 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
678 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
679 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
680 |
+
)
|
681 |
+
(mlp): Qwen2MLP(
|
682 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
683 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
684 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
685 |
+
(act_fn): SiLU()
|
686 |
+
)
|
687 |
+
(input_layernorm): Qwen2RMSNorm()
|
688 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
689 |
+
)
|
690 |
+
(6): Qwen2DecoderLayer(
|
691 |
+
(self_attn): Qwen2Attention(
|
692 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
693 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
694 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
695 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
696 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
697 |
+
)
|
698 |
+
(mlp): Qwen2MLP(
|
699 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
700 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
701 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
702 |
+
(act_fn): SiLU()
|
703 |
+
)
|
704 |
+
(input_layernorm): Qwen2RMSNorm()
|
705 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
706 |
+
)
|
707 |
+
(7): Qwen2DecoderLayer(
|
708 |
+
(self_attn): Qwen2Attention(
|
709 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
710 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
711 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
712 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
713 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
714 |
+
)
|
715 |
+
(mlp): Qwen2MLP(
|
716 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
717 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
718 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
719 |
+
(act_fn): SiLU()
|
720 |
+
)
|
721 |
+
(input_layernorm): Qwen2RMSNorm()
|
722 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
723 |
+
)
|
724 |
+
(8): Qwen2DecoderLayer(
|
725 |
+
(self_attn): Qwen2Attention(
|
726 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
727 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
728 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
729 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
730 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
731 |
+
)
|
732 |
+
(mlp): Qwen2MLP(
|
733 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
734 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
735 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
736 |
+
(act_fn): SiLU()
|
737 |
+
)
|
738 |
+
(input_layernorm): Qwen2RMSNorm()
|
739 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
740 |
+
)
|
741 |
+
(9): Qwen2DecoderLayer(
|
742 |
+
(self_attn): Qwen2Attention(
|
743 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
744 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
745 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
746 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
747 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
748 |
+
)
|
749 |
+
(mlp): Qwen2MLP(
|
750 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
751 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
752 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
753 |
+
(act_fn): SiLU()
|
754 |
+
)
|
755 |
+
(input_layernorm): Qwen2RMSNorm()
|
756 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
757 |
+
)
|
758 |
+
(10): Qwen2DecoderLayer(
|
759 |
+
(self_attn): Qwen2Attention(
|
760 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
761 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
762 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
763 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
764 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
765 |
+
)
|
766 |
+
(mlp): Qwen2MLP(
|
767 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
768 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
769 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
770 |
+
(act_fn): SiLU()
|
771 |
+
)
|
772 |
+
(input_layernorm): Qwen2RMSNorm()
|
773 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
774 |
+
)
|
775 |
+
(11): Qwen2DecoderLayer(
|
776 |
+
(self_attn): Qwen2Attention(
|
777 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
778 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
779 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
780 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
781 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
782 |
+
)
|
783 |
+
(mlp): Qwen2MLP(
|
784 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
785 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
786 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
787 |
+
(act_fn): SiLU()
|
788 |
+
)
|
789 |
+
(input_layernorm): Qwen2RMSNorm()
|
790 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
791 |
+
)
|
792 |
+
(12): Qwen2DecoderLayer(
|
793 |
+
(self_attn): Qwen2Attention(
|
794 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
795 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
796 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
797 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
798 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
799 |
+
)
|
800 |
+
(mlp): Qwen2MLP(
|
801 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
802 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
803 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
804 |
+
(act_fn): SiLU()
|
805 |
+
)
|
806 |
+
(input_layernorm): Qwen2RMSNorm()
|
807 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
808 |
+
)
|
809 |
+
(13): Qwen2DecoderLayer(
|
810 |
+
(self_attn): Qwen2Attention(
|
811 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
812 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
813 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
814 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
815 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
816 |
+
)
|
817 |
+
(mlp): Qwen2MLP(
|
818 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
819 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
820 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
821 |
+
(act_fn): SiLU()
|
822 |
+
)
|
823 |
+
(input_layernorm): Qwen2RMSNorm()
|
824 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
825 |
+
)
|
826 |
+
(14): Qwen2DecoderLayer(
|
827 |
+
(self_attn): Qwen2Attention(
|
828 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
829 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
830 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
831 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
832 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
833 |
+
)
|
834 |
+
(mlp): Qwen2MLP(
|
835 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
836 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
837 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
838 |
+
(act_fn): SiLU()
|
839 |
+
)
|
840 |
+
(input_layernorm): Qwen2RMSNorm()
|
841 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
842 |
+
)
|
843 |
+
(15): Qwen2DecoderLayer(
|
844 |
+
(self_attn): Qwen2Attention(
|
845 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
846 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
847 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
848 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
849 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
850 |
+
)
|
851 |
+
(mlp): Qwen2MLP(
|
852 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
853 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
854 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
855 |
+
(act_fn): SiLU()
|
856 |
+
)
|
857 |
+
(input_layernorm): Qwen2RMSNorm()
|
858 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
859 |
+
)
|
860 |
+
(16): Qwen2DecoderLayer(
|
861 |
+
(self_attn): Qwen2Attention(
|
862 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
863 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
864 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
865 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
866 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
867 |
+
)
|
868 |
+
(mlp): Qwen2MLP(
|
869 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
870 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
871 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
872 |
+
(act_fn): SiLU()
|
873 |
+
)
|
874 |
+
(input_layernorm): Qwen2RMSNorm()
|
875 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
876 |
+
)
|
877 |
+
(17): Qwen2DecoderLayer(
|
878 |
+
(self_attn): Qwen2Attention(
|
879 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
880 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
881 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
882 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
883 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
884 |
+
)
|
885 |
+
(mlp): Qwen2MLP(
|
886 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
887 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
888 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
889 |
+
(act_fn): SiLU()
|
890 |
+
)
|
891 |
+
(input_layernorm): Qwen2RMSNorm()
|
892 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
893 |
+
)
|
894 |
+
(18): Qwen2DecoderLayer(
|
895 |
+
(self_attn): Qwen2Attention(
|
896 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
897 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
898 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
899 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
900 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
901 |
+
)
|
902 |
+
(mlp): Qwen2MLP(
|
903 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
904 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
905 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
906 |
+
(act_fn): SiLU()
|
907 |
+
)
|
908 |
+
(input_layernorm): Qwen2RMSNorm()
|
909 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
910 |
+
)
|
911 |
+
(19): Qwen2DecoderLayer(
|
912 |
+
(self_attn): Qwen2Attention(
|
913 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
914 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
915 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
916 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
917 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
918 |
+
)
|
919 |
+
(mlp): Qwen2MLP(
|
920 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
921 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
922 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
923 |
+
(act_fn): SiLU()
|
924 |
+
)
|
925 |
+
(input_layernorm): Qwen2RMSNorm()
|
926 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
927 |
+
)
|
928 |
+
(20): Qwen2DecoderLayer(
|
929 |
+
(self_attn): Qwen2Attention(
|
930 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
931 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
932 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
933 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
934 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
935 |
+
)
|
936 |
+
(mlp): Qwen2MLP(
|
937 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
938 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
939 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
940 |
+
(act_fn): SiLU()
|
941 |
+
)
|
942 |
+
(input_layernorm): Qwen2RMSNorm()
|
943 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
944 |
+
)
|
945 |
+
(21): Qwen2DecoderLayer(
|
946 |
+
(self_attn): Qwen2Attention(
|
947 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
948 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
949 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
950 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
951 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
952 |
+
)
|
953 |
+
(mlp): Qwen2MLP(
|
954 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
955 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
956 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
957 |
+
(act_fn): SiLU()
|
958 |
+
)
|
959 |
+
(input_layernorm): Qwen2RMSNorm()
|
960 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
961 |
+
)
|
962 |
+
(22): Qwen2DecoderLayer(
|
963 |
+
(self_attn): Qwen2Attention(
|
964 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
965 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
966 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
967 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
968 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
969 |
+
)
|
970 |
+
(mlp): Qwen2MLP(
|
971 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
972 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
973 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
974 |
+
(act_fn): SiLU()
|
975 |
+
)
|
976 |
+
(input_layernorm): Qwen2RMSNorm()
|
977 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
978 |
+
)
|
979 |
+
(23): Qwen2DecoderLayer(
|
980 |
+
(self_attn): Qwen2Attention(
|
981 |
+
(q_proj): Linear(in_features=896, out_features=896, bias=True)
|
982 |
+
(k_proj): Linear(in_features=896, out_features=128, bias=True)
|
983 |
+
(v_proj): Linear(in_features=896, out_features=128, bias=True)
|
984 |
+
(o_proj): Linear(in_features=896, out_features=896, bias=False)
|
985 |
+
(rotary_emb): Qwen2RotaryEmbedding()
|
986 |
+
)
|
987 |
+
(mlp): Qwen2MLP(
|
988 |
+
(gate_proj): Linear(in_features=896, out_features=4864, bias=False)
|
989 |
+
(up_proj): Linear(in_features=896, out_features=4864, bias=False)
|
990 |
+
(down_proj): Linear(in_features=4864, out_features=896, bias=False)
|
991 |
+
(act_fn): SiLU()
|
992 |
+
)
|
993 |
+
(input_layernorm): Qwen2RMSNorm()
|
994 |
+
(post_attention_layernorm): Qwen2RMSNorm()
|
995 |
+
)
|
996 |
+
)
|
997 |
+
(norm): Qwen2RMSNorm()
|
998 |
+
)
|
999 |
+
(lm_head): Linear(in_features=896, out_features=151936, bias=False)
|
1000 |
+
)
|
1001 |
+
(input_projetor): Linear(in_features=1024, out_features=896, bias=True)
|
1002 |
+
)
|
josie_dict.txt
ADDED
@@ -0,0 +1,693 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
Model's state_dict:
|
2 |
+
encoder.modality_preprocessors.vision.cls_token torch.Size([1, 1, 768])
|
3 |
+
encoder.modality_preprocessors.vision.rgbt_stem.proj.1.weight torch.Size([768, 3, 2, 14, 14])
|
4 |
+
encoder.modality_preprocessors.vision.pos_embedding_helper.pos_embed torch.Size([1, 7681, 768])
|
5 |
+
encoder.modality_preprocessors.audio.cls_token torch.Size([1, 1, 768])
|
6 |
+
encoder.modality_preprocessors.audio.rgbt_stem.proj.weight torch.Size([768, 1, 16, 16])
|
7 |
+
encoder.modality_preprocessors.audio.rgbt_stem.norm_layer.weight torch.Size([768])
|
8 |
+
encoder.modality_preprocessors.audio.pos_embedding_helper.pos_embed torch.Size([1, 229, 768])
|
9 |
+
encoder.modality_preprocessors.depth.cls_token torch.Size([1, 1, 384])
|
10 |
+
encoder.modality_preprocessors.depth.depth_stem.proj.weight torch.Size([384, 1, 16, 16])
|
11 |
+
encoder.modality_preprocessors.depth.depth_stem.norm_layer.weight torch.Size([384])
|
12 |
+
encoder.modality_preprocessors.depth.pos_embedding_helper.pos_embed torch.Size([1, 197, 384])
|
13 |
+
encoder.modality_preprocessors.thermal.cls_token torch.Size([1, 1, 768])
|
14 |
+
encoder.modality_preprocessors.thermal.rgbt_stem.proj.weight torch.Size([768, 1, 16, 16])
|
15 |
+
encoder.modality_preprocessors.thermal.rgbt_stem.norm_layer.weight torch.Size([768])
|
16 |
+
encoder.modality_preprocessors.thermal.pos_embedding_helper.pos_embed torch.Size([1, 197, 768])
|
17 |
+
encoder.modality_transformers.vision.pre_transformer_layer.0.weight torch.Size([768])
|
18 |
+
encoder.modality_transformers.vision.blocks.0.attn.in_proj_weight torch.Size([2304, 768])
|
19 |
+
encoder.modality_transformers.vision.blocks.0.attn.in_proj_bias torch.Size([2304])
|
20 |
+
encoder.modality_transformers.vision.blocks.0.attn.out_proj.weight torch.Size([768, 768])
|
21 |
+
encoder.modality_transformers.vision.blocks.0.attn.out_proj.bias torch.Size([768])
|
22 |
+
encoder.modality_transformers.vision.blocks.0.norm1.weight torch.Size([768])
|
23 |
+
encoder.modality_transformers.vision.blocks.0.norm2.weight torch.Size([768])
|
24 |
+
encoder.modality_transformers.vision.blocks.0.mlp.w1.weight torch.Size([512, 768])
|
25 |
+
encoder.modality_transformers.vision.blocks.0.mlp.w2.weight torch.Size([768, 512])
|
26 |
+
encoder.modality_transformers.vision.blocks.0.mlp.w3.weight torch.Size([512, 768])
|
27 |
+
encoder.modality_transformers.vision.blocks.1.attn.in_proj_weight torch.Size([2304, 768])
|
28 |
+
encoder.modality_transformers.vision.blocks.1.attn.in_proj_bias torch.Size([2304])
|
29 |
+
encoder.modality_transformers.vision.blocks.1.attn.out_proj.weight torch.Size([768, 768])
|
30 |
+
encoder.modality_transformers.vision.blocks.1.attn.out_proj.bias torch.Size([768])
|
31 |
+
encoder.modality_transformers.vision.blocks.1.norm1.weight torch.Size([768])
|
32 |
+
encoder.modality_transformers.vision.blocks.1.norm2.weight torch.Size([768])
|
33 |
+
encoder.modality_transformers.vision.blocks.1.mlp.w1.weight torch.Size([512, 768])
|
34 |
+
encoder.modality_transformers.vision.blocks.1.mlp.w2.weight torch.Size([768, 512])
|
35 |
+
encoder.modality_transformers.vision.blocks.1.mlp.w3.weight torch.Size([512, 768])
|
36 |
+
encoder.modality_transformers.vision.blocks.2.attn.in_proj_weight torch.Size([2304, 768])
|
37 |
+
encoder.modality_transformers.vision.blocks.2.attn.in_proj_bias torch.Size([2304])
|
38 |
+
encoder.modality_transformers.vision.blocks.2.attn.out_proj.weight torch.Size([768, 768])
|
39 |
+
encoder.modality_transformers.vision.blocks.2.attn.out_proj.bias torch.Size([768])
|
40 |
+
encoder.modality_transformers.vision.blocks.2.norm1.weight torch.Size([768])
|
41 |
+
encoder.modality_transformers.vision.blocks.2.norm2.weight torch.Size([768])
|
42 |
+
encoder.modality_transformers.vision.blocks.2.mlp.w1.weight torch.Size([512, 768])
|
43 |
+
encoder.modality_transformers.vision.blocks.2.mlp.w2.weight torch.Size([768, 512])
|
44 |
+
encoder.modality_transformers.vision.blocks.2.mlp.w3.weight torch.Size([512, 768])
|
45 |
+
encoder.modality_transformers.vision.blocks.3.attn.in_proj_weight torch.Size([2304, 768])
|
46 |
+
encoder.modality_transformers.vision.blocks.3.attn.in_proj_bias torch.Size([2304])
|
47 |
+
encoder.modality_transformers.vision.blocks.3.attn.out_proj.weight torch.Size([768, 768])
|
48 |
+
encoder.modality_transformers.vision.blocks.3.attn.out_proj.bias torch.Size([768])
|
49 |
+
encoder.modality_transformers.vision.blocks.3.norm1.weight torch.Size([768])
|
50 |
+
encoder.modality_transformers.vision.blocks.3.norm2.weight torch.Size([768])
|
51 |
+
encoder.modality_transformers.vision.blocks.3.mlp.w1.weight torch.Size([512, 768])
|
52 |
+
encoder.modality_transformers.vision.blocks.3.mlp.w2.weight torch.Size([768, 512])
|
53 |
+
encoder.modality_transformers.vision.blocks.3.mlp.w3.weight torch.Size([512, 768])
|
54 |
+
encoder.modality_transformers.vision.blocks.4.attn.in_proj_weight torch.Size([2304, 768])
|
55 |
+
encoder.modality_transformers.vision.blocks.4.attn.in_proj_bias torch.Size([2304])
|
56 |
+
encoder.modality_transformers.vision.blocks.4.attn.out_proj.weight torch.Size([768, 768])
|
57 |
+
encoder.modality_transformers.vision.blocks.4.attn.out_proj.bias torch.Size([768])
|
58 |
+
encoder.modality_transformers.vision.blocks.4.norm1.weight torch.Size([768])
|
59 |
+
encoder.modality_transformers.vision.blocks.4.norm2.weight torch.Size([768])
|
60 |
+
encoder.modality_transformers.vision.blocks.4.mlp.w1.weight torch.Size([512, 768])
|
61 |
+
encoder.modality_transformers.vision.blocks.4.mlp.w2.weight torch.Size([768, 512])
|
62 |
+
encoder.modality_transformers.vision.blocks.4.mlp.w3.weight torch.Size([512, 768])
|
63 |
+
encoder.modality_transformers.vision.blocks.5.attn.in_proj_weight torch.Size([2304, 768])
|
64 |
+
encoder.modality_transformers.vision.blocks.5.attn.in_proj_bias torch.Size([2304])
|
65 |
+
encoder.modality_transformers.vision.blocks.5.attn.out_proj.weight torch.Size([768, 768])
|
66 |
+
encoder.modality_transformers.vision.blocks.5.attn.out_proj.bias torch.Size([768])
|
67 |
+
encoder.modality_transformers.vision.blocks.5.norm1.weight torch.Size([768])
|
68 |
+
encoder.modality_transformers.vision.blocks.5.norm2.weight torch.Size([768])
|
69 |
+
encoder.modality_transformers.vision.blocks.5.mlp.w1.weight torch.Size([512, 768])
|
70 |
+
encoder.modality_transformers.vision.blocks.5.mlp.w2.weight torch.Size([768, 512])
|
71 |
+
encoder.modality_transformers.vision.blocks.5.mlp.w3.weight torch.Size([512, 768])
|
72 |
+
encoder.modality_transformers.vision.blocks.6.attn.in_proj_weight torch.Size([2304, 768])
|
73 |
+
encoder.modality_transformers.vision.blocks.6.attn.in_proj_bias torch.Size([2304])
|
74 |
+
encoder.modality_transformers.vision.blocks.6.attn.out_proj.weight torch.Size([768, 768])
|
75 |
+
encoder.modality_transformers.vision.blocks.6.attn.out_proj.bias torch.Size([768])
|
76 |
+
encoder.modality_transformers.vision.blocks.6.norm1.weight torch.Size([768])
|
77 |
+
encoder.modality_transformers.vision.blocks.6.norm2.weight torch.Size([768])
|
78 |
+
encoder.modality_transformers.vision.blocks.6.mlp.w1.weight torch.Size([512, 768])
|
79 |
+
encoder.modality_transformers.vision.blocks.6.mlp.w2.weight torch.Size([768, 512])
|
80 |
+
encoder.modality_transformers.vision.blocks.6.mlp.w3.weight torch.Size([512, 768])
|
81 |
+
encoder.modality_transformers.vision.blocks.7.attn.in_proj_weight torch.Size([2304, 768])
|
82 |
+
encoder.modality_transformers.vision.blocks.7.attn.in_proj_bias torch.Size([2304])
|
83 |
+
encoder.modality_transformers.vision.blocks.7.attn.out_proj.weight torch.Size([768, 768])
|
84 |
+
encoder.modality_transformers.vision.blocks.7.attn.out_proj.bias torch.Size([768])
|
85 |
+
encoder.modality_transformers.vision.blocks.7.norm1.weight torch.Size([768])
|
86 |
+
encoder.modality_transformers.vision.blocks.7.norm2.weight torch.Size([768])
|
87 |
+
encoder.modality_transformers.vision.blocks.7.mlp.w1.weight torch.Size([512, 768])
|
88 |
+
encoder.modality_transformers.vision.blocks.7.mlp.w2.weight torch.Size([768, 512])
|
89 |
+
encoder.modality_transformers.vision.blocks.7.mlp.w3.weight torch.Size([512, 768])
|
90 |
+
encoder.modality_transformers.vision.blocks.8.attn.in_proj_weight torch.Size([2304, 768])
|
91 |
+
encoder.modality_transformers.vision.blocks.8.attn.in_proj_bias torch.Size([2304])
|
92 |
+
encoder.modality_transformers.vision.blocks.8.attn.out_proj.weight torch.Size([768, 768])
|
93 |
+
encoder.modality_transformers.vision.blocks.8.attn.out_proj.bias torch.Size([768])
|
94 |
+
encoder.modality_transformers.vision.blocks.8.norm1.weight torch.Size([768])
|
95 |
+
encoder.modality_transformers.vision.blocks.8.norm2.weight torch.Size([768])
|
96 |
+
encoder.modality_transformers.vision.blocks.8.mlp.w1.weight torch.Size([512, 768])
|
97 |
+
encoder.modality_transformers.vision.blocks.8.mlp.w2.weight torch.Size([768, 512])
|
98 |
+
encoder.modality_transformers.vision.blocks.8.mlp.w3.weight torch.Size([512, 768])
|
99 |
+
encoder.modality_transformers.vision.blocks.9.attn.in_proj_weight torch.Size([2304, 768])
|
100 |
+
encoder.modality_transformers.vision.blocks.9.attn.in_proj_bias torch.Size([2304])
|
101 |
+
encoder.modality_transformers.vision.blocks.9.attn.out_proj.weight torch.Size([768, 768])
|
102 |
+
encoder.modality_transformers.vision.blocks.9.attn.out_proj.bias torch.Size([768])
|
103 |
+
encoder.modality_transformers.vision.blocks.9.norm1.weight torch.Size([768])
|
104 |
+
encoder.modality_transformers.vision.blocks.9.norm2.weight torch.Size([768])
|
105 |
+
encoder.modality_transformers.vision.blocks.9.mlp.w1.weight torch.Size([512, 768])
|
106 |
+
encoder.modality_transformers.vision.blocks.9.mlp.w2.weight torch.Size([768, 512])
|
107 |
+
encoder.modality_transformers.vision.blocks.9.mlp.w3.weight torch.Size([512, 768])
|
108 |
+
encoder.modality_transformers.vision.blocks.10.attn.in_proj_weight torch.Size([2304, 768])
|
109 |
+
encoder.modality_transformers.vision.blocks.10.attn.in_proj_bias torch.Size([2304])
|
110 |
+
encoder.modality_transformers.vision.blocks.10.attn.out_proj.weight torch.Size([768, 768])
|
111 |
+
encoder.modality_transformers.vision.blocks.10.attn.out_proj.bias torch.Size([768])
|
112 |
+
encoder.modality_transformers.vision.blocks.10.norm1.weight torch.Size([768])
|
113 |
+
encoder.modality_transformers.vision.blocks.10.norm2.weight torch.Size([768])
|
114 |
+
encoder.modality_transformers.vision.blocks.10.mlp.w1.weight torch.Size([512, 768])
|
115 |
+
encoder.modality_transformers.vision.blocks.10.mlp.w2.weight torch.Size([768, 512])
|
116 |
+
encoder.modality_transformers.vision.blocks.10.mlp.w3.weight torch.Size([512, 768])
|
117 |
+
encoder.modality_transformers.vision.blocks.11.attn.in_proj_weight torch.Size([2304, 768])
|
118 |
+
encoder.modality_transformers.vision.blocks.11.attn.in_proj_bias torch.Size([2304])
|
119 |
+
encoder.modality_transformers.vision.blocks.11.attn.out_proj.weight torch.Size([768, 768])
|
120 |
+
encoder.modality_transformers.vision.blocks.11.attn.out_proj.bias torch.Size([768])
|
121 |
+
encoder.modality_transformers.vision.blocks.11.norm1.weight torch.Size([768])
|
122 |
+
encoder.modality_transformers.vision.blocks.11.norm2.weight torch.Size([768])
|
123 |
+
encoder.modality_transformers.vision.blocks.11.mlp.w1.weight torch.Size([512, 768])
|
124 |
+
encoder.modality_transformers.vision.blocks.11.mlp.w2.weight torch.Size([768, 512])
|
125 |
+
encoder.modality_transformers.vision.blocks.11.mlp.w3.weight torch.Size([512, 768])
|
126 |
+
encoder.modality_transformers.audio.pre_transformer_layer.0.weight torch.Size([768])
|
127 |
+
encoder.modality_transformers.audio.blocks.0.attn.in_proj_weight torch.Size([2304, 768])
|
128 |
+
encoder.modality_transformers.audio.blocks.0.attn.in_proj_bias torch.Size([2304])
|
129 |
+
encoder.modality_transformers.audio.blocks.0.attn.bias_k torch.Size([1, 1, 768])
|
130 |
+
encoder.modality_transformers.audio.blocks.0.attn.bias_v torch.Size([1, 1, 768])
|
131 |
+
encoder.modality_transformers.audio.blocks.0.attn.out_proj.weight torch.Size([768, 768])
|
132 |
+
encoder.modality_transformers.audio.blocks.0.attn.out_proj.bias torch.Size([768])
|
133 |
+
encoder.modality_transformers.audio.blocks.0.norm1.weight torch.Size([768])
|
134 |
+
encoder.modality_transformers.audio.blocks.0.norm2.weight torch.Size([768])
|
135 |
+
encoder.modality_transformers.audio.blocks.0.mlp.w1.weight torch.Size([512, 768])
|
136 |
+
encoder.modality_transformers.audio.blocks.0.mlp.w2.weight torch.Size([768, 512])
|
137 |
+
encoder.modality_transformers.audio.blocks.0.mlp.w3.weight torch.Size([512, 768])
|
138 |
+
encoder.modality_transformers.audio.blocks.1.attn.in_proj_weight torch.Size([2304, 768])
|
139 |
+
encoder.modality_transformers.audio.blocks.1.attn.in_proj_bias torch.Size([2304])
|
140 |
+
encoder.modality_transformers.audio.blocks.1.attn.bias_k torch.Size([1, 1, 768])
|
141 |
+
encoder.modality_transformers.audio.blocks.1.attn.bias_v torch.Size([1, 1, 768])
|
142 |
+
encoder.modality_transformers.audio.blocks.1.attn.out_proj.weight torch.Size([768, 768])
|
143 |
+
encoder.modality_transformers.audio.blocks.1.attn.out_proj.bias torch.Size([768])
|
144 |
+
encoder.modality_transformers.audio.blocks.1.norm1.weight torch.Size([768])
|
145 |
+
encoder.modality_transformers.audio.blocks.1.norm2.weight torch.Size([768])
|
146 |
+
encoder.modality_transformers.audio.blocks.1.mlp.w1.weight torch.Size([512, 768])
|
147 |
+
encoder.modality_transformers.audio.blocks.1.mlp.w2.weight torch.Size([768, 512])
|
148 |
+
encoder.modality_transformers.audio.blocks.1.mlp.w3.weight torch.Size([512, 768])
|
149 |
+
encoder.modality_transformers.audio.blocks.2.attn.in_proj_weight torch.Size([2304, 768])
|
150 |
+
encoder.modality_transformers.audio.blocks.2.attn.in_proj_bias torch.Size([2304])
|
151 |
+
encoder.modality_transformers.audio.blocks.2.attn.bias_k torch.Size([1, 1, 768])
|
152 |
+
encoder.modality_transformers.audio.blocks.2.attn.bias_v torch.Size([1, 1, 768])
|
153 |
+
encoder.modality_transformers.audio.blocks.2.attn.out_proj.weight torch.Size([768, 768])
|
154 |
+
encoder.modality_transformers.audio.blocks.2.attn.out_proj.bias torch.Size([768])
|
155 |
+
encoder.modality_transformers.audio.blocks.2.norm1.weight torch.Size([768])
|
156 |
+
encoder.modality_transformers.audio.blocks.2.norm2.weight torch.Size([768])
|
157 |
+
encoder.modality_transformers.audio.blocks.2.mlp.w1.weight torch.Size([512, 768])
|
158 |
+
encoder.modality_transformers.audio.blocks.2.mlp.w2.weight torch.Size([768, 512])
|
159 |
+
encoder.modality_transformers.audio.blocks.2.mlp.w3.weight torch.Size([512, 768])
|
160 |
+
encoder.modality_transformers.audio.blocks.3.attn.in_proj_weight torch.Size([2304, 768])
|
161 |
+
encoder.modality_transformers.audio.blocks.3.attn.in_proj_bias torch.Size([2304])
|
162 |
+
encoder.modality_transformers.audio.blocks.3.attn.bias_k torch.Size([1, 1, 768])
|
163 |
+
encoder.modality_transformers.audio.blocks.3.attn.bias_v torch.Size([1, 1, 768])
|
164 |
+
encoder.modality_transformers.audio.blocks.3.attn.out_proj.weight torch.Size([768, 768])
|
165 |
+
encoder.modality_transformers.audio.blocks.3.attn.out_proj.bias torch.Size([768])
|
166 |
+
encoder.modality_transformers.audio.blocks.3.norm1.weight torch.Size([768])
|
167 |
+
encoder.modality_transformers.audio.blocks.3.norm2.weight torch.Size([768])
|
168 |
+
encoder.modality_transformers.audio.blocks.3.mlp.w1.weight torch.Size([512, 768])
|
169 |
+
encoder.modality_transformers.audio.blocks.3.mlp.w2.weight torch.Size([768, 512])
|
170 |
+
encoder.modality_transformers.audio.blocks.3.mlp.w3.weight torch.Size([512, 768])
|
171 |
+
encoder.modality_transformers.audio.blocks.4.attn.in_proj_weight torch.Size([2304, 768])
|
172 |
+
encoder.modality_transformers.audio.blocks.4.attn.in_proj_bias torch.Size([2304])
|
173 |
+
encoder.modality_transformers.audio.blocks.4.attn.bias_k torch.Size([1, 1, 768])
|
174 |
+
encoder.modality_transformers.audio.blocks.4.attn.bias_v torch.Size([1, 1, 768])
|
175 |
+
encoder.modality_transformers.audio.blocks.4.attn.out_proj.weight torch.Size([768, 768])
|
176 |
+
encoder.modality_transformers.audio.blocks.4.attn.out_proj.bias torch.Size([768])
|
177 |
+
encoder.modality_transformers.audio.blocks.4.norm1.weight torch.Size([768])
|
178 |
+
encoder.modality_transformers.audio.blocks.4.norm2.weight torch.Size([768])
|
179 |
+
encoder.modality_transformers.audio.blocks.4.mlp.w1.weight torch.Size([512, 768])
|
180 |
+
encoder.modality_transformers.audio.blocks.4.mlp.w2.weight torch.Size([768, 512])
|
181 |
+
encoder.modality_transformers.audio.blocks.4.mlp.w3.weight torch.Size([512, 768])
|
182 |
+
encoder.modality_transformers.audio.blocks.5.attn.in_proj_weight torch.Size([2304, 768])
|
183 |
+
encoder.modality_transformers.audio.blocks.5.attn.in_proj_bias torch.Size([2304])
|
184 |
+
encoder.modality_transformers.audio.blocks.5.attn.bias_k torch.Size([1, 1, 768])
|
185 |
+
encoder.modality_transformers.audio.blocks.5.attn.bias_v torch.Size([1, 1, 768])
|
186 |
+
encoder.modality_transformers.audio.blocks.5.attn.out_proj.weight torch.Size([768, 768])
|
187 |
+
encoder.modality_transformers.audio.blocks.5.attn.out_proj.bias torch.Size([768])
|
188 |
+
encoder.modality_transformers.audio.blocks.5.norm1.weight torch.Size([768])
|
189 |
+
encoder.modality_transformers.audio.blocks.5.norm2.weight torch.Size([768])
|
190 |
+
encoder.modality_transformers.audio.blocks.5.mlp.w1.weight torch.Size([512, 768])
|
191 |
+
encoder.modality_transformers.audio.blocks.5.mlp.w2.weight torch.Size([768, 512])
|
192 |
+
encoder.modality_transformers.audio.blocks.5.mlp.w3.weight torch.Size([512, 768])
|
193 |
+
encoder.modality_transformers.audio.blocks.6.attn.in_proj_weight torch.Size([2304, 768])
|
194 |
+
encoder.modality_transformers.audio.blocks.6.attn.in_proj_bias torch.Size([2304])
|
195 |
+
encoder.modality_transformers.audio.blocks.6.attn.bias_k torch.Size([1, 1, 768])
|
196 |
+
encoder.modality_transformers.audio.blocks.6.attn.bias_v torch.Size([1, 1, 768])
|
197 |
+
encoder.modality_transformers.audio.blocks.6.attn.out_proj.weight torch.Size([768, 768])
|
198 |
+
encoder.modality_transformers.audio.blocks.6.attn.out_proj.bias torch.Size([768])
|
199 |
+
encoder.modality_transformers.audio.blocks.6.norm1.weight torch.Size([768])
|
200 |
+
encoder.modality_transformers.audio.blocks.6.norm2.weight torch.Size([768])
|
201 |
+
encoder.modality_transformers.audio.blocks.6.mlp.w1.weight torch.Size([512, 768])
|
202 |
+
encoder.modality_transformers.audio.blocks.6.mlp.w2.weight torch.Size([768, 512])
|
203 |
+
encoder.modality_transformers.audio.blocks.6.mlp.w3.weight torch.Size([512, 768])
|
204 |
+
encoder.modality_transformers.audio.blocks.7.attn.in_proj_weight torch.Size([2304, 768])
|
205 |
+
encoder.modality_transformers.audio.blocks.7.attn.in_proj_bias torch.Size([2304])
|
206 |
+
encoder.modality_transformers.audio.blocks.7.attn.bias_k torch.Size([1, 1, 768])
|
207 |
+
encoder.modality_transformers.audio.blocks.7.attn.bias_v torch.Size([1, 1, 768])
|
208 |
+
encoder.modality_transformers.audio.blocks.7.attn.out_proj.weight torch.Size([768, 768])
|
209 |
+
encoder.modality_transformers.audio.blocks.7.attn.out_proj.bias torch.Size([768])
|
210 |
+
encoder.modality_transformers.audio.blocks.7.norm1.weight torch.Size([768])
|
211 |
+
encoder.modality_transformers.audio.blocks.7.norm2.weight torch.Size([768])
|
212 |
+
encoder.modality_transformers.audio.blocks.7.mlp.w1.weight torch.Size([512, 768])
|
213 |
+
encoder.modality_transformers.audio.blocks.7.mlp.w2.weight torch.Size([768, 512])
|
214 |
+
encoder.modality_transformers.audio.blocks.7.mlp.w3.weight torch.Size([512, 768])
|
215 |
+
encoder.modality_transformers.audio.blocks.8.attn.in_proj_weight torch.Size([2304, 768])
|
216 |
+
encoder.modality_transformers.audio.blocks.8.attn.in_proj_bias torch.Size([2304])
|
217 |
+
encoder.modality_transformers.audio.blocks.8.attn.bias_k torch.Size([1, 1, 768])
|
218 |
+
encoder.modality_transformers.audio.blocks.8.attn.bias_v torch.Size([1, 1, 768])
|
219 |
+
encoder.modality_transformers.audio.blocks.8.attn.out_proj.weight torch.Size([768, 768])
|
220 |
+
encoder.modality_transformers.audio.blocks.8.attn.out_proj.bias torch.Size([768])
|
221 |
+
encoder.modality_transformers.audio.blocks.8.norm1.weight torch.Size([768])
|
222 |
+
encoder.modality_transformers.audio.blocks.8.norm2.weight torch.Size([768])
|
223 |
+
encoder.modality_transformers.audio.blocks.8.mlp.w1.weight torch.Size([512, 768])
|
224 |
+
encoder.modality_transformers.audio.blocks.8.mlp.w2.weight torch.Size([768, 512])
|
225 |
+
encoder.modality_transformers.audio.blocks.8.mlp.w3.weight torch.Size([512, 768])
|
226 |
+
encoder.modality_transformers.audio.blocks.9.attn.in_proj_weight torch.Size([2304, 768])
|
227 |
+
encoder.modality_transformers.audio.blocks.9.attn.in_proj_bias torch.Size([2304])
|
228 |
+
encoder.modality_transformers.audio.blocks.9.attn.bias_k torch.Size([1, 1, 768])
|
229 |
+
encoder.modality_transformers.audio.blocks.9.attn.bias_v torch.Size([1, 1, 768])
|
230 |
+
encoder.modality_transformers.audio.blocks.9.attn.out_proj.weight torch.Size([768, 768])
|
231 |
+
encoder.modality_transformers.audio.blocks.9.attn.out_proj.bias torch.Size([768])
|
232 |
+
encoder.modality_transformers.audio.blocks.9.norm1.weight torch.Size([768])
|
233 |
+
encoder.modality_transformers.audio.blocks.9.norm2.weight torch.Size([768])
|
234 |
+
encoder.modality_transformers.audio.blocks.9.mlp.w1.weight torch.Size([512, 768])
|
235 |
+
encoder.modality_transformers.audio.blocks.9.mlp.w2.weight torch.Size([768, 512])
|
236 |
+
encoder.modality_transformers.audio.blocks.9.mlp.w3.weight torch.Size([512, 768])
|
237 |
+
encoder.modality_transformers.audio.blocks.10.attn.in_proj_weight torch.Size([2304, 768])
|
238 |
+
encoder.modality_transformers.audio.blocks.10.attn.in_proj_bias torch.Size([2304])
|
239 |
+
encoder.modality_transformers.audio.blocks.10.attn.bias_k torch.Size([1, 1, 768])
|
240 |
+
encoder.modality_transformers.audio.blocks.10.attn.bias_v torch.Size([1, 1, 768])
|
241 |
+
encoder.modality_transformers.audio.blocks.10.attn.out_proj.weight torch.Size([768, 768])
|
242 |
+
encoder.modality_transformers.audio.blocks.10.attn.out_proj.bias torch.Size([768])
|
243 |
+
encoder.modality_transformers.audio.blocks.10.norm1.weight torch.Size([768])
|
244 |
+
encoder.modality_transformers.audio.blocks.10.norm2.weight torch.Size([768])
|
245 |
+
encoder.modality_transformers.audio.blocks.10.mlp.w1.weight torch.Size([512, 768])
|
246 |
+
encoder.modality_transformers.audio.blocks.10.mlp.w2.weight torch.Size([768, 512])
|
247 |
+
encoder.modality_transformers.audio.blocks.10.mlp.w3.weight torch.Size([512, 768])
|
248 |
+
encoder.modality_transformers.audio.blocks.11.attn.in_proj_weight torch.Size([2304, 768])
|
249 |
+
encoder.modality_transformers.audio.blocks.11.attn.in_proj_bias torch.Size([2304])
|
250 |
+
encoder.modality_transformers.audio.blocks.11.attn.bias_k torch.Size([1, 1, 768])
|
251 |
+
encoder.modality_transformers.audio.blocks.11.attn.bias_v torch.Size([1, 1, 768])
|
252 |
+
encoder.modality_transformers.audio.blocks.11.attn.out_proj.weight torch.Size([768, 768])
|
253 |
+
encoder.modality_transformers.audio.blocks.11.attn.out_proj.bias torch.Size([768])
|
254 |
+
encoder.modality_transformers.audio.blocks.11.norm1.weight torch.Size([768])
|
255 |
+
encoder.modality_transformers.audio.blocks.11.norm2.weight torch.Size([768])
|
256 |
+
encoder.modality_transformers.audio.blocks.11.mlp.w1.weight torch.Size([512, 768])
|
257 |
+
encoder.modality_transformers.audio.blocks.11.mlp.w2.weight torch.Size([768, 512])
|
258 |
+
encoder.modality_transformers.audio.blocks.11.mlp.w3.weight torch.Size([512, 768])
|
259 |
+
encoder.modality_transformers.depth.pre_transformer_layer.0.weight torch.Size([384])
|
260 |
+
encoder.modality_transformers.depth.blocks.0.attn.in_proj_weight torch.Size([1152, 384])
|
261 |
+
encoder.modality_transformers.depth.blocks.0.attn.in_proj_bias torch.Size([1152])
|
262 |
+
encoder.modality_transformers.depth.blocks.0.attn.bias_k torch.Size([1, 1, 384])
|
263 |
+
encoder.modality_transformers.depth.blocks.0.attn.bias_v torch.Size([1, 1, 384])
|
264 |
+
encoder.modality_transformers.depth.blocks.0.attn.out_proj.weight torch.Size([384, 384])
|
265 |
+
encoder.modality_transformers.depth.blocks.0.attn.out_proj.bias torch.Size([384])
|
266 |
+
encoder.modality_transformers.depth.blocks.0.norm1.weight torch.Size([384])
|
267 |
+
encoder.modality_transformers.depth.blocks.0.norm2.weight torch.Size([384])
|
268 |
+
encoder.modality_transformers.depth.blocks.0.mlp.w1.weight torch.Size([256, 384])
|
269 |
+
encoder.modality_transformers.depth.blocks.0.mlp.w2.weight torch.Size([384, 256])
|
270 |
+
encoder.modality_transformers.depth.blocks.0.mlp.w3.weight torch.Size([256, 384])
|
271 |
+
encoder.modality_transformers.depth.blocks.1.attn.in_proj_weight torch.Size([1152, 384])
|
272 |
+
encoder.modality_transformers.depth.blocks.1.attn.in_proj_bias torch.Size([1152])
|
273 |
+
encoder.modality_transformers.depth.blocks.1.attn.bias_k torch.Size([1, 1, 384])
|
274 |
+
encoder.modality_transformers.depth.blocks.1.attn.bias_v torch.Size([1, 1, 384])
|
275 |
+
encoder.modality_transformers.depth.blocks.1.attn.out_proj.weight torch.Size([384, 384])
|
276 |
+
encoder.modality_transformers.depth.blocks.1.attn.out_proj.bias torch.Size([384])
|
277 |
+
encoder.modality_transformers.depth.blocks.1.norm1.weight torch.Size([384])
|
278 |
+
encoder.modality_transformers.depth.blocks.1.norm2.weight torch.Size([384])
|
279 |
+
encoder.modality_transformers.depth.blocks.1.mlp.w1.weight torch.Size([256, 384])
|
280 |
+
encoder.modality_transformers.depth.blocks.1.mlp.w2.weight torch.Size([384, 256])
|
281 |
+
encoder.modality_transformers.depth.blocks.1.mlp.w3.weight torch.Size([256, 384])
|
282 |
+
encoder.modality_transformers.depth.blocks.2.attn.in_proj_weight torch.Size([1152, 384])
|
283 |
+
encoder.modality_transformers.depth.blocks.2.attn.in_proj_bias torch.Size([1152])
|
284 |
+
encoder.modality_transformers.depth.blocks.2.attn.bias_k torch.Size([1, 1, 384])
|
285 |
+
encoder.modality_transformers.depth.blocks.2.attn.bias_v torch.Size([1, 1, 384])
|
286 |
+
encoder.modality_transformers.depth.blocks.2.attn.out_proj.weight torch.Size([384, 384])
|
287 |
+
encoder.modality_transformers.depth.blocks.2.attn.out_proj.bias torch.Size([384])
|
288 |
+
encoder.modality_transformers.depth.blocks.2.norm1.weight torch.Size([384])
|
289 |
+
encoder.modality_transformers.depth.blocks.2.norm2.weight torch.Size([384])
|
290 |
+
encoder.modality_transformers.depth.blocks.2.mlp.w1.weight torch.Size([256, 384])
|
291 |
+
encoder.modality_transformers.depth.blocks.2.mlp.w2.weight torch.Size([384, 256])
|
292 |
+
encoder.modality_transformers.depth.blocks.2.mlp.w3.weight torch.Size([256, 384])
|
293 |
+
encoder.modality_transformers.depth.blocks.3.attn.in_proj_weight torch.Size([1152, 384])
|
294 |
+
encoder.modality_transformers.depth.blocks.3.attn.in_proj_bias torch.Size([1152])
|
295 |
+
encoder.modality_transformers.depth.blocks.3.attn.bias_k torch.Size([1, 1, 384])
|
296 |
+
encoder.modality_transformers.depth.blocks.3.attn.bias_v torch.Size([1, 1, 384])
|
297 |
+
encoder.modality_transformers.depth.blocks.3.attn.out_proj.weight torch.Size([384, 384])
|
298 |
+
encoder.modality_transformers.depth.blocks.3.attn.out_proj.bias torch.Size([384])
|
299 |
+
encoder.modality_transformers.depth.blocks.3.norm1.weight torch.Size([384])
|
300 |
+
encoder.modality_transformers.depth.blocks.3.norm2.weight torch.Size([384])
|
301 |
+
encoder.modality_transformers.depth.blocks.3.mlp.w1.weight torch.Size([256, 384])
|
302 |
+
encoder.modality_transformers.depth.blocks.3.mlp.w2.weight torch.Size([384, 256])
|
303 |
+
encoder.modality_transformers.depth.blocks.3.mlp.w3.weight torch.Size([256, 384])
|
304 |
+
encoder.modality_transformers.depth.blocks.4.attn.in_proj_weight torch.Size([1152, 384])
|
305 |
+
encoder.modality_transformers.depth.blocks.4.attn.in_proj_bias torch.Size([1152])
|
306 |
+
encoder.modality_transformers.depth.blocks.4.attn.bias_k torch.Size([1, 1, 384])
|
307 |
+
encoder.modality_transformers.depth.blocks.4.attn.bias_v torch.Size([1, 1, 384])
|
308 |
+
encoder.modality_transformers.depth.blocks.4.attn.out_proj.weight torch.Size([384, 384])
|
309 |
+
encoder.modality_transformers.depth.blocks.4.attn.out_proj.bias torch.Size([384])
|
310 |
+
encoder.modality_transformers.depth.blocks.4.norm1.weight torch.Size([384])
|
311 |
+
encoder.modality_transformers.depth.blocks.4.norm2.weight torch.Size([384])
|
312 |
+
encoder.modality_transformers.depth.blocks.4.mlp.w1.weight torch.Size([256, 384])
|
313 |
+
encoder.modality_transformers.depth.blocks.4.mlp.w2.weight torch.Size([384, 256])
|
314 |
+
encoder.modality_transformers.depth.blocks.4.mlp.w3.weight torch.Size([256, 384])
|
315 |
+
encoder.modality_transformers.depth.blocks.5.attn.in_proj_weight torch.Size([1152, 384])
|
316 |
+
encoder.modality_transformers.depth.blocks.5.attn.in_proj_bias torch.Size([1152])
|
317 |
+
encoder.modality_transformers.depth.blocks.5.attn.bias_k torch.Size([1, 1, 384])
|
318 |
+
encoder.modality_transformers.depth.blocks.5.attn.bias_v torch.Size([1, 1, 384])
|
319 |
+
encoder.modality_transformers.depth.blocks.5.attn.out_proj.weight torch.Size([384, 384])
|
320 |
+
encoder.modality_transformers.depth.blocks.5.attn.out_proj.bias torch.Size([384])
|
321 |
+
encoder.modality_transformers.depth.blocks.5.norm1.weight torch.Size([384])
|
322 |
+
encoder.modality_transformers.depth.blocks.5.norm2.weight torch.Size([384])
|
323 |
+
encoder.modality_transformers.depth.blocks.5.mlp.w1.weight torch.Size([256, 384])
|
324 |
+
encoder.modality_transformers.depth.blocks.5.mlp.w2.weight torch.Size([384, 256])
|
325 |
+
encoder.modality_transformers.depth.blocks.5.mlp.w3.weight torch.Size([256, 384])
|
326 |
+
encoder.modality_transformers.thermal.pre_transformer_layer.0.weight torch.Size([768])
|
327 |
+
encoder.modality_transformers.thermal.blocks.0.attn.in_proj_weight torch.Size([2304, 768])
|
328 |
+
encoder.modality_transformers.thermal.blocks.0.attn.in_proj_bias torch.Size([2304])
|
329 |
+
encoder.modality_transformers.thermal.blocks.0.attn.bias_k torch.Size([1, 1, 768])
|
330 |
+
encoder.modality_transformers.thermal.blocks.0.attn.bias_v torch.Size([1, 1, 768])
|
331 |
+
encoder.modality_transformers.thermal.blocks.0.attn.out_proj.weight torch.Size([768, 768])
|
332 |
+
encoder.modality_transformers.thermal.blocks.0.attn.out_proj.bias torch.Size([768])
|
333 |
+
encoder.modality_transformers.thermal.blocks.0.norm1.weight torch.Size([768])
|
334 |
+
encoder.modality_transformers.thermal.blocks.0.norm2.weight torch.Size([768])
|
335 |
+
encoder.modality_transformers.thermal.blocks.0.mlp.w1.weight torch.Size([512, 768])
|
336 |
+
encoder.modality_transformers.thermal.blocks.0.mlp.w2.weight torch.Size([768, 512])
|
337 |
+
encoder.modality_transformers.thermal.blocks.0.mlp.w3.weight torch.Size([512, 768])
|
338 |
+
encoder.modality_transformers.thermal.blocks.1.attn.in_proj_weight torch.Size([2304, 768])
|
339 |
+
encoder.modality_transformers.thermal.blocks.1.attn.in_proj_bias torch.Size([2304])
|
340 |
+
encoder.modality_transformers.thermal.blocks.1.attn.bias_k torch.Size([1, 1, 768])
|
341 |
+
encoder.modality_transformers.thermal.blocks.1.attn.bias_v torch.Size([1, 1, 768])
|
342 |
+
encoder.modality_transformers.thermal.blocks.1.attn.out_proj.weight torch.Size([768, 768])
|
343 |
+
encoder.modality_transformers.thermal.blocks.1.attn.out_proj.bias torch.Size([768])
|
344 |
+
encoder.modality_transformers.thermal.blocks.1.norm1.weight torch.Size([768])
|
345 |
+
encoder.modality_transformers.thermal.blocks.1.norm2.weight torch.Size([768])
|
346 |
+
encoder.modality_transformers.thermal.blocks.1.mlp.w1.weight torch.Size([512, 768])
|
347 |
+
encoder.modality_transformers.thermal.blocks.1.mlp.w2.weight torch.Size([768, 512])
|
348 |
+
encoder.modality_transformers.thermal.blocks.1.mlp.w3.weight torch.Size([512, 768])
|
349 |
+
encoder.modality_transformers.thermal.blocks.2.attn.in_proj_weight torch.Size([2304, 768])
|
350 |
+
encoder.modality_transformers.thermal.blocks.2.attn.in_proj_bias torch.Size([2304])
|
351 |
+
encoder.modality_transformers.thermal.blocks.2.attn.bias_k torch.Size([1, 1, 768])
|
352 |
+
encoder.modality_transformers.thermal.blocks.2.attn.bias_v torch.Size([1, 1, 768])
|
353 |
+
encoder.modality_transformers.thermal.blocks.2.attn.out_proj.weight torch.Size([768, 768])
|
354 |
+
encoder.modality_transformers.thermal.blocks.2.attn.out_proj.bias torch.Size([768])
|
355 |
+
encoder.modality_transformers.thermal.blocks.2.norm1.weight torch.Size([768])
|
356 |
+
encoder.modality_transformers.thermal.blocks.2.norm2.weight torch.Size([768])
|
357 |
+
encoder.modality_transformers.thermal.blocks.2.mlp.w1.weight torch.Size([512, 768])
|
358 |
+
encoder.modality_transformers.thermal.blocks.2.mlp.w2.weight torch.Size([768, 512])
|
359 |
+
encoder.modality_transformers.thermal.blocks.2.mlp.w3.weight torch.Size([512, 768])
|
360 |
+
encoder.modality_transformers.thermal.blocks.3.attn.in_proj_weight torch.Size([2304, 768])
|
361 |
+
encoder.modality_transformers.thermal.blocks.3.attn.in_proj_bias torch.Size([2304])
|
362 |
+
encoder.modality_transformers.thermal.blocks.3.attn.bias_k torch.Size([1, 1, 768])
|
363 |
+
encoder.modality_transformers.thermal.blocks.3.attn.bias_v torch.Size([1, 1, 768])
|
364 |
+
encoder.modality_transformers.thermal.blocks.3.attn.out_proj.weight torch.Size([768, 768])
|
365 |
+
encoder.modality_transformers.thermal.blocks.3.attn.out_proj.bias torch.Size([768])
|
366 |
+
encoder.modality_transformers.thermal.blocks.3.norm1.weight torch.Size([768])
|
367 |
+
encoder.modality_transformers.thermal.blocks.3.norm2.weight torch.Size([768])
|
368 |
+
encoder.modality_transformers.thermal.blocks.3.mlp.w1.weight torch.Size([512, 768])
|
369 |
+
encoder.modality_transformers.thermal.blocks.3.mlp.w2.weight torch.Size([768, 512])
|
370 |
+
encoder.modality_transformers.thermal.blocks.3.mlp.w3.weight torch.Size([512, 768])
|
371 |
+
encoder.modality_transformers.thermal.blocks.4.attn.in_proj_weight torch.Size([2304, 768])
|
372 |
+
encoder.modality_transformers.thermal.blocks.4.attn.in_proj_bias torch.Size([2304])
|
373 |
+
encoder.modality_transformers.thermal.blocks.4.attn.bias_k torch.Size([1, 1, 768])
|
374 |
+
encoder.modality_transformers.thermal.blocks.4.attn.bias_v torch.Size([1, 1, 768])
|
375 |
+
encoder.modality_transformers.thermal.blocks.4.attn.out_proj.weight torch.Size([768, 768])
|
376 |
+
encoder.modality_transformers.thermal.blocks.4.attn.out_proj.bias torch.Size([768])
|
377 |
+
encoder.modality_transformers.thermal.blocks.4.norm1.weight torch.Size([768])
|
378 |
+
encoder.modality_transformers.thermal.blocks.4.norm2.weight torch.Size([768])
|
379 |
+
encoder.modality_transformers.thermal.blocks.4.mlp.w1.weight torch.Size([512, 768])
|
380 |
+
encoder.modality_transformers.thermal.blocks.4.mlp.w2.weight torch.Size([768, 512])
|
381 |
+
encoder.modality_transformers.thermal.blocks.4.mlp.w3.weight torch.Size([512, 768])
|
382 |
+
encoder.modality_transformers.thermal.blocks.5.attn.in_proj_weight torch.Size([2304, 768])
|
383 |
+
encoder.modality_transformers.thermal.blocks.5.attn.in_proj_bias torch.Size([2304])
|
384 |
+
encoder.modality_transformers.thermal.blocks.5.attn.bias_k torch.Size([1, 1, 768])
|
385 |
+
encoder.modality_transformers.thermal.blocks.5.attn.bias_v torch.Size([1, 1, 768])
|
386 |
+
encoder.modality_transformers.thermal.blocks.5.attn.out_proj.weight torch.Size([768, 768])
|
387 |
+
encoder.modality_transformers.thermal.blocks.5.attn.out_proj.bias torch.Size([768])
|
388 |
+
encoder.modality_transformers.thermal.blocks.5.norm1.weight torch.Size([768])
|
389 |
+
encoder.modality_transformers.thermal.blocks.5.norm2.weight torch.Size([768])
|
390 |
+
encoder.modality_transformers.thermal.blocks.5.mlp.w1.weight torch.Size([512, 768])
|
391 |
+
encoder.modality_transformers.thermal.blocks.5.mlp.w2.weight torch.Size([768, 512])
|
392 |
+
encoder.modality_transformers.thermal.blocks.5.mlp.w3.weight torch.Size([512, 768])
|
393 |
+
encoder.modality_heads.vision.0.weight torch.Size([768])
|
394 |
+
encoder.modality_heads.vision.2.weight torch.Size([1024, 768])
|
395 |
+
encoder.modality_heads.audio.0.weight torch.Size([768])
|
396 |
+
encoder.modality_heads.audio.2.weight torch.Size([1024, 768])
|
397 |
+
encoder.modality_heads.depth.0.weight torch.Size([384])
|
398 |
+
encoder.modality_heads.depth.2.weight torch.Size([1024, 384])
|
399 |
+
encoder.modality_heads.thermal.0.weight torch.Size([768])
|
400 |
+
encoder.modality_heads.thermal.2.weight torch.Size([1024, 768])
|
401 |
+
reasoner.model.embed_tokens.weight torch.Size([151936, 896])
|
402 |
+
reasoner.model.layers.0.self_attn.q_proj.weight torch.Size([896, 896])
|
403 |
+
reasoner.model.layers.0.self_attn.q_proj.bias torch.Size([896])
|
404 |
+
reasoner.model.layers.0.self_attn.k_proj.weight torch.Size([128, 896])
|
405 |
+
reasoner.model.layers.0.self_attn.k_proj.bias torch.Size([128])
|
406 |
+
reasoner.model.layers.0.self_attn.v_proj.weight torch.Size([128, 896])
|
407 |
+
reasoner.model.layers.0.self_attn.v_proj.bias torch.Size([128])
|
408 |
+
reasoner.model.layers.0.self_attn.o_proj.weight torch.Size([896, 896])
|
409 |
+
reasoner.model.layers.0.mlp.gate_proj.weight torch.Size([4864, 896])
|
410 |
+
reasoner.model.layers.0.mlp.up_proj.weight torch.Size([4864, 896])
|
411 |
+
reasoner.model.layers.0.mlp.down_proj.weight torch.Size([896, 4864])
|
412 |
+
reasoner.model.layers.0.input_layernorm.weight torch.Size([896])
|
413 |
+
reasoner.model.layers.0.post_attention_layernorm.weight torch.Size([896])
|
414 |
+
reasoner.model.layers.1.self_attn.q_proj.weight torch.Size([896, 896])
|
415 |
+
reasoner.model.layers.1.self_attn.q_proj.bias torch.Size([896])
|
416 |
+
reasoner.model.layers.1.self_attn.k_proj.weight torch.Size([128, 896])
|
417 |
+
reasoner.model.layers.1.self_attn.k_proj.bias torch.Size([128])
|
418 |
+
reasoner.model.layers.1.self_attn.v_proj.weight torch.Size([128, 896])
|
419 |
+
reasoner.model.layers.1.self_attn.v_proj.bias torch.Size([128])
|
420 |
+
reasoner.model.layers.1.self_attn.o_proj.weight torch.Size([896, 896])
|
421 |
+
reasoner.model.layers.1.mlp.gate_proj.weight torch.Size([4864, 896])
|
422 |
+
reasoner.model.layers.1.mlp.up_proj.weight torch.Size([4864, 896])
|
423 |
+
reasoner.model.layers.1.mlp.down_proj.weight torch.Size([896, 4864])
|
424 |
+
reasoner.model.layers.1.input_layernorm.weight torch.Size([896])
|
425 |
+
reasoner.model.layers.1.post_attention_layernorm.weight torch.Size([896])
|
426 |
+
reasoner.model.layers.2.self_attn.q_proj.weight torch.Size([896, 896])
|
427 |
+
reasoner.model.layers.2.self_attn.q_proj.bias torch.Size([896])
|
428 |
+
reasoner.model.layers.2.self_attn.k_proj.weight torch.Size([128, 896])
|
429 |
+
reasoner.model.layers.2.self_attn.k_proj.bias torch.Size([128])
|
430 |
+
reasoner.model.layers.2.self_attn.v_proj.weight torch.Size([128, 896])
|
431 |
+
reasoner.model.layers.2.self_attn.v_proj.bias torch.Size([128])
|
432 |
+
reasoner.model.layers.2.self_attn.o_proj.weight torch.Size([896, 896])
|
433 |
+
reasoner.model.layers.2.mlp.gate_proj.weight torch.Size([4864, 896])
|
434 |
+
reasoner.model.layers.2.mlp.up_proj.weight torch.Size([4864, 896])
|
435 |
+
reasoner.model.layers.2.mlp.down_proj.weight torch.Size([896, 4864])
|
436 |
+
reasoner.model.layers.2.input_layernorm.weight torch.Size([896])
|
437 |
+
reasoner.model.layers.2.post_attention_layernorm.weight torch.Size([896])
|
438 |
+
reasoner.model.layers.3.self_attn.q_proj.weight torch.Size([896, 896])
|
439 |
+
reasoner.model.layers.3.self_attn.q_proj.bias torch.Size([896])
|
440 |
+
reasoner.model.layers.3.self_attn.k_proj.weight torch.Size([128, 896])
|
441 |
+
reasoner.model.layers.3.self_attn.k_proj.bias torch.Size([128])
|
442 |
+
reasoner.model.layers.3.self_attn.v_proj.weight torch.Size([128, 896])
|
443 |
+
reasoner.model.layers.3.self_attn.v_proj.bias torch.Size([128])
|
444 |
+
reasoner.model.layers.3.self_attn.o_proj.weight torch.Size([896, 896])
|
445 |
+
reasoner.model.layers.3.mlp.gate_proj.weight torch.Size([4864, 896])
|
446 |
+
reasoner.model.layers.3.mlp.up_proj.weight torch.Size([4864, 896])
|
447 |
+
reasoner.model.layers.3.mlp.down_proj.weight torch.Size([896, 4864])
|
448 |
+
reasoner.model.layers.3.input_layernorm.weight torch.Size([896])
|
449 |
+
reasoner.model.layers.3.post_attention_layernorm.weight torch.Size([896])
|
450 |
+
reasoner.model.layers.4.self_attn.q_proj.weight torch.Size([896, 896])
|
451 |
+
reasoner.model.layers.4.self_attn.q_proj.bias torch.Size([896])
|
452 |
+
reasoner.model.layers.4.self_attn.k_proj.weight torch.Size([128, 896])
|
453 |
+
reasoner.model.layers.4.self_attn.k_proj.bias torch.Size([128])
|
454 |
+
reasoner.model.layers.4.self_attn.v_proj.weight torch.Size([128, 896])
|
455 |
+
reasoner.model.layers.4.self_attn.v_proj.bias torch.Size([128])
|
456 |
+
reasoner.model.layers.4.self_attn.o_proj.weight torch.Size([896, 896])
|
457 |
+
reasoner.model.layers.4.mlp.gate_proj.weight torch.Size([4864, 896])
|
458 |
+
reasoner.model.layers.4.mlp.up_proj.weight torch.Size([4864, 896])
|
459 |
+
reasoner.model.layers.4.mlp.down_proj.weight torch.Size([896, 4864])
|
460 |
+
reasoner.model.layers.4.input_layernorm.weight torch.Size([896])
|
461 |
+
reasoner.model.layers.4.post_attention_layernorm.weight torch.Size([896])
|
462 |
+
reasoner.model.layers.5.self_attn.q_proj.weight torch.Size([896, 896])
|
463 |
+
reasoner.model.layers.5.self_attn.q_proj.bias torch.Size([896])
|
464 |
+
reasoner.model.layers.5.self_attn.k_proj.weight torch.Size([128, 896])
|
465 |
+
reasoner.model.layers.5.self_attn.k_proj.bias torch.Size([128])
|
466 |
+
reasoner.model.layers.5.self_attn.v_proj.weight torch.Size([128, 896])
|
467 |
+
reasoner.model.layers.5.self_attn.v_proj.bias torch.Size([128])
|
468 |
+
reasoner.model.layers.5.self_attn.o_proj.weight torch.Size([896, 896])
|
469 |
+
reasoner.model.layers.5.mlp.gate_proj.weight torch.Size([4864, 896])
|
470 |
+
reasoner.model.layers.5.mlp.up_proj.weight torch.Size([4864, 896])
|
471 |
+
reasoner.model.layers.5.mlp.down_proj.weight torch.Size([896, 4864])
|
472 |
+
reasoner.model.layers.5.input_layernorm.weight torch.Size([896])
|
473 |
+
reasoner.model.layers.5.post_attention_layernorm.weight torch.Size([896])
|
474 |
+
reasoner.model.layers.6.self_attn.q_proj.weight torch.Size([896, 896])
|
475 |
+
reasoner.model.layers.6.self_attn.q_proj.bias torch.Size([896])
|
476 |
+
reasoner.model.layers.6.self_attn.k_proj.weight torch.Size([128, 896])
|
477 |
+
reasoner.model.layers.6.self_attn.k_proj.bias torch.Size([128])
|
478 |
+
reasoner.model.layers.6.self_attn.v_proj.weight torch.Size([128, 896])
|
479 |
+
reasoner.model.layers.6.self_attn.v_proj.bias torch.Size([128])
|
480 |
+
reasoner.model.layers.6.self_attn.o_proj.weight torch.Size([896, 896])
|
481 |
+
reasoner.model.layers.6.mlp.gate_proj.weight torch.Size([4864, 896])
|
482 |
+
reasoner.model.layers.6.mlp.up_proj.weight torch.Size([4864, 896])
|
483 |
+
reasoner.model.layers.6.mlp.down_proj.weight torch.Size([896, 4864])
|
484 |
+
reasoner.model.layers.6.input_layernorm.weight torch.Size([896])
|
485 |
+
reasoner.model.layers.6.post_attention_layernorm.weight torch.Size([896])
|
486 |
+
reasoner.model.layers.7.self_attn.q_proj.weight torch.Size([896, 896])
|
487 |
+
reasoner.model.layers.7.self_attn.q_proj.bias torch.Size([896])
|
488 |
+
reasoner.model.layers.7.self_attn.k_proj.weight torch.Size([128, 896])
|
489 |
+
reasoner.model.layers.7.self_attn.k_proj.bias torch.Size([128])
|
490 |
+
reasoner.model.layers.7.self_attn.v_proj.weight torch.Size([128, 896])
|
491 |
+
reasoner.model.layers.7.self_attn.v_proj.bias torch.Size([128])
|
492 |
+
reasoner.model.layers.7.self_attn.o_proj.weight torch.Size([896, 896])
|
493 |
+
reasoner.model.layers.7.mlp.gate_proj.weight torch.Size([4864, 896])
|
494 |
+
reasoner.model.layers.7.mlp.up_proj.weight torch.Size([4864, 896])
|
495 |
+
reasoner.model.layers.7.mlp.down_proj.weight torch.Size([896, 4864])
|
496 |
+
reasoner.model.layers.7.input_layernorm.weight torch.Size([896])
|
497 |
+
reasoner.model.layers.7.post_attention_layernorm.weight torch.Size([896])
|
498 |
+
reasoner.model.layers.8.self_attn.q_proj.weight torch.Size([896, 896])
|
499 |
+
reasoner.model.layers.8.self_attn.q_proj.bias torch.Size([896])
|
500 |
+
reasoner.model.layers.8.self_attn.k_proj.weight torch.Size([128, 896])
|
501 |
+
reasoner.model.layers.8.self_attn.k_proj.bias torch.Size([128])
|
502 |
+
reasoner.model.layers.8.self_attn.v_proj.weight torch.Size([128, 896])
|
503 |
+
reasoner.model.layers.8.self_attn.v_proj.bias torch.Size([128])
|
504 |
+
reasoner.model.layers.8.self_attn.o_proj.weight torch.Size([896, 896])
|
505 |
+
reasoner.model.layers.8.mlp.gate_proj.weight torch.Size([4864, 896])
|
506 |
+
reasoner.model.layers.8.mlp.up_proj.weight torch.Size([4864, 896])
|
507 |
+
reasoner.model.layers.8.mlp.down_proj.weight torch.Size([896, 4864])
|
508 |
+
reasoner.model.layers.8.input_layernorm.weight torch.Size([896])
|
509 |
+
reasoner.model.layers.8.post_attention_layernorm.weight torch.Size([896])
|
510 |
+
reasoner.model.layers.9.self_attn.q_proj.weight torch.Size([896, 896])
|
511 |
+
reasoner.model.layers.9.self_attn.q_proj.bias torch.Size([896])
|
512 |
+
reasoner.model.layers.9.self_attn.k_proj.weight torch.Size([128, 896])
|
513 |
+
reasoner.model.layers.9.self_attn.k_proj.bias torch.Size([128])
|
514 |
+
reasoner.model.layers.9.self_attn.v_proj.weight torch.Size([128, 896])
|
515 |
+
reasoner.model.layers.9.self_attn.v_proj.bias torch.Size([128])
|
516 |
+
reasoner.model.layers.9.self_attn.o_proj.weight torch.Size([896, 896])
|
517 |
+
reasoner.model.layers.9.mlp.gate_proj.weight torch.Size([4864, 896])
|
518 |
+
reasoner.model.layers.9.mlp.up_proj.weight torch.Size([4864, 896])
|
519 |
+
reasoner.model.layers.9.mlp.down_proj.weight torch.Size([896, 4864])
|
520 |
+
reasoner.model.layers.9.input_layernorm.weight torch.Size([896])
|
521 |
+
reasoner.model.layers.9.post_attention_layernorm.weight torch.Size([896])
|
522 |
+
reasoner.model.layers.10.self_attn.q_proj.weight torch.Size([896, 896])
|
523 |
+
reasoner.model.layers.10.self_attn.q_proj.bias torch.Size([896])
|
524 |
+
reasoner.model.layers.10.self_attn.k_proj.weight torch.Size([128, 896])
|
525 |
+
reasoner.model.layers.10.self_attn.k_proj.bias torch.Size([128])
|
526 |
+
reasoner.model.layers.10.self_attn.v_proj.weight torch.Size([128, 896])
|
527 |
+
reasoner.model.layers.10.self_attn.v_proj.bias torch.Size([128])
|
528 |
+
reasoner.model.layers.10.self_attn.o_proj.weight torch.Size([896, 896])
|
529 |
+
reasoner.model.layers.10.mlp.gate_proj.weight torch.Size([4864, 896])
|
530 |
+
reasoner.model.layers.10.mlp.up_proj.weight torch.Size([4864, 896])
|
531 |
+
reasoner.model.layers.10.mlp.down_proj.weight torch.Size([896, 4864])
|
532 |
+
reasoner.model.layers.10.input_layernorm.weight torch.Size([896])
|
533 |
+
reasoner.model.layers.10.post_attention_layernorm.weight torch.Size([896])
|
534 |
+
reasoner.model.layers.11.self_attn.q_proj.weight torch.Size([896, 896])
|
535 |
+
reasoner.model.layers.11.self_attn.q_proj.bias torch.Size([896])
|
536 |
+
reasoner.model.layers.11.self_attn.k_proj.weight torch.Size([128, 896])
|
537 |
+
reasoner.model.layers.11.self_attn.k_proj.bias torch.Size([128])
|
538 |
+
reasoner.model.layers.11.self_attn.v_proj.weight torch.Size([128, 896])
|
539 |
+
reasoner.model.layers.11.self_attn.v_proj.bias torch.Size([128])
|
540 |
+
reasoner.model.layers.11.self_attn.o_proj.weight torch.Size([896, 896])
|
541 |
+
reasoner.model.layers.11.mlp.gate_proj.weight torch.Size([4864, 896])
|
542 |
+
reasoner.model.layers.11.mlp.up_proj.weight torch.Size([4864, 896])
|
543 |
+
reasoner.model.layers.11.mlp.down_proj.weight torch.Size([896, 4864])
|
544 |
+
reasoner.model.layers.11.input_layernorm.weight torch.Size([896])
|
545 |
+
reasoner.model.layers.11.post_attention_layernorm.weight torch.Size([896])
|
546 |
+
reasoner.model.layers.12.self_attn.q_proj.weight torch.Size([896, 896])
|
547 |
+
reasoner.model.layers.12.self_attn.q_proj.bias torch.Size([896])
|
548 |
+
reasoner.model.layers.12.self_attn.k_proj.weight torch.Size([128, 896])
|
549 |
+
reasoner.model.layers.12.self_attn.k_proj.bias torch.Size([128])
|
550 |
+
reasoner.model.layers.12.self_attn.v_proj.weight torch.Size([128, 896])
|
551 |
+
reasoner.model.layers.12.self_attn.v_proj.bias torch.Size([128])
|
552 |
+
reasoner.model.layers.12.self_attn.o_proj.weight torch.Size([896, 896])
|
553 |
+
reasoner.model.layers.12.mlp.gate_proj.weight torch.Size([4864, 896])
|
554 |
+
reasoner.model.layers.12.mlp.up_proj.weight torch.Size([4864, 896])
|
555 |
+
reasoner.model.layers.12.mlp.down_proj.weight torch.Size([896, 4864])
|
556 |
+
reasoner.model.layers.12.input_layernorm.weight torch.Size([896])
|
557 |
+
reasoner.model.layers.12.post_attention_layernorm.weight torch.Size([896])
|
558 |
+
reasoner.model.layers.13.self_attn.q_proj.weight torch.Size([896, 896])
|
559 |
+
reasoner.model.layers.13.self_attn.q_proj.bias torch.Size([896])
|
560 |
+
reasoner.model.layers.13.self_attn.k_proj.weight torch.Size([128, 896])
|
561 |
+
reasoner.model.layers.13.self_attn.k_proj.bias torch.Size([128])
|
562 |
+
reasoner.model.layers.13.self_attn.v_proj.weight torch.Size([128, 896])
|
563 |
+
reasoner.model.layers.13.self_attn.v_proj.bias torch.Size([128])
|
564 |
+
reasoner.model.layers.13.self_attn.o_proj.weight torch.Size([896, 896])
|
565 |
+
reasoner.model.layers.13.mlp.gate_proj.weight torch.Size([4864, 896])
|
566 |
+
reasoner.model.layers.13.mlp.up_proj.weight torch.Size([4864, 896])
|
567 |
+
reasoner.model.layers.13.mlp.down_proj.weight torch.Size([896, 4864])
|
568 |
+
reasoner.model.layers.13.input_layernorm.weight torch.Size([896])
|
569 |
+
reasoner.model.layers.13.post_attention_layernorm.weight torch.Size([896])
|
570 |
+
reasoner.model.layers.14.self_attn.q_proj.weight torch.Size([896, 896])
|
571 |
+
reasoner.model.layers.14.self_attn.q_proj.bias torch.Size([896])
|
572 |
+
reasoner.model.layers.14.self_attn.k_proj.weight torch.Size([128, 896])
|
573 |
+
reasoner.model.layers.14.self_attn.k_proj.bias torch.Size([128])
|
574 |
+
reasoner.model.layers.14.self_attn.v_proj.weight torch.Size([128, 896])
|
575 |
+
reasoner.model.layers.14.self_attn.v_proj.bias torch.Size([128])
|
576 |
+
reasoner.model.layers.14.self_attn.o_proj.weight torch.Size([896, 896])
|
577 |
+
reasoner.model.layers.14.mlp.gate_proj.weight torch.Size([4864, 896])
|
578 |
+
reasoner.model.layers.14.mlp.up_proj.weight torch.Size([4864, 896])
|
579 |
+
reasoner.model.layers.14.mlp.down_proj.weight torch.Size([896, 4864])
|
580 |
+
reasoner.model.layers.14.input_layernorm.weight torch.Size([896])
|
581 |
+
reasoner.model.layers.14.post_attention_layernorm.weight torch.Size([896])
|
582 |
+
reasoner.model.layers.15.self_attn.q_proj.weight torch.Size([896, 896])
|
583 |
+
reasoner.model.layers.15.self_attn.q_proj.bias torch.Size([896])
|
584 |
+
reasoner.model.layers.15.self_attn.k_proj.weight torch.Size([128, 896])
|
585 |
+
reasoner.model.layers.15.self_attn.k_proj.bias torch.Size([128])
|
586 |
+
reasoner.model.layers.15.self_attn.v_proj.weight torch.Size([128, 896])
|
587 |
+
reasoner.model.layers.15.self_attn.v_proj.bias torch.Size([128])
|
588 |
+
reasoner.model.layers.15.self_attn.o_proj.weight torch.Size([896, 896])
|
589 |
+
reasoner.model.layers.15.mlp.gate_proj.weight torch.Size([4864, 896])
|
590 |
+
reasoner.model.layers.15.mlp.up_proj.weight torch.Size([4864, 896])
|
591 |
+
reasoner.model.layers.15.mlp.down_proj.weight torch.Size([896, 4864])
|
592 |
+
reasoner.model.layers.15.input_layernorm.weight torch.Size([896])
|
593 |
+
reasoner.model.layers.15.post_attention_layernorm.weight torch.Size([896])
|
594 |
+
reasoner.model.layers.16.self_attn.q_proj.weight torch.Size([896, 896])
|
595 |
+
reasoner.model.layers.16.self_attn.q_proj.bias torch.Size([896])
|
596 |
+
reasoner.model.layers.16.self_attn.k_proj.weight torch.Size([128, 896])
|
597 |
+
reasoner.model.layers.16.self_attn.k_proj.bias torch.Size([128])
|
598 |
+
reasoner.model.layers.16.self_attn.v_proj.weight torch.Size([128, 896])
|
599 |
+
reasoner.model.layers.16.self_attn.v_proj.bias torch.Size([128])
|
600 |
+
reasoner.model.layers.16.self_attn.o_proj.weight torch.Size([896, 896])
|
601 |
+
reasoner.model.layers.16.mlp.gate_proj.weight torch.Size([4864, 896])
|
602 |
+
reasoner.model.layers.16.mlp.up_proj.weight torch.Size([4864, 896])
|
603 |
+
reasoner.model.layers.16.mlp.down_proj.weight torch.Size([896, 4864])
|
604 |
+
reasoner.model.layers.16.input_layernorm.weight torch.Size([896])
|
605 |
+
reasoner.model.layers.16.post_attention_layernorm.weight torch.Size([896])
|
606 |
+
reasoner.model.layers.17.self_attn.q_proj.weight torch.Size([896, 896])
|
607 |
+
reasoner.model.layers.17.self_attn.q_proj.bias torch.Size([896])
|
608 |
+
reasoner.model.layers.17.self_attn.k_proj.weight torch.Size([128, 896])
|
609 |
+
reasoner.model.layers.17.self_attn.k_proj.bias torch.Size([128])
|
610 |
+
reasoner.model.layers.17.self_attn.v_proj.weight torch.Size([128, 896])
|
611 |
+
reasoner.model.layers.17.self_attn.v_proj.bias torch.Size([128])
|
612 |
+
reasoner.model.layers.17.self_attn.o_proj.weight torch.Size([896, 896])
|
613 |
+
reasoner.model.layers.17.mlp.gate_proj.weight torch.Size([4864, 896])
|
614 |
+
reasoner.model.layers.17.mlp.up_proj.weight torch.Size([4864, 896])
|
615 |
+
reasoner.model.layers.17.mlp.down_proj.weight torch.Size([896, 4864])
|
616 |
+
reasoner.model.layers.17.input_layernorm.weight torch.Size([896])
|
617 |
+
reasoner.model.layers.17.post_attention_layernorm.weight torch.Size([896])
|
618 |
+
reasoner.model.layers.18.self_attn.q_proj.weight torch.Size([896, 896])
|
619 |
+
reasoner.model.layers.18.self_attn.q_proj.bias torch.Size([896])
|
620 |
+
reasoner.model.layers.18.self_attn.k_proj.weight torch.Size([128, 896])
|
621 |
+
reasoner.model.layers.18.self_attn.k_proj.bias torch.Size([128])
|
622 |
+
reasoner.model.layers.18.self_attn.v_proj.weight torch.Size([128, 896])
|
623 |
+
reasoner.model.layers.18.self_attn.v_proj.bias torch.Size([128])
|
624 |
+
reasoner.model.layers.18.self_attn.o_proj.weight torch.Size([896, 896])
|
625 |
+
reasoner.model.layers.18.mlp.gate_proj.weight torch.Size([4864, 896])
|
626 |
+
reasoner.model.layers.18.mlp.up_proj.weight torch.Size([4864, 896])
|
627 |
+
reasoner.model.layers.18.mlp.down_proj.weight torch.Size([896, 4864])
|
628 |
+
reasoner.model.layers.18.input_layernorm.weight torch.Size([896])
|
629 |
+
reasoner.model.layers.18.post_attention_layernorm.weight torch.Size([896])
|
630 |
+
reasoner.model.layers.19.self_attn.q_proj.weight torch.Size([896, 896])
|
631 |
+
reasoner.model.layers.19.self_attn.q_proj.bias torch.Size([896])
|
632 |
+
reasoner.model.layers.19.self_attn.k_proj.weight torch.Size([128, 896])
|
633 |
+
reasoner.model.layers.19.self_attn.k_proj.bias torch.Size([128])
|
634 |
+
reasoner.model.layers.19.self_attn.v_proj.weight torch.Size([128, 896])
|
635 |
+
reasoner.model.layers.19.self_attn.v_proj.bias torch.Size([128])
|
636 |
+
reasoner.model.layers.19.self_attn.o_proj.weight torch.Size([896, 896])
|
637 |
+
reasoner.model.layers.19.mlp.gate_proj.weight torch.Size([4864, 896])
|
638 |
+
reasoner.model.layers.19.mlp.up_proj.weight torch.Size([4864, 896])
|
639 |
+
reasoner.model.layers.19.mlp.down_proj.weight torch.Size([896, 4864])
|
640 |
+
reasoner.model.layers.19.input_layernorm.weight torch.Size([896])
|
641 |
+
reasoner.model.layers.19.post_attention_layernorm.weight torch.Size([896])
|
642 |
+
reasoner.model.layers.20.self_attn.q_proj.weight torch.Size([896, 896])
|
643 |
+
reasoner.model.layers.20.self_attn.q_proj.bias torch.Size([896])
|
644 |
+
reasoner.model.layers.20.self_attn.k_proj.weight torch.Size([128, 896])
|
645 |
+
reasoner.model.layers.20.self_attn.k_proj.bias torch.Size([128])
|
646 |
+
reasoner.model.layers.20.self_attn.v_proj.weight torch.Size([128, 896])
|
647 |
+
reasoner.model.layers.20.self_attn.v_proj.bias torch.Size([128])
|
648 |
+
reasoner.model.layers.20.self_attn.o_proj.weight torch.Size([896, 896])
|
649 |
+
reasoner.model.layers.20.mlp.gate_proj.weight torch.Size([4864, 896])
|
650 |
+
reasoner.model.layers.20.mlp.up_proj.weight torch.Size([4864, 896])
|
651 |
+
reasoner.model.layers.20.mlp.down_proj.weight torch.Size([896, 4864])
|
652 |
+
reasoner.model.layers.20.input_layernorm.weight torch.Size([896])
|
653 |
+
reasoner.model.layers.20.post_attention_layernorm.weight torch.Size([896])
|
654 |
+
reasoner.model.layers.21.self_attn.q_proj.weight torch.Size([896, 896])
|
655 |
+
reasoner.model.layers.21.self_attn.q_proj.bias torch.Size([896])
|
656 |
+
reasoner.model.layers.21.self_attn.k_proj.weight torch.Size([128, 896])
|
657 |
+
reasoner.model.layers.21.self_attn.k_proj.bias torch.Size([128])
|
658 |
+
reasoner.model.layers.21.self_attn.v_proj.weight torch.Size([128, 896])
|
659 |
+
reasoner.model.layers.21.self_attn.v_proj.bias torch.Size([128])
|
660 |
+
reasoner.model.layers.21.self_attn.o_proj.weight torch.Size([896, 896])
|
661 |
+
reasoner.model.layers.21.mlp.gate_proj.weight torch.Size([4864, 896])
|
662 |
+
reasoner.model.layers.21.mlp.up_proj.weight torch.Size([4864, 896])
|
663 |
+
reasoner.model.layers.21.mlp.down_proj.weight torch.Size([896, 4864])
|
664 |
+
reasoner.model.layers.21.input_layernorm.weight torch.Size([896])
|
665 |
+
reasoner.model.layers.21.post_attention_layernorm.weight torch.Size([896])
|
666 |
+
reasoner.model.layers.22.self_attn.q_proj.weight torch.Size([896, 896])
|
667 |
+
reasoner.model.layers.22.self_attn.q_proj.bias torch.Size([896])
|
668 |
+
reasoner.model.layers.22.self_attn.k_proj.weight torch.Size([128, 896])
|
669 |
+
reasoner.model.layers.22.self_attn.k_proj.bias torch.Size([128])
|
670 |
+
reasoner.model.layers.22.self_attn.v_proj.weight torch.Size([128, 896])
|
671 |
+
reasoner.model.layers.22.self_attn.v_proj.bias torch.Size([128])
|
672 |
+
reasoner.model.layers.22.self_attn.o_proj.weight torch.Size([896, 896])
|
673 |
+
reasoner.model.layers.22.mlp.gate_proj.weight torch.Size([4864, 896])
|
674 |
+
reasoner.model.layers.22.mlp.up_proj.weight torch.Size([4864, 896])
|
675 |
+
reasoner.model.layers.22.mlp.down_proj.weight torch.Size([896, 4864])
|
676 |
+
reasoner.model.layers.22.input_layernorm.weight torch.Size([896])
|
677 |
+
reasoner.model.layers.22.post_attention_layernorm.weight torch.Size([896])
|
678 |
+
reasoner.model.layers.23.self_attn.q_proj.weight torch.Size([896, 896])
|
679 |
+
reasoner.model.layers.23.self_attn.q_proj.bias torch.Size([896])
|
680 |
+
reasoner.model.layers.23.self_attn.k_proj.weight torch.Size([128, 896])
|
681 |
+
reasoner.model.layers.23.self_attn.k_proj.bias torch.Size([128])
|
682 |
+
reasoner.model.layers.23.self_attn.v_proj.weight torch.Size([128, 896])
|
683 |
+
reasoner.model.layers.23.self_attn.v_proj.bias torch.Size([128])
|
684 |
+
reasoner.model.layers.23.self_attn.o_proj.weight torch.Size([896, 896])
|
685 |
+
reasoner.model.layers.23.mlp.gate_proj.weight torch.Size([4864, 896])
|
686 |
+
reasoner.model.layers.23.mlp.up_proj.weight torch.Size([4864, 896])
|
687 |
+
reasoner.model.layers.23.mlp.down_proj.weight torch.Size([896, 4864])
|
688 |
+
reasoner.model.layers.23.input_layernorm.weight torch.Size([896])
|
689 |
+
reasoner.model.layers.23.post_attention_layernorm.weight torch.Size([896])
|
690 |
+
reasoner.model.norm.weight torch.Size([896])
|
691 |
+
reasoner.lm_head.weight torch.Size([151936, 896])
|
692 |
+
input_projetor.weight torch.Size([896, 1024])
|
693 |
+
input_projetor.bias torch.Size([896])
|