Create keys.txt
Browse files- T5_text_encoder/keys.txt +223 -0
T5_text_encoder/keys.txt
ADDED
@@ -0,0 +1,223 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
model-1 keys:
|
2 |
+
encoder.block.0.layer.0.SelfAttention.k.weight
|
3 |
+
encoder.block.0.layer.0.SelfAttention.o.weight
|
4 |
+
encoder.block.0.layer.0.SelfAttention.q.weight
|
5 |
+
encoder.block.0.layer.0.SelfAttention.relative_attention_bias.weight
|
6 |
+
encoder.block.0.layer.0.SelfAttention.v.weight
|
7 |
+
encoder.block.0.layer.0.layer_norm.weight
|
8 |
+
encoder.block.0.layer.1.DenseReluDense.wi_0.weight
|
9 |
+
encoder.block.0.layer.1.DenseReluDense.wi_1.weight
|
10 |
+
encoder.block.0.layer.1.DenseReluDense.wo.weight
|
11 |
+
encoder.block.0.layer.1.layer_norm.weight
|
12 |
+
encoder.block.1.layer.0.SelfAttention.k.weight
|
13 |
+
encoder.block.1.layer.0.SelfAttention.o.weight
|
14 |
+
encoder.block.1.layer.0.SelfAttention.q.weight
|
15 |
+
encoder.block.1.layer.0.SelfAttention.v.weight
|
16 |
+
encoder.block.1.layer.0.layer_norm.weight
|
17 |
+
encoder.block.1.layer.1.DenseReluDense.wi_0.weight
|
18 |
+
encoder.block.1.layer.1.DenseReluDense.wi_1.weight
|
19 |
+
encoder.block.1.layer.1.DenseReluDense.wo.weight
|
20 |
+
encoder.block.1.layer.1.layer_norm.weight
|
21 |
+
encoder.block.10.layer.0.SelfAttention.k.weight
|
22 |
+
encoder.block.10.layer.0.SelfAttention.o.weight
|
23 |
+
encoder.block.10.layer.0.SelfAttention.q.weight
|
24 |
+
encoder.block.10.layer.0.SelfAttention.v.weight
|
25 |
+
encoder.block.10.layer.0.layer_norm.weight
|
26 |
+
encoder.block.10.layer.1.DenseReluDense.wi_0.weight
|
27 |
+
encoder.block.10.layer.1.DenseReluDense.wi_1.weight
|
28 |
+
encoder.block.10.layer.1.DenseReluDense.wo.weight
|
29 |
+
encoder.block.10.layer.1.layer_norm.weight
|
30 |
+
encoder.block.11.layer.0.SelfAttention.k.weight
|
31 |
+
encoder.block.11.layer.0.SelfAttention.o.weight
|
32 |
+
encoder.block.11.layer.0.SelfAttention.q.weight
|
33 |
+
encoder.block.11.layer.0.SelfAttention.v.weight
|
34 |
+
encoder.block.11.layer.0.layer_norm.weight
|
35 |
+
encoder.block.11.layer.1.DenseReluDense.wi_0.weight
|
36 |
+
encoder.block.11.layer.1.DenseReluDense.wi_1.weight
|
37 |
+
encoder.block.11.layer.1.DenseReluDense.wo.weight
|
38 |
+
encoder.block.11.layer.1.layer_norm.weight
|
39 |
+
encoder.block.12.layer.0.SelfAttention.k.weight
|
40 |
+
encoder.block.12.layer.0.SelfAttention.q.weight
|
41 |
+
encoder.block.12.layer.0.SelfAttention.v.weight
|
42 |
+
encoder.block.2.layer.0.SelfAttention.k.weight
|
43 |
+
encoder.block.2.layer.0.SelfAttention.o.weight
|
44 |
+
encoder.block.2.layer.0.SelfAttention.q.weight
|
45 |
+
encoder.block.2.layer.0.SelfAttention.v.weight
|
46 |
+
encoder.block.2.layer.0.layer_norm.weight
|
47 |
+
encoder.block.2.layer.1.DenseReluDense.wi_0.weight
|
48 |
+
encoder.block.2.layer.1.DenseReluDense.wi_1.weight
|
49 |
+
encoder.block.2.layer.1.DenseReluDense.wo.weight
|
50 |
+
encoder.block.2.layer.1.layer_norm.weight
|
51 |
+
encoder.block.3.layer.0.SelfAttention.k.weight
|
52 |
+
encoder.block.3.layer.0.SelfAttention.o.weight
|
53 |
+
encoder.block.3.layer.0.SelfAttention.q.weight
|
54 |
+
encoder.block.3.layer.0.SelfAttention.v.weight
|
55 |
+
encoder.block.3.layer.0.layer_norm.weight
|
56 |
+
encoder.block.3.layer.1.DenseReluDense.wi_0.weight
|
57 |
+
encoder.block.3.layer.1.DenseReluDense.wi_1.weight
|
58 |
+
encoder.block.3.layer.1.DenseReluDense.wo.weight
|
59 |
+
encoder.block.3.layer.1.layer_norm.weight
|
60 |
+
encoder.block.4.layer.0.SelfAttention.k.weight
|
61 |
+
encoder.block.4.layer.0.SelfAttention.o.weight
|
62 |
+
encoder.block.4.layer.0.SelfAttention.q.weight
|
63 |
+
encoder.block.4.layer.0.SelfAttention.v.weight
|
64 |
+
encoder.block.4.layer.0.layer_norm.weight
|
65 |
+
encoder.block.4.layer.1.DenseReluDense.wi_0.weight
|
66 |
+
encoder.block.4.layer.1.DenseReluDense.wi_1.weight
|
67 |
+
encoder.block.4.layer.1.DenseReluDense.wo.weight
|
68 |
+
encoder.block.4.layer.1.layer_norm.weight
|
69 |
+
encoder.block.5.layer.0.SelfAttention.k.weight
|
70 |
+
encoder.block.5.layer.0.SelfAttention.o.weight
|
71 |
+
encoder.block.5.layer.0.SelfAttention.q.weight
|
72 |
+
encoder.block.5.layer.0.SelfAttention.v.weight
|
73 |
+
encoder.block.5.layer.0.layer_norm.weight
|
74 |
+
encoder.block.5.layer.1.DenseReluDense.wi_0.weight
|
75 |
+
encoder.block.5.layer.1.DenseReluDense.wi_1.weight
|
76 |
+
encoder.block.5.layer.1.DenseReluDense.wo.weight
|
77 |
+
encoder.block.5.layer.1.layer_norm.weight
|
78 |
+
encoder.block.6.layer.0.SelfAttention.k.weight
|
79 |
+
encoder.block.6.layer.0.SelfAttention.o.weight
|
80 |
+
encoder.block.6.layer.0.SelfAttention.q.weight
|
81 |
+
encoder.block.6.layer.0.SelfAttention.v.weight
|
82 |
+
encoder.block.6.layer.0.layer_norm.weight
|
83 |
+
encoder.block.6.layer.1.DenseReluDense.wi_0.weight
|
84 |
+
encoder.block.6.layer.1.DenseReluDense.wi_1.weight
|
85 |
+
encoder.block.6.layer.1.DenseReluDense.wo.weight
|
86 |
+
encoder.block.6.layer.1.layer_norm.weight
|
87 |
+
encoder.block.7.layer.0.SelfAttention.k.weight
|
88 |
+
encoder.block.7.layer.0.SelfAttention.o.weight
|
89 |
+
encoder.block.7.layer.0.SelfAttention.q.weight
|
90 |
+
encoder.block.7.layer.0.SelfAttention.v.weight
|
91 |
+
encoder.block.7.layer.0.layer_norm.weight
|
92 |
+
encoder.block.7.layer.1.DenseReluDense.wi_0.weight
|
93 |
+
encoder.block.7.layer.1.DenseReluDense.wi_1.weight
|
94 |
+
encoder.block.7.layer.1.DenseReluDense.wo.weight
|
95 |
+
encoder.block.7.layer.1.layer_norm.weight
|
96 |
+
encoder.block.8.layer.0.SelfAttention.k.weight
|
97 |
+
encoder.block.8.layer.0.SelfAttention.o.weight
|
98 |
+
encoder.block.8.layer.0.SelfAttention.q.weight
|
99 |
+
encoder.block.8.layer.0.SelfAttention.v.weight
|
100 |
+
encoder.block.8.layer.0.layer_norm.weight
|
101 |
+
encoder.block.8.layer.1.DenseReluDense.wi_0.weight
|
102 |
+
encoder.block.8.layer.1.DenseReluDense.wi_1.weight
|
103 |
+
encoder.block.8.layer.1.DenseReluDense.wo.weight
|
104 |
+
encoder.block.8.layer.1.layer_norm.weight
|
105 |
+
encoder.block.9.layer.0.SelfAttention.k.weight
|
106 |
+
encoder.block.9.layer.0.SelfAttention.o.weight
|
107 |
+
encoder.block.9.layer.0.SelfAttention.q.weight
|
108 |
+
encoder.block.9.layer.0.SelfAttention.v.weight
|
109 |
+
encoder.block.9.layer.0.layer_norm.weight
|
110 |
+
encoder.block.9.layer.1.DenseReluDense.wi_0.weight
|
111 |
+
encoder.block.9.layer.1.DenseReluDense.wi_1.weight
|
112 |
+
encoder.block.9.layer.1.DenseReluDense.wo.weight
|
113 |
+
encoder.block.9.layer.1.layer_norm.weight
|
114 |
+
shared.weight
|
115 |
+
|
116 |
+
|
117 |
+
model-2 keys:
|
118 |
+
encoder.block.12.layer.0.SelfAttention.o.weight
|
119 |
+
encoder.block.12.layer.0.layer_norm.weight
|
120 |
+
encoder.block.12.layer.1.DenseReluDense.wi_0.weight
|
121 |
+
encoder.block.12.layer.1.DenseReluDense.wi_1.weight
|
122 |
+
encoder.block.12.layer.1.DenseReluDense.wo.weight
|
123 |
+
encoder.block.12.layer.1.layer_norm.weight
|
124 |
+
encoder.block.13.layer.0.SelfAttention.k.weight
|
125 |
+
encoder.block.13.layer.0.SelfAttention.o.weight
|
126 |
+
encoder.block.13.layer.0.SelfAttention.q.weight
|
127 |
+
encoder.block.13.layer.0.SelfAttention.v.weight
|
128 |
+
encoder.block.13.layer.0.layer_norm.weight
|
129 |
+
encoder.block.13.layer.1.DenseReluDense.wi_0.weight
|
130 |
+
encoder.block.13.layer.1.DenseReluDense.wi_1.weight
|
131 |
+
encoder.block.13.layer.1.DenseReluDense.wo.weight
|
132 |
+
encoder.block.13.layer.1.layer_norm.weight
|
133 |
+
encoder.block.14.layer.0.SelfAttention.k.weight
|
134 |
+
encoder.block.14.layer.0.SelfAttention.o.weight
|
135 |
+
encoder.block.14.layer.0.SelfAttention.q.weight
|
136 |
+
encoder.block.14.layer.0.SelfAttention.v.weight
|
137 |
+
encoder.block.14.layer.0.layer_norm.weight
|
138 |
+
encoder.block.14.layer.1.DenseReluDense.wi_0.weight
|
139 |
+
encoder.block.14.layer.1.DenseReluDense.wi_1.weight
|
140 |
+
encoder.block.14.layer.1.DenseReluDense.wo.weight
|
141 |
+
encoder.block.14.layer.1.layer_norm.weight
|
142 |
+
encoder.block.15.layer.0.SelfAttention.k.weight
|
143 |
+
encoder.block.15.layer.0.SelfAttention.o.weight
|
144 |
+
encoder.block.15.layer.0.SelfAttention.q.weight
|
145 |
+
encoder.block.15.layer.0.SelfAttention.v.weight
|
146 |
+
encoder.block.15.layer.0.layer_norm.weight
|
147 |
+
encoder.block.15.layer.1.DenseReluDense.wi_0.weight
|
148 |
+
encoder.block.15.layer.1.DenseReluDense.wi_1.weight
|
149 |
+
encoder.block.15.layer.1.DenseReluDense.wo.weight
|
150 |
+
encoder.block.15.layer.1.layer_norm.weight
|
151 |
+
encoder.block.16.layer.0.SelfAttention.k.weight
|
152 |
+
encoder.block.16.layer.0.SelfAttention.o.weight
|
153 |
+
encoder.block.16.layer.0.SelfAttention.q.weight
|
154 |
+
encoder.block.16.layer.0.SelfAttention.v.weight
|
155 |
+
encoder.block.16.layer.0.layer_norm.weight
|
156 |
+
encoder.block.16.layer.1.DenseReluDense.wi_0.weight
|
157 |
+
encoder.block.16.layer.1.DenseReluDense.wi_1.weight
|
158 |
+
encoder.block.16.layer.1.DenseReluDense.wo.weight
|
159 |
+
encoder.block.16.layer.1.layer_norm.weight
|
160 |
+
encoder.block.17.layer.0.SelfAttention.k.weight
|
161 |
+
encoder.block.17.layer.0.SelfAttention.o.weight
|
162 |
+
encoder.block.17.layer.0.SelfAttention.q.weight
|
163 |
+
encoder.block.17.layer.0.SelfAttention.v.weight
|
164 |
+
encoder.block.17.layer.0.layer_norm.weight
|
165 |
+
encoder.block.17.layer.1.DenseReluDense.wi_0.weight
|
166 |
+
encoder.block.17.layer.1.DenseReluDense.wi_1.weight
|
167 |
+
encoder.block.17.layer.1.DenseReluDense.wo.weight
|
168 |
+
encoder.block.17.layer.1.layer_norm.weight
|
169 |
+
encoder.block.18.layer.0.SelfAttention.k.weight
|
170 |
+
encoder.block.18.layer.0.SelfAttention.o.weight
|
171 |
+
encoder.block.18.layer.0.SelfAttention.q.weight
|
172 |
+
encoder.block.18.layer.0.SelfAttention.v.weight
|
173 |
+
encoder.block.18.layer.0.layer_norm.weight
|
174 |
+
encoder.block.18.layer.1.DenseReluDense.wi_0.weight
|
175 |
+
encoder.block.18.layer.1.DenseReluDense.wi_1.weight
|
176 |
+
encoder.block.18.layer.1.DenseReluDense.wo.weight
|
177 |
+
encoder.block.18.layer.1.layer_norm.weight
|
178 |
+
encoder.block.19.layer.0.SelfAttention.k.weight
|
179 |
+
encoder.block.19.layer.0.SelfAttention.o.weight
|
180 |
+
encoder.block.19.layer.0.SelfAttention.q.weight
|
181 |
+
encoder.block.19.layer.0.SelfAttention.v.weight
|
182 |
+
encoder.block.19.layer.0.layer_norm.weight
|
183 |
+
encoder.block.19.layer.1.DenseReluDense.wi_0.weight
|
184 |
+
encoder.block.19.layer.1.DenseReluDense.wi_1.weight
|
185 |
+
encoder.block.19.layer.1.DenseReluDense.wo.weight
|
186 |
+
encoder.block.19.layer.1.layer_norm.weight
|
187 |
+
encoder.block.20.layer.0.SelfAttention.k.weight
|
188 |
+
encoder.block.20.layer.0.SelfAttention.o.weight
|
189 |
+
encoder.block.20.layer.0.SelfAttention.q.weight
|
190 |
+
encoder.block.20.layer.0.SelfAttention.v.weight
|
191 |
+
encoder.block.20.layer.0.layer_norm.weight
|
192 |
+
encoder.block.20.layer.1.DenseReluDense.wi_0.weight
|
193 |
+
encoder.block.20.layer.1.DenseReluDense.wi_1.weight
|
194 |
+
encoder.block.20.layer.1.DenseReluDense.wo.weight
|
195 |
+
encoder.block.20.layer.1.layer_norm.weight
|
196 |
+
encoder.block.21.layer.0.SelfAttention.k.weight
|
197 |
+
encoder.block.21.layer.0.SelfAttention.o.weight
|
198 |
+
encoder.block.21.layer.0.SelfAttention.q.weight
|
199 |
+
encoder.block.21.layer.0.SelfAttention.v.weight
|
200 |
+
encoder.block.21.layer.0.layer_norm.weight
|
201 |
+
encoder.block.21.layer.1.DenseReluDense.wi_0.weight
|
202 |
+
encoder.block.21.layer.1.DenseReluDense.wi_1.weight
|
203 |
+
encoder.block.21.layer.1.DenseReluDense.wo.weight
|
204 |
+
encoder.block.21.layer.1.layer_norm.weight
|
205 |
+
encoder.block.22.layer.0.SelfAttention.k.weight
|
206 |
+
encoder.block.22.layer.0.SelfAttention.o.weight
|
207 |
+
encoder.block.22.layer.0.SelfAttention.q.weight
|
208 |
+
encoder.block.22.layer.0.SelfAttention.v.weight
|
209 |
+
encoder.block.22.layer.0.layer_norm.weight
|
210 |
+
encoder.block.22.layer.1.DenseReluDense.wi_0.weight
|
211 |
+
encoder.block.22.layer.1.DenseReluDense.wi_1.weight
|
212 |
+
encoder.block.22.layer.1.DenseReluDense.wo.weight
|
213 |
+
encoder.block.22.layer.1.layer_norm.weight
|
214 |
+
encoder.block.23.layer.0.SelfAttention.k.weight
|
215 |
+
encoder.block.23.layer.0.SelfAttention.o.weight
|
216 |
+
encoder.block.23.layer.0.SelfAttention.q.weight
|
217 |
+
encoder.block.23.layer.0.SelfAttention.v.weight
|
218 |
+
encoder.block.23.layer.0.layer_norm.weight
|
219 |
+
encoder.block.23.layer.1.DenseReluDense.wi_0.weight
|
220 |
+
encoder.block.23.layer.1.DenseReluDense.wi_1.weight
|
221 |
+
encoder.block.23.layer.1.DenseReluDense.wo.weight
|
222 |
+
encoder.block.23.layer.1.layer_norm.weight
|
223 |
+
encoder.final_layer_norm.weight
|