Upload config.json
Browse files- config.json +270 -0
config.json
ADDED
@@ -0,0 +1,270 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"Transformer": {
|
3 |
+
"encoder": {
|
4 |
+
"sentence_embedding": {
|
5 |
+
"embedding": {
|
6 |
+
"num_embeddings": 71,
|
7 |
+
"embedding_dim": 512
|
8 |
+
},
|
9 |
+
"position_encoder": {},
|
10 |
+
"dropout": 0.1
|
11 |
+
},
|
12 |
+
"layers": [
|
13 |
+
{
|
14 |
+
"attention": {
|
15 |
+
"qkv_layer": {
|
16 |
+
"in_features": 512,
|
17 |
+
"out_features": 1536,
|
18 |
+
"bias": true
|
19 |
+
},
|
20 |
+
"linear_layer": {
|
21 |
+
"in_features": 512,
|
22 |
+
"out_features": 512,
|
23 |
+
"bias": true
|
24 |
+
}
|
25 |
+
},
|
26 |
+
"norm1": {},
|
27 |
+
"dropout1": 0.1,
|
28 |
+
"ffn": {
|
29 |
+
"linear1": {
|
30 |
+
"in_features": 512,
|
31 |
+
"out_features": 2048,
|
32 |
+
"bias": true
|
33 |
+
},
|
34 |
+
"linear2": {
|
35 |
+
"in_features": 2048,
|
36 |
+
"out_features": 512,
|
37 |
+
"bias": true
|
38 |
+
}
|
39 |
+
},
|
40 |
+
"norm2": {},
|
41 |
+
"dropout2": 0.1
|
42 |
+
},
|
43 |
+
{
|
44 |
+
"attention": {
|
45 |
+
"qkv_layer": {
|
46 |
+
"in_features": 512,
|
47 |
+
"out_features": 1536,
|
48 |
+
"bias": true
|
49 |
+
},
|
50 |
+
"linear_layer": {
|
51 |
+
"in_features": 512,
|
52 |
+
"out_features": 512,
|
53 |
+
"bias": true
|
54 |
+
}
|
55 |
+
},
|
56 |
+
"norm1": {},
|
57 |
+
"dropout1": 0.1,
|
58 |
+
"ffn": {
|
59 |
+
"linear1": {
|
60 |
+
"in_features": 512,
|
61 |
+
"out_features": 2048,
|
62 |
+
"bias": true
|
63 |
+
},
|
64 |
+
"linear2": {
|
65 |
+
"in_features": 2048,
|
66 |
+
"out_features": 512,
|
67 |
+
"bias": true
|
68 |
+
}
|
69 |
+
},
|
70 |
+
"norm2": {},
|
71 |
+
"dropout2": 0.1
|
72 |
+
},
|
73 |
+
{
|
74 |
+
"attention": {
|
75 |
+
"qkv_layer": {
|
76 |
+
"in_features": 512,
|
77 |
+
"out_features": 1536,
|
78 |
+
"bias": true
|
79 |
+
},
|
80 |
+
"linear_layer": {
|
81 |
+
"in_features": 512,
|
82 |
+
"out_features": 512,
|
83 |
+
"bias": true
|
84 |
+
}
|
85 |
+
},
|
86 |
+
"norm1": {},
|
87 |
+
"dropout1": 0.1,
|
88 |
+
"ffn": {
|
89 |
+
"linear1": {
|
90 |
+
"in_features": 512,
|
91 |
+
"out_features": 2048,
|
92 |
+
"bias": true
|
93 |
+
},
|
94 |
+
"linear2": {
|
95 |
+
"in_features": 2048,
|
96 |
+
"out_features": 512,
|
97 |
+
"bias": true
|
98 |
+
}
|
99 |
+
},
|
100 |
+
"norm2": {},
|
101 |
+
"dropout2": 0.1
|
102 |
+
}
|
103 |
+
]
|
104 |
+
},
|
105 |
+
"decoder": {
|
106 |
+
"sentence_embedding": {
|
107 |
+
"embedding": {
|
108 |
+
"num_embeddings": 125,
|
109 |
+
"embedding_dim": 512
|
110 |
+
},
|
111 |
+
"position_encoder": {},
|
112 |
+
"dropout": 0.1
|
113 |
+
},
|
114 |
+
"layers": [
|
115 |
+
{
|
116 |
+
"self_attention": {
|
117 |
+
"qkv_layer": {
|
118 |
+
"in_features": 512,
|
119 |
+
"out_features": 1536,
|
120 |
+
"bias": true
|
121 |
+
},
|
122 |
+
"linear_layer": {
|
123 |
+
"in_features": 512,
|
124 |
+
"out_features": 512,
|
125 |
+
"bias": true
|
126 |
+
}
|
127 |
+
},
|
128 |
+
"layer_norm1": {},
|
129 |
+
"dropout1": 0.1,
|
130 |
+
"encoder_decoder_attention": {
|
131 |
+
"kv_layer": {
|
132 |
+
"in_features": 512,
|
133 |
+
"out_features": 1024,
|
134 |
+
"bias": true
|
135 |
+
},
|
136 |
+
"q_layer": {
|
137 |
+
"in_features": 512,
|
138 |
+
"out_features": 512,
|
139 |
+
"bias": true
|
140 |
+
},
|
141 |
+
"linear_layer": {
|
142 |
+
"in_features": 512,
|
143 |
+
"out_features": 512,
|
144 |
+
"bias": true
|
145 |
+
}
|
146 |
+
},
|
147 |
+
"layer_norm2": {},
|
148 |
+
"dropout2": 0.1,
|
149 |
+
"ffn": {
|
150 |
+
"linear1": {
|
151 |
+
"in_features": 512,
|
152 |
+
"out_features": 2048,
|
153 |
+
"bias": true
|
154 |
+
},
|
155 |
+
"linear2": {
|
156 |
+
"in_features": 2048,
|
157 |
+
"out_features": 512,
|
158 |
+
"bias": true
|
159 |
+
}
|
160 |
+
},
|
161 |
+
"layer_norm3": {},
|
162 |
+
"dropout3": 0.1
|
163 |
+
},
|
164 |
+
{
|
165 |
+
"self_attention": {
|
166 |
+
"qkv_layer": {
|
167 |
+
"in_features": 512,
|
168 |
+
"out_features": 1536,
|
169 |
+
"bias": true
|
170 |
+
},
|
171 |
+
"linear_layer": {
|
172 |
+
"in_features": 512,
|
173 |
+
"out_features": 512,
|
174 |
+
"bias": true
|
175 |
+
}
|
176 |
+
},
|
177 |
+
"layer_norm1": {},
|
178 |
+
"dropout1": 0.1,
|
179 |
+
"encoder_decoder_attention": {
|
180 |
+
"kv_layer": {
|
181 |
+
"in_features": 512,
|
182 |
+
"out_features": 1024,
|
183 |
+
"bias": true
|
184 |
+
},
|
185 |
+
"q_layer": {
|
186 |
+
"in_features": 512,
|
187 |
+
"out_features": 512,
|
188 |
+
"bias": true
|
189 |
+
},
|
190 |
+
"linear_layer": {
|
191 |
+
"in_features": 512,
|
192 |
+
"out_features": 512,
|
193 |
+
"bias": true
|
194 |
+
}
|
195 |
+
},
|
196 |
+
"layer_norm2": {},
|
197 |
+
"dropout2": 0.1,
|
198 |
+
"ffn": {
|
199 |
+
"linear1": {
|
200 |
+
"in_features": 512,
|
201 |
+
"out_features": 2048,
|
202 |
+
"bias": true
|
203 |
+
},
|
204 |
+
"linear2": {
|
205 |
+
"in_features": 2048,
|
206 |
+
"out_features": 512,
|
207 |
+
"bias": true
|
208 |
+
}
|
209 |
+
},
|
210 |
+
"layer_norm3": {},
|
211 |
+
"dropout3": 0.1
|
212 |
+
},
|
213 |
+
{
|
214 |
+
"self_attention": {
|
215 |
+
"qkv_layer": {
|
216 |
+
"in_features": 512,
|
217 |
+
"out_features": 1536,
|
218 |
+
"bias": true
|
219 |
+
},
|
220 |
+
"linear_layer": {
|
221 |
+
"in_features": 512,
|
222 |
+
"out_features": 512,
|
223 |
+
"bias": true
|
224 |
+
}
|
225 |
+
},
|
226 |
+
"layer_norm1": {},
|
227 |
+
"dropout1": 0.1,
|
228 |
+
"encoder_decoder_attention": {
|
229 |
+
"kv_layer": {
|
230 |
+
"in_features": 512,
|
231 |
+
"out_features": 1024,
|
232 |
+
"bias": true
|
233 |
+
},
|
234 |
+
"q_layer": {
|
235 |
+
"in_features": 512,
|
236 |
+
"out_features": 512,
|
237 |
+
"bias": true
|
238 |
+
},
|
239 |
+
"linear_layer": {
|
240 |
+
"in_features": 512,
|
241 |
+
"out_features": 512,
|
242 |
+
"bias": true
|
243 |
+
}
|
244 |
+
},
|
245 |
+
"layer_norm2": {},
|
246 |
+
"dropout2": 0.1,
|
247 |
+
"ffn": {
|
248 |
+
"linear1": {
|
249 |
+
"in_features": 512,
|
250 |
+
"out_features": 2048,
|
251 |
+
"bias": true
|
252 |
+
},
|
253 |
+
"linear2": {
|
254 |
+
"in_features": 2048,
|
255 |
+
"out_features": 512,
|
256 |
+
"bias": true
|
257 |
+
}
|
258 |
+
},
|
259 |
+
"layer_norm3": {},
|
260 |
+
"dropout3": 0.1
|
261 |
+
}
|
262 |
+
]
|
263 |
+
},
|
264 |
+
"linear": {
|
265 |
+
"in_features": 512,
|
266 |
+
"out_features": 125,
|
267 |
+
"bias": true
|
268 |
+
}
|
269 |
+
}
|
270 |
+
}
|