Hugging Face's logo

Spaces:
pokeberrypie
/
Trying
Runtime error

App Files Community
pokeberrypie commited on
Commit
2e5718e
·
verified ·
1 Parent(s): 74a0653

Delete model.safetensors.index.json

Browse files
Files changed (1) hide show
  1. model.safetensors.index.json +0 -651
model.safetensors.index.json DELETED
@@ -1,651 +0,0 @@
1
- {
2
- "metadata": {
3
- "total_size": 359045130240
4
- },
5
- "weight_map": {
6
- "lm_head.weight": "model-00081-of-00081.safetensors",
7
- "transformer.h.0.ln_attn.bias": "model-00002-of-00081.safetensors",
8
- "transformer.h.0.ln_attn.weight": "model-00002-of-00081.safetensors",
9
- "transformer.h.0.ln_mlp.bias": "model-00002-of-00081.safetensors",
10
- "transformer.h.0.ln_mlp.weight": "model-00002-of-00081.safetensors",
11
- "transformer.h.0.mlp.dense_4h_to_h.weight": "model-00002-of-00081.safetensors",
12
- "transformer.h.0.mlp.dense_h_to_4h.weight": "model-00001-of-00081.safetensors",
13
- "transformer.h.0.self_attention.dense.weight": "model-00001-of-00081.safetensors",
14
- "transformer.h.0.self_attention.query_key_value.weight": "model-00001-of-00081.safetensors",
15
- "transformer.h.1.ln_attn.bias": "model-00003-of-00081.safetensors",
16
- "transformer.h.1.ln_attn.weight": "model-00003-of-00081.safetensors",
17
- "transformer.h.1.ln_mlp.bias": "model-00003-of-00081.safetensors",
18
- "transformer.h.1.ln_mlp.weight": "model-00003-of-00081.safetensors",
19
- "transformer.h.1.mlp.dense_4h_to_h.weight": "model-00003-of-00081.safetensors",
20
- "transformer.h.1.mlp.dense_h_to_4h.weight": "model-00002-of-00081.safetensors",
21
- "transformer.h.1.self_attention.dense.weight": "model-00002-of-00081.safetensors",
22
- "transformer.h.1.self_attention.query_key_value.weight": "model-00002-of-00081.safetensors",
23
- "transformer.h.10.ln_attn.bias": "model-00012-of-00081.safetensors",
24
- "transformer.h.10.ln_attn.weight": "model-00012-of-00081.safetensors",
25
- "transformer.h.10.ln_mlp.bias": "model-00012-of-00081.safetensors",
26
- "transformer.h.10.ln_mlp.weight": "model-00012-of-00081.safetensors",
27
- "transformer.h.10.mlp.dense_4h_to_h.weight": "model-00012-of-00081.safetensors",
28
- "transformer.h.10.mlp.dense_h_to_4h.weight": "model-00011-of-00081.safetensors",
29
- "transformer.h.10.self_attention.dense.weight": "model-00011-of-00081.safetensors",
30
- "transformer.h.10.self_attention.query_key_value.weight": "model-00011-of-00081.safetensors",
31
- "transformer.h.11.ln_attn.bias": "model-00013-of-00081.safetensors",
32
- "transformer.h.11.ln_attn.weight": "model-00013-of-00081.safetensors",
33
- "transformer.h.11.ln_mlp.bias": "model-00013-of-00081.safetensors",
34
- "transformer.h.11.ln_mlp.weight": "model-00013-of-00081.safetensors",
35
- "transformer.h.11.mlp.dense_4h_to_h.weight": "model-00013-of-00081.safetensors",
36
- "transformer.h.11.mlp.dense_h_to_4h.weight": "model-00012-of-00081.safetensors",
37
- "transformer.h.11.self_attention.dense.weight": "model-00012-of-00081.safetensors",
38
- "transformer.h.11.self_attention.query_key_value.weight": "model-00012-of-00081.safetensors",
39
- "transformer.h.12.ln_attn.bias": "model-00014-of-00081.safetensors",
40
- "transformer.h.12.ln_attn.weight": "model-00014-of-00081.safetensors",
41
- "transformer.h.12.ln_mlp.bias": "model-00014-of-00081.safetensors",
42
- "transformer.h.12.ln_mlp.weight": "model-00014-of-00081.safetensors",
43
- "transformer.h.12.mlp.dense_4h_to_h.weight": "model-00014-of-00081.safetensors",
44
- "transformer.h.12.mlp.dense_h_to_4h.weight": "model-00013-of-00081.safetensors",
45
- "transformer.h.12.self_attention.dense.weight": "model-00013-of-00081.safetensors",
46
- "transformer.h.12.self_attention.query_key_value.weight": "model-00013-of-00081.safetensors",
47
- "transformer.h.13.ln_attn.bias": "model-00015-of-00081.safetensors",
48
- "transformer.h.13.ln_attn.weight": "model-00015-of-00081.safetensors",
49
- "transformer.h.13.ln_mlp.bias": "model-00015-of-00081.safetensors",
50
- "transformer.h.13.ln_mlp.weight": "model-00015-of-00081.safetensors",
51
- "transformer.h.13.mlp.dense_4h_to_h.weight": "model-00015-of-00081.safetensors",
52
- "transformer.h.13.mlp.dense_h_to_4h.weight": "model-00014-of-00081.safetensors",
53
- "transformer.h.13.self_attention.dense.weight": "model-00014-of-00081.safetensors",
54
- "transformer.h.13.self_attention.query_key_value.weight": "model-00014-of-00081.safetensors",
55
- "transformer.h.14.ln_attn.bias": "model-00016-of-00081.safetensors",
56
- "transformer.h.14.ln_attn.weight": "model-00016-of-00081.safetensors",
57
- "transformer.h.14.ln_mlp.bias": "model-00016-of-00081.safetensors",
58
- "transformer.h.14.ln_mlp.weight": "model-00016-of-00081.safetensors",
59
- "transformer.h.14.mlp.dense_4h_to_h.weight": "model-00016-of-00081.safetensors",
60
- "transformer.h.14.mlp.dense_h_to_4h.weight": "model-00015-of-00081.safetensors",
61
- "transformer.h.14.self_attention.dense.weight": "model-00015-of-00081.safetensors",
62
- "transformer.h.14.self_attention.query_key_value.weight": "model-00015-of-00081.safetensors",
63
- "transformer.h.15.ln_attn.bias": "model-00017-of-00081.safetensors",
64
- "transformer.h.15.ln_attn.weight": "model-00017-of-00081.safetensors",
65
- "transformer.h.15.ln_mlp.bias": "model-00017-of-00081.safetensors",
66
- "transformer.h.15.ln_mlp.weight": "model-00017-of-00081.safetensors",
67
- "transformer.h.15.mlp.dense_4h_to_h.weight": "model-00017-of-00081.safetensors",
68
- "transformer.h.15.mlp.dense_h_to_4h.weight": "model-00016-of-00081.safetensors",
69
- "transformer.h.15.self_attention.dense.weight": "model-00016-of-00081.safetensors",
70
- "transformer.h.15.self_attention.query_key_value.weight": "model-00016-of-00081.safetensors",
71
- "transformer.h.16.ln_attn.bias": "model-00018-of-00081.safetensors",
72
- "transformer.h.16.ln_attn.weight": "model-00018-of-00081.safetensors",
73
- "transformer.h.16.ln_mlp.bias": "model-00018-of-00081.safetensors",
74
- "transformer.h.16.ln_mlp.weight": "model-00018-of-00081.safetensors",
75
- "transformer.h.16.mlp.dense_4h_to_h.weight": "model-00018-of-00081.safetensors",
76
- "transformer.h.16.mlp.dense_h_to_4h.weight": "model-00017-of-00081.safetensors",
77
- "transformer.h.16.self_attention.dense.weight": "model-00017-of-00081.safetensors",
78
- "transformer.h.16.self_attention.query_key_value.weight": "model-00017-of-00081.safetensors",
79
- "transformer.h.17.ln_attn.bias": "model-00019-of-00081.safetensors",
80
- "transformer.h.17.ln_attn.weight": "model-00019-of-00081.safetensors",
81
- "transformer.h.17.ln_mlp.bias": "model-00019-of-00081.safetensors",
82
- "transformer.h.17.ln_mlp.weight": "model-00019-of-00081.safetensors",
83
- "transformer.h.17.mlp.dense_4h_to_h.weight": "model-00019-of-00081.safetensors",
84
- "transformer.h.17.mlp.dense_h_to_4h.weight": "model-00018-of-00081.safetensors",
85
- "transformer.h.17.self_attention.dense.weight": "model-00018-of-00081.safetensors",
86
- "transformer.h.17.self_attention.query_key_value.weight": "model-00018-of-00081.safetensors",
87
- "transformer.h.18.ln_attn.bias": "model-00020-of-00081.safetensors",
88
- "transformer.h.18.ln_attn.weight": "model-00020-of-00081.safetensors",
89
- "transformer.h.18.ln_mlp.bias": "model-00020-of-00081.safetensors",
90
- "transformer.h.18.ln_mlp.weight": "model-00020-of-00081.safetensors",
91
- "transformer.h.18.mlp.dense_4h_to_h.weight": "model-00020-of-00081.safetensors",
92
- "transformer.h.18.mlp.dense_h_to_4h.weight": "model-00019-of-00081.safetensors",
93
- "transformer.h.18.self_attention.dense.weight": "model-00019-of-00081.safetensors",
94
- "transformer.h.18.self_attention.query_key_value.weight": "model-00019-of-00081.safetensors",
95
- "transformer.h.19.ln_attn.bias": "model-00021-of-00081.safetensors",
96
- "transformer.h.19.ln_attn.weight": "model-00021-of-00081.safetensors",
97
- "transformer.h.19.ln_mlp.bias": "model-00021-of-00081.safetensors",
98
- "transformer.h.19.ln_mlp.weight": "model-00021-of-00081.safetensors",
99
- "transformer.h.19.mlp.dense_4h_to_h.weight": "model-00021-of-00081.safetensors",
100
- "transformer.h.19.mlp.dense_h_to_4h.weight": "model-00020-of-00081.safetensors",
101
- "transformer.h.19.self_attention.dense.weight": "model-00020-of-00081.safetensors",
102
- "transformer.h.19.self_attention.query_key_value.weight": "model-00020-of-00081.safetensors",
103
- "transformer.h.2.ln_attn.bias": "model-00004-of-00081.safetensors",
104
- "transformer.h.2.ln_attn.weight": "model-00004-of-00081.safetensors",
105
- "transformer.h.2.ln_mlp.bias": "model-00004-of-00081.safetensors",
106
- "transformer.h.2.ln_mlp.weight": "model-00004-of-00081.safetensors",
107
- "transformer.h.2.mlp.dense_4h_to_h.weight": "model-00004-of-00081.safetensors",
108
- "transformer.h.2.mlp.dense_h_to_4h.weight": "model-00003-of-00081.safetensors",
109
- "transformer.h.2.self_attention.dense.weight": "model-00003-of-00081.safetensors",
110
- "transformer.h.2.self_attention.query_key_value.weight": "model-00003-of-00081.safetensors",
111
- "transformer.h.20.ln_attn.bias": "model-00022-of-00081.safetensors",
112
- "transformer.h.20.ln_attn.weight": "model-00022-of-00081.safetensors",
113
- "transformer.h.20.ln_mlp.bias": "model-00022-of-00081.safetensors",
114
- "transformer.h.20.ln_mlp.weight": "model-00022-of-00081.safetensors",
115
- "transformer.h.20.mlp.dense_4h_to_h.weight": "model-00022-of-00081.safetensors",
116
- "transformer.h.20.mlp.dense_h_to_4h.weight": "model-00021-of-00081.safetensors",
117
- "transformer.h.20.self_attention.dense.weight": "model-00021-of-00081.safetensors",
118
- "transformer.h.20.self_attention.query_key_value.weight": "model-00021-of-00081.safetensors",
119
- "transformer.h.21.ln_attn.bias": "model-00023-of-00081.safetensors",
120
- "transformer.h.21.ln_attn.weight": "model-00023-of-00081.safetensors",
121
- "transformer.h.21.ln_mlp.bias": "model-00023-of-00081.safetensors",
122
- "transformer.h.21.ln_mlp.weight": "model-00023-of-00081.safetensors",
123
- "transformer.h.21.mlp.dense_4h_to_h.weight": "model-00023-of-00081.safetensors",
124
- "transformer.h.21.mlp.dense_h_to_4h.weight": "model-00022-of-00081.safetensors",
125
- "transformer.h.21.self_attention.dense.weight": "model-00022-of-00081.safetensors",
126
- "transformer.h.21.self_attention.query_key_value.weight": "model-00022-of-00081.safetensors",
127
- "transformer.h.22.ln_attn.bias": "model-00024-of-00081.safetensors",
128
- "transformer.h.22.ln_attn.weight": "model-00024-of-00081.safetensors",
129
- "transformer.h.22.ln_mlp.bias": "model-00024-of-00081.safetensors",
130
- "transformer.h.22.ln_mlp.weight": "model-00024-of-00081.safetensors",
131
- "transformer.h.22.mlp.dense_4h_to_h.weight": "model-00024-of-00081.safetensors",
132
- "transformer.h.22.mlp.dense_h_to_4h.weight": "model-00023-of-00081.safetensors",
133
- "transformer.h.22.self_attention.dense.weight": "model-00023-of-00081.safetensors",
134
- "transformer.h.22.self_attention.query_key_value.weight": "model-00023-of-00081.safetensors",
135
- "transformer.h.23.ln_attn.bias": "model-00025-of-00081.safetensors",
136
- "transformer.h.23.ln_attn.weight": "model-00025-of-00081.safetensors",
137
- "transformer.h.23.ln_mlp.bias": "model-00025-of-00081.safetensors",
138
- "transformer.h.23.ln_mlp.weight": "model-00025-of-00081.safetensors",
139
- "transformer.h.23.mlp.dense_4h_to_h.weight": "model-00025-of-00081.safetensors",
140
- "transformer.h.23.mlp.dense_h_to_4h.weight": "model-00024-of-00081.safetensors",
141
- "transformer.h.23.self_attention.dense.weight": "model-00024-of-00081.safetensors",
142
- "transformer.h.23.self_attention.query_key_value.weight": "model-00024-of-00081.safetensors",
143
- "transformer.h.24.ln_attn.bias": "model-00026-of-00081.safetensors",
144
- "transformer.h.24.ln_attn.weight": "model-00026-of-00081.safetensors",
145
- "transformer.h.24.ln_mlp.bias": "model-00026-of-00081.safetensors",
146
- "transformer.h.24.ln_mlp.weight": "model-00026-of-00081.safetensors",
147
- "transformer.h.24.mlp.dense_4h_to_h.weight": "model-00026-of-00081.safetensors",
148
- "transformer.h.24.mlp.dense_h_to_4h.weight": "model-00025-of-00081.safetensors",
149
- "transformer.h.24.self_attention.dense.weight": "model-00025-of-00081.safetensors",
150
- "transformer.h.24.self_attention.query_key_value.weight": "model-00025-of-00081.safetensors",
151
- "transformer.h.25.ln_attn.bias": "model-00027-of-00081.safetensors",
152
- "transformer.h.25.ln_attn.weight": "model-00027-of-00081.safetensors",
153
- "transformer.h.25.ln_mlp.bias": "model-00027-of-00081.safetensors",
154
- "transformer.h.25.ln_mlp.weight": "model-00027-of-00081.safetensors",
155
- "transformer.h.25.mlp.dense_4h_to_h.weight": "model-00027-of-00081.safetensors",
156
- "transformer.h.25.mlp.dense_h_to_4h.weight": "model-00026-of-00081.safetensors",
157
- "transformer.h.25.self_attention.dense.weight": "model-00026-of-00081.safetensors",
158
- "transformer.h.25.self_attention.query_key_value.weight": "model-00026-of-00081.safetensors",
159
- "transformer.h.26.ln_attn.bias": "model-00028-of-00081.safetensors",
160
- "transformer.h.26.ln_attn.weight": "model-00028-of-00081.safetensors",
161
- "transformer.h.26.ln_mlp.bias": "model-00028-of-00081.safetensors",
162
- "transformer.h.26.ln_mlp.weight": "model-00028-of-00081.safetensors",
163
- "transformer.h.26.mlp.dense_4h_to_h.weight": "model-00028-of-00081.safetensors",
164
- "transformer.h.26.mlp.dense_h_to_4h.weight": "model-00027-of-00081.safetensors",
165
- "transformer.h.26.self_attention.dense.weight": "model-00027-of-00081.safetensors",
166
- "transformer.h.26.self_attention.query_key_value.weight": "model-00027-of-00081.safetensors",
167
- "transformer.h.27.ln_attn.bias": "model-00029-of-00081.safetensors",
168
- "transformer.h.27.ln_attn.weight": "model-00029-of-00081.safetensors",
169
- "transformer.h.27.ln_mlp.bias": "model-00029-of-00081.safetensors",
170
- "transformer.h.27.ln_mlp.weight": "model-00029-of-00081.safetensors",
171
- "transformer.h.27.mlp.dense_4h_to_h.weight": "model-00029-of-00081.safetensors",
172
- "transformer.h.27.mlp.dense_h_to_4h.weight": "model-00028-of-00081.safetensors",
173
- "transformer.h.27.self_attention.dense.weight": "model-00028-of-00081.safetensors",
174
- "transformer.h.27.self_attention.query_key_value.weight": "model-00028-of-00081.safetensors",
175
- "transformer.h.28.ln_attn.bias": "model-00030-of-00081.safetensors",
176
- "transformer.h.28.ln_attn.weight": "model-00030-of-00081.safetensors",
177
- "transformer.h.28.ln_mlp.bias": "model-00030-of-00081.safetensors",
178
- "transformer.h.28.ln_mlp.weight": "model-00030-of-00081.safetensors",
179
- "transformer.h.28.mlp.dense_4h_to_h.weight": "model-00030-of-00081.safetensors",
180
- "transformer.h.28.mlp.dense_h_to_4h.weight": "model-00029-of-00081.safetensors",
181
- "transformer.h.28.self_attention.dense.weight": "model-00029-of-00081.safetensors",
182
- "transformer.h.28.self_attention.query_key_value.weight": "model-00029-of-00081.safetensors",
183
- "transformer.h.29.ln_attn.bias": "model-00031-of-00081.safetensors",
184
- "transformer.h.29.ln_attn.weight": "model-00031-of-00081.safetensors",
185
- "transformer.h.29.ln_mlp.bias": "model-00031-of-00081.safetensors",
186
- "transformer.h.29.ln_mlp.weight": "model-00031-of-00081.safetensors",
187
- "transformer.h.29.mlp.dense_4h_to_h.weight": "model-00031-of-00081.safetensors",
188
- "transformer.h.29.mlp.dense_h_to_4h.weight": "model-00030-of-00081.safetensors",
189
- "transformer.h.29.self_attention.dense.weight": "model-00030-of-00081.safetensors",
190
- "transformer.h.29.self_attention.query_key_value.weight": "model-00030-of-00081.safetensors",
191
- "transformer.h.3.ln_attn.bias": "model-00005-of-00081.safetensors",
192
- "transformer.h.3.ln_attn.weight": "model-00005-of-00081.safetensors",
193
- "transformer.h.3.ln_mlp.bias": "model-00005-of-00081.safetensors",
194
- "transformer.h.3.ln_mlp.weight": "model-00005-of-00081.safetensors",
195
- "transformer.h.3.mlp.dense_4h_to_h.weight": "model-00005-of-00081.safetensors",
196
- "transformer.h.3.mlp.dense_h_to_4h.weight": "model-00004-of-00081.safetensors",
197
- "transformer.h.3.self_attention.dense.weight": "model-00004-of-00081.safetensors",
198
- "transformer.h.3.self_attention.query_key_value.weight": "model-00004-of-00081.safetensors",
199
- "transformer.h.30.ln_attn.bias": "model-00032-of-00081.safetensors",
200
- "transformer.h.30.ln_attn.weight": "model-00032-of-00081.safetensors",
201
- "transformer.h.30.ln_mlp.bias": "model-00032-of-00081.safetensors",
202
- "transformer.h.30.ln_mlp.weight": "model-00032-of-00081.safetensors",
203
- "transformer.h.30.mlp.dense_4h_to_h.weight": "model-00032-of-00081.safetensors",
204
- "transformer.h.30.mlp.dense_h_to_4h.weight": "model-00031-of-00081.safetensors",
205
- "transformer.h.30.self_attention.dense.weight": "model-00031-of-00081.safetensors",
206
- "transformer.h.30.self_attention.query_key_value.weight": "model-00031-of-00081.safetensors",
207
- "transformer.h.31.ln_attn.bias": "model-00033-of-00081.safetensors",
208
- "transformer.h.31.ln_attn.weight": "model-00033-of-00081.safetensors",
209
- "transformer.h.31.ln_mlp.bias": "model-00033-of-00081.safetensors",
210
- "transformer.h.31.ln_mlp.weight": "model-00033-of-00081.safetensors",
211
- "transformer.h.31.mlp.dense_4h_to_h.weight": "model-00033-of-00081.safetensors",
212
- "transformer.h.31.mlp.dense_h_to_4h.weight": "model-00032-of-00081.safetensors",
213
- "transformer.h.31.self_attention.dense.weight": "model-00032-of-00081.safetensors",
214
- "transformer.h.31.self_attention.query_key_value.weight": "model-00032-of-00081.safetensors",
215
- "transformer.h.32.ln_attn.bias": "model-00034-of-00081.safetensors",
216
- "transformer.h.32.ln_attn.weight": "model-00034-of-00081.safetensors",
217
- "transformer.h.32.ln_mlp.bias": "model-00034-of-00081.safetensors",
218
- "transformer.h.32.ln_mlp.weight": "model-00034-of-00081.safetensors",
219
- "transformer.h.32.mlp.dense_4h_to_h.weight": "model-00034-of-00081.safetensors",
220
- "transformer.h.32.mlp.dense_h_to_4h.weight": "model-00033-of-00081.safetensors",
221
- "transformer.h.32.self_attention.dense.weight": "model-00033-of-00081.safetensors",
222
- "transformer.h.32.self_attention.query_key_value.weight": "model-00033-of-00081.safetensors",
223
- "transformer.h.33.ln_attn.bias": "model-00035-of-00081.safetensors",
224
- "transformer.h.33.ln_attn.weight": "model-00035-of-00081.safetensors",
225
- "transformer.h.33.ln_mlp.bias": "model-00035-of-00081.safetensors",
226
- "transformer.h.33.ln_mlp.weight": "model-00035-of-00081.safetensors",
227
- "transformer.h.33.mlp.dense_4h_to_h.weight": "model-00035-of-00081.safetensors",
228
- "transformer.h.33.mlp.dense_h_to_4h.weight": "model-00034-of-00081.safetensors",
229
- "transformer.h.33.self_attention.dense.weight": "model-00034-of-00081.safetensors",
230
- "transformer.h.33.self_attention.query_key_value.weight": "model-00034-of-00081.safetensors",
231
- "transformer.h.34.ln_attn.bias": "model-00036-of-00081.safetensors",
232
- "transformer.h.34.ln_attn.weight": "model-00036-of-00081.safetensors",
233
- "transformer.h.34.ln_mlp.bias": "model-00036-of-00081.safetensors",
234
- "transformer.h.34.ln_mlp.weight": "model-00036-of-00081.safetensors",
235
- "transformer.h.34.mlp.dense_4h_to_h.weight": "model-00036-of-00081.safetensors",
236
- "transformer.h.34.mlp.dense_h_to_4h.weight": "model-00035-of-00081.safetensors",
237
- "transformer.h.34.self_attention.dense.weight": "model-00035-of-00081.safetensors",
238
- "transformer.h.34.self_attention.query_key_value.weight": "model-00035-of-00081.safetensors",
239
- "transformer.h.35.ln_attn.bias": "model-00037-of-00081.safetensors",
240
- "transformer.h.35.ln_attn.weight": "model-00037-of-00081.safetensors",
241
- "transformer.h.35.ln_mlp.bias": "model-00037-of-00081.safetensors",
242
- "transformer.h.35.ln_mlp.weight": "model-00037-of-00081.safetensors",
243
- "transformer.h.35.mlp.dense_4h_to_h.weight": "model-00037-of-00081.safetensors",
244
- "transformer.h.35.mlp.dense_h_to_4h.weight": "model-00036-of-00081.safetensors",
245
- "transformer.h.35.self_attention.dense.weight": "model-00036-of-00081.safetensors",
246
- "transformer.h.35.self_attention.query_key_value.weight": "model-00036-of-00081.safetensors",
247
- "transformer.h.36.ln_attn.bias": "model-00038-of-00081.safetensors",
248
- "transformer.h.36.ln_attn.weight": "model-00038-of-00081.safetensors",
249
- "transformer.h.36.ln_mlp.bias": "model-00038-of-00081.safetensors",
250
- "transformer.h.36.ln_mlp.weight": "model-00038-of-00081.safetensors",
251
- "transformer.h.36.mlp.dense_4h_to_h.weight": "model-00038-of-00081.safetensors",
252
- "transformer.h.36.mlp.dense_h_to_4h.weight": "model-00037-of-00081.safetensors",
253
- "transformer.h.36.self_attention.dense.weight": "model-00037-of-00081.safetensors",
254
- "transformer.h.36.self_attention.query_key_value.weight": "model-00037-of-00081.safetensors",
255
- "transformer.h.37.ln_attn.bias": "model-00039-of-00081.safetensors",
256
- "transformer.h.37.ln_attn.weight": "model-00039-of-00081.safetensors",
257
- "transformer.h.37.ln_mlp.bias": "model-00039-of-00081.safetensors",
258
- "transformer.h.37.ln_mlp.weight": "model-00039-of-00081.safetensors",
259
- "transformer.h.37.mlp.dense_4h_to_h.weight": "model-00039-of-00081.safetensors",
260
- "transformer.h.37.mlp.dense_h_to_4h.weight": "model-00038-of-00081.safetensors",
261
- "transformer.h.37.self_attention.dense.weight": "model-00038-of-00081.safetensors",
262
- "transformer.h.37.self_attention.query_key_value.weight": "model-00038-of-00081.safetensors",
263
- "transformer.h.38.ln_attn.bias": "model-00040-of-00081.safetensors",
264
- "transformer.h.38.ln_attn.weight": "model-00040-of-00081.safetensors",
265
- "transformer.h.38.ln_mlp.bias": "model-00040-of-00081.safetensors",
266
- "transformer.h.38.ln_mlp.weight": "model-00040-of-00081.safetensors",
267
- "transformer.h.38.mlp.dense_4h_to_h.weight": "model-00040-of-00081.safetensors",
268
- "transformer.h.38.mlp.dense_h_to_4h.weight": "model-00039-of-00081.safetensors",
269
- "transformer.h.38.self_attention.dense.weight": "model-00039-of-00081.safetensors",
270
- "transformer.h.38.self_attention.query_key_value.weight": "model-00039-of-00081.safetensors",
271
- "transformer.h.39.ln_attn.bias": "model-00041-of-00081.safetensors",
272
- "transformer.h.39.ln_attn.weight": "model-00041-of-00081.safetensors",
273
- "transformer.h.39.ln_mlp.bias": "model-00041-of-00081.safetensors",
274
- "transformer.h.39.ln_mlp.weight": "model-00041-of-00081.safetensors",
275
- "transformer.h.39.mlp.dense_4h_to_h.weight": "model-00041-of-00081.safetensors",
276
- "transformer.h.39.mlp.dense_h_to_4h.weight": "model-00040-of-00081.safetensors",
277
- "transformer.h.39.self_attention.dense.weight": "model-00040-of-00081.safetensors",
278
- "transformer.h.39.self_attention.query_key_value.weight": "model-00040-of-00081.safetensors",
279
- "transformer.h.4.ln_attn.bias": "model-00006-of-00081.safetensors",
280
- "transformer.h.4.ln_attn.weight": "model-00006-of-00081.safetensors",
281
- "transformer.h.4.ln_mlp.bias": "model-00006-of-00081.safetensors",
282
- "transformer.h.4.ln_mlp.weight": "model-00006-of-00081.safetensors",
283
- "transformer.h.4.mlp.dense_4h_to_h.weight": "model-00006-of-00081.safetensors",
284
- "transformer.h.4.mlp.dense_h_to_4h.weight": "model-00005-of-00081.safetensors",
285
- "transformer.h.4.self_attention.dense.weight": "model-00005-of-00081.safetensors",
286
- "transformer.h.4.self_attention.query_key_value.weight": "model-00005-of-00081.safetensors",
287
- "transformer.h.40.ln_attn.bias": "model-00042-of-00081.safetensors",
288
- "transformer.h.40.ln_attn.weight": "model-00042-of-00081.safetensors",
289
- "transformer.h.40.ln_mlp.bias": "model-00042-of-00081.safetensors",
290
- "transformer.h.40.ln_mlp.weight": "model-00042-of-00081.safetensors",
291
- "transformer.h.40.mlp.dense_4h_to_h.weight": "model-00042-of-00081.safetensors",
292
- "transformer.h.40.mlp.dense_h_to_4h.weight": "model-00041-of-00081.safetensors",
293
- "transformer.h.40.self_attention.dense.weight": "model-00041-of-00081.safetensors",
294
- "transformer.h.40.self_attention.query_key_value.weight": "model-00041-of-00081.safetensors",
295
- "transformer.h.41.ln_attn.bias": "model-00043-of-00081.safetensors",
296
- "transformer.h.41.ln_attn.weight": "model-00043-of-00081.safetensors",
297
- "transformer.h.41.ln_mlp.bias": "model-00043-of-00081.safetensors",
298
- "transformer.h.41.ln_mlp.weight": "model-00043-of-00081.safetensors",
299
- "transformer.h.41.mlp.dense_4h_to_h.weight": "model-00043-of-00081.safetensors",
300
- "transformer.h.41.mlp.dense_h_to_4h.weight": "model-00042-of-00081.safetensors",
301
- "transformer.h.41.self_attention.dense.weight": "model-00042-of-00081.safetensors",
302
- "transformer.h.41.self_attention.query_key_value.weight": "model-00042-of-00081.safetensors",
303
- "transformer.h.42.ln_attn.bias": "model-00044-of-00081.safetensors",
304
- "transformer.h.42.ln_attn.weight": "model-00044-of-00081.safetensors",
305
- "transformer.h.42.ln_mlp.bias": "model-00044-of-00081.safetensors",
306
- "transformer.h.42.ln_mlp.weight": "model-00044-of-00081.safetensors",
307
- "transformer.h.42.mlp.dense_4h_to_h.weight": "model-00044-of-00081.safetensors",
308
- "transformer.h.42.mlp.dense_h_to_4h.weight": "model-00043-of-00081.safetensors",
309
- "transformer.h.42.self_attention.dense.weight": "model-00043-of-00081.safetensors",
310
- "transformer.h.42.self_attention.query_key_value.weight": "model-00043-of-00081.safetensors",
311
- "transformer.h.43.ln_attn.bias": "model-00045-of-00081.safetensors",
312
- "transformer.h.43.ln_attn.weight": "model-00045-of-00081.safetensors",
313
- "transformer.h.43.ln_mlp.bias": "model-00045-of-00081.safetensors",
314
- "transformer.h.43.ln_mlp.weight": "model-00045-of-00081.safetensors",
315
- "transformer.h.43.mlp.dense_4h_to_h.weight": "model-00045-of-00081.safetensors",
316
- "transformer.h.43.mlp.dense_h_to_4h.weight": "model-00044-of-00081.safetensors",
317
- "transformer.h.43.self_attention.dense.weight": "model-00044-of-00081.safetensors",
318
- "transformer.h.43.self_attention.query_key_value.weight": "model-00044-of-00081.safetensors",
319
- "transformer.h.44.ln_attn.bias": "model-00046-of-00081.safetensors",
320
- "transformer.h.44.ln_attn.weight": "model-00046-of-00081.safetensors",
321
- "transformer.h.44.ln_mlp.bias": "model-00046-of-00081.safetensors",
322
- "transformer.h.44.ln_mlp.weight": "model-00046-of-00081.safetensors",
323
- "transformer.h.44.mlp.dense_4h_to_h.weight": "model-00046-of-00081.safetensors",
324
- "transformer.h.44.mlp.dense_h_to_4h.weight": "model-00045-of-00081.safetensors",
325
- "transformer.h.44.self_attention.dense.weight": "model-00045-of-00081.safetensors",
326
- "transformer.h.44.self_attention.query_key_value.weight": "model-00045-of-00081.safetensors",
327
- "transformer.h.45.ln_attn.bias": "model-00047-of-00081.safetensors",
328
- "transformer.h.45.ln_attn.weight": "model-00047-of-00081.safetensors",
329
- "transformer.h.45.ln_mlp.bias": "model-00047-of-00081.safetensors",
330
- "transformer.h.45.ln_mlp.weight": "model-00047-of-00081.safetensors",
331
- "transformer.h.45.mlp.dense_4h_to_h.weight": "model-00047-of-00081.safetensors",
332
- "transformer.h.45.mlp.dense_h_to_4h.weight": "model-00046-of-00081.safetensors",
333
- "transformer.h.45.self_attention.dense.weight": "model-00046-of-00081.safetensors",
334
- "transformer.h.45.self_attention.query_key_value.weight": "model-00046-of-00081.safetensors",
335
- "transformer.h.46.ln_attn.bias": "model-00048-of-00081.safetensors",
336
- "transformer.h.46.ln_attn.weight": "model-00048-of-00081.safetensors",
337
- "transformer.h.46.ln_mlp.bias": "model-00048-of-00081.safetensors",
338
- "transformer.h.46.ln_mlp.weight": "model-00048-of-00081.safetensors",
339
- "transformer.h.46.mlp.dense_4h_to_h.weight": "model-00048-of-00081.safetensors",
340
- "transformer.h.46.mlp.dense_h_to_4h.weight": "model-00047-of-00081.safetensors",
341
- "transformer.h.46.self_attention.dense.weight": "model-00047-of-00081.safetensors",
342
- "transformer.h.46.self_attention.query_key_value.weight": "model-00047-of-00081.safetensors",
343
- "transformer.h.47.ln_attn.bias": "model-00049-of-00081.safetensors",
344
- "transformer.h.47.ln_attn.weight": "model-00049-of-00081.safetensors",
345
- "transformer.h.47.ln_mlp.bias": "model-00049-of-00081.safetensors",
346
- "transformer.h.47.ln_mlp.weight": "model-00049-of-00081.safetensors",
347
- "transformer.h.47.mlp.dense_4h_to_h.weight": "model-00049-of-00081.safetensors",
348
- "transformer.h.47.mlp.dense_h_to_4h.weight": "model-00048-of-00081.safetensors",
349
- "transformer.h.47.self_attention.dense.weight": "model-00048-of-00081.safetensors",
350
- "transformer.h.47.self_attention.query_key_value.weight": "model-00048-of-00081.safetensors",
351
- "transformer.h.48.ln_attn.bias": "model-00050-of-00081.safetensors",
352
- "transformer.h.48.ln_attn.weight": "model-00050-of-00081.safetensors",
353
- "transformer.h.48.ln_mlp.bias": "model-00050-of-00081.safetensors",
354
- "transformer.h.48.ln_mlp.weight": "model-00050-of-00081.safetensors",
355
- "transformer.h.48.mlp.dense_4h_to_h.weight": "model-00050-of-00081.safetensors",
356
- "transformer.h.48.mlp.dense_h_to_4h.weight": "model-00049-of-00081.safetensors",
357
- "transformer.h.48.self_attention.dense.weight": "model-00049-of-00081.safetensors",
358
- "transformer.h.48.self_attention.query_key_value.weight": "model-00049-of-00081.safetensors",
359
- "transformer.h.49.ln_attn.bias": "model-00051-of-00081.safetensors",
360
- "transformer.h.49.ln_attn.weight": "model-00051-of-00081.safetensors",
361
- "transformer.h.49.ln_mlp.bias": "model-00051-of-00081.safetensors",
362
- "transformer.h.49.ln_mlp.weight": "model-00051-of-00081.safetensors",
363
- "transformer.h.49.mlp.dense_4h_to_h.weight": "model-00051-of-00081.safetensors",
364
- "transformer.h.49.mlp.dense_h_to_4h.weight": "model-00050-of-00081.safetensors",
365
- "transformer.h.49.self_attention.dense.weight": "model-00050-of-00081.safetensors",
366
- "transformer.h.49.self_attention.query_key_value.weight": "model-00050-of-00081.safetensors",
367
- "transformer.h.5.ln_attn.bias": "model-00007-of-00081.safetensors",
368
- "transformer.h.5.ln_attn.weight": "model-00007-of-00081.safetensors",
369
- "transformer.h.5.ln_mlp.bias": "model-00007-of-00081.safetensors",
370
- "transformer.h.5.ln_mlp.weight": "model-00007-of-00081.safetensors",
371
- "transformer.h.5.mlp.dense_4h_to_h.weight": "model-00007-of-00081.safetensors",
372
- "transformer.h.5.mlp.dense_h_to_4h.weight": "model-00006-of-00081.safetensors",
373
- "transformer.h.5.self_attention.dense.weight": "model-00006-of-00081.safetensors",
374
- "transformer.h.5.self_attention.query_key_value.weight": "model-00006-of-00081.safetensors",
375
- "transformer.h.50.ln_attn.bias": "model-00052-of-00081.safetensors",
376
- "transformer.h.50.ln_attn.weight": "model-00052-of-00081.safetensors",
377
- "transformer.h.50.ln_mlp.bias": "model-00052-of-00081.safetensors",
378
- "transformer.h.50.ln_mlp.weight": "model-00052-of-00081.safetensors",
379
- "transformer.h.50.mlp.dense_4h_to_h.weight": "model-00052-of-00081.safetensors",
380
- "transformer.h.50.mlp.dense_h_to_4h.weight": "model-00051-of-00081.safetensors",
381
- "transformer.h.50.self_attention.dense.weight": "model-00051-of-00081.safetensors",
382
- "transformer.h.50.self_attention.query_key_value.weight": "model-00051-of-00081.safetensors",
383
- "transformer.h.51.ln_attn.bias": "model-00053-of-00081.safetensors",
384
- "transformer.h.51.ln_attn.weight": "model-00053-of-00081.safetensors",
385
- "transformer.h.51.ln_mlp.bias": "model-00053-of-00081.safetensors",
386
- "transformer.h.51.ln_mlp.weight": "model-00053-of-00081.safetensors",
387
- "transformer.h.51.mlp.dense_4h_to_h.weight": "model-00053-of-00081.safetensors",
388
- "transformer.h.51.mlp.dense_h_to_4h.weight": "model-00052-of-00081.safetensors",
389
- "transformer.h.51.self_attention.dense.weight": "model-00052-of-00081.safetensors",
390
- "transformer.h.51.self_attention.query_key_value.weight": "model-00052-of-00081.safetensors",
391
- "transformer.h.52.ln_attn.bias": "model-00054-of-00081.safetensors",
392
- "transformer.h.52.ln_attn.weight": "model-00054-of-00081.safetensors",
393
- "transformer.h.52.ln_mlp.bias": "model-00054-of-00081.safetensors",
394
- "transformer.h.52.ln_mlp.weight": "model-00054-of-00081.safetensors",
395
- "transformer.h.52.mlp.dense_4h_to_h.weight": "model-00054-of-00081.safetensors",
396
- "transformer.h.52.mlp.dense_h_to_4h.weight": "model-00053-of-00081.safetensors",
397
- "transformer.h.52.self_attention.dense.weight": "model-00053-of-00081.safetensors",
398
- "transformer.h.52.self_attention.query_key_value.weight": "model-00053-of-00081.safetensors",
399
- "transformer.h.53.ln_attn.bias": "model-00055-of-00081.safetensors",
400
- "transformer.h.53.ln_attn.weight": "model-00055-of-00081.safetensors",
401
- "transformer.h.53.ln_mlp.bias": "model-00055-of-00081.safetensors",
402
- "transformer.h.53.ln_mlp.weight": "model-00055-of-00081.safetensors",
403
- "transformer.h.53.mlp.dense_4h_to_h.weight": "model-00055-of-00081.safetensors",
404
- "transformer.h.53.mlp.dense_h_to_4h.weight": "model-00054-of-00081.safetensors",
405
- "transformer.h.53.self_attention.dense.weight": "model-00054-of-00081.safetensors",
406
- "transformer.h.53.self_attention.query_key_value.weight": "model-00054-of-00081.safetensors",
407
- "transformer.h.54.ln_attn.bias": "model-00056-of-00081.safetensors",
408
- "transformer.h.54.ln_attn.weight": "model-00056-of-00081.safetensors",
409
- "transformer.h.54.ln_mlp.bias": "model-00056-of-00081.safetensors",
410
- "transformer.h.54.ln_mlp.weight": "model-00056-of-00081.safetensors",
411
- "transformer.h.54.mlp.dense_4h_to_h.weight": "model-00056-of-00081.safetensors",
412
- "transformer.h.54.mlp.dense_h_to_4h.weight": "model-00055-of-00081.safetensors",
413
- "transformer.h.54.self_attention.dense.weight": "model-00055-of-00081.safetensors",
414
- "transformer.h.54.self_attention.query_key_value.weight": "model-00055-of-00081.safetensors",
415
- "transformer.h.55.ln_attn.bias": "model-00057-of-00081.safetensors",
416
- "transformer.h.55.ln_attn.weight": "model-00057-of-00081.safetensors",
417
- "transformer.h.55.ln_mlp.bias": "model-00057-of-00081.safetensors",
418
- "transformer.h.55.ln_mlp.weight": "model-00057-of-00081.safetensors",
419
- "transformer.h.55.mlp.dense_4h_to_h.weight": "model-00057-of-00081.safetensors",
420
- "transformer.h.55.mlp.dense_h_to_4h.weight": "model-00056-of-00081.safetensors",
421
- "transformer.h.55.self_attention.dense.weight": "model-00056-of-00081.safetensors",
422
- "transformer.h.55.self_attention.query_key_value.weight": "model-00056-of-00081.safetensors",
423
- "transformer.h.56.ln_attn.bias": "model-00058-of-00081.safetensors",
424
- "transformer.h.56.ln_attn.weight": "model-00058-of-00081.safetensors",
425
- "transformer.h.56.ln_mlp.bias": "model-00058-of-00081.safetensors",
426
- "transformer.h.56.ln_mlp.weight": "model-00058-of-00081.safetensors",
427
- "transformer.h.56.mlp.dense_4h_to_h.weight": "model-00058-of-00081.safetensors",
428
- "transformer.h.56.mlp.dense_h_to_4h.weight": "model-00057-of-00081.safetensors",
429
- "transformer.h.56.self_attention.dense.weight": "model-00057-of-00081.safetensors",
430
- "transformer.h.56.self_attention.query_key_value.weight": "model-00057-of-00081.safetensors",
431
- "transformer.h.57.ln_attn.bias": "model-00059-of-00081.safetensors",
432
- "transformer.h.57.ln_attn.weight": "model-00059-of-00081.safetensors",
433
- "transformer.h.57.ln_mlp.bias": "model-00059-of-00081.safetensors",
434
- "transformer.h.57.ln_mlp.weight": "model-00059-of-00081.safetensors",
435
- "transformer.h.57.mlp.dense_4h_to_h.weight": "model-00059-of-00081.safetensors",
436
- "transformer.h.57.mlp.dense_h_to_4h.weight": "model-00058-of-00081.safetensors",
437
- "transformer.h.57.self_attention.dense.weight": "model-00058-of-00081.safetensors",
438
- "transformer.h.57.self_attention.query_key_value.weight": "model-00058-of-00081.safetensors",
439
- "transformer.h.58.ln_attn.bias": "model-00060-of-00081.safetensors",
440
- "transformer.h.58.ln_attn.weight": "model-00060-of-00081.safetensors",
441
- "transformer.h.58.ln_mlp.bias": "model-00060-of-00081.safetensors",
442
- "transformer.h.58.ln_mlp.weight": "model-00060-of-00081.safetensors",
443
- "transformer.h.58.mlp.dense_4h_to_h.weight": "model-00060-of-00081.safetensors",
444
- "transformer.h.58.mlp.dense_h_to_4h.weight": "model-00059-of-00081.safetensors",
445
- "transformer.h.58.self_attention.dense.weight": "model-00059-of-00081.safetensors",
446
- "transformer.h.58.self_attention.query_key_value.weight": "model-00059-of-00081.safetensors",
447
- "transformer.h.59.ln_attn.bias": "model-00061-of-00081.safetensors",
448
- "transformer.h.59.ln_attn.weight": "model-00061-of-00081.safetensors",
449
- "transformer.h.59.ln_mlp.bias": "model-00061-of-00081.safetensors",
450
- "transformer.h.59.ln_mlp.weight": "model-00061-of-00081.safetensors",
451
- "transformer.h.59.mlp.dense_4h_to_h.weight": "model-00061-of-00081.safetensors",
452
- "transformer.h.59.mlp.dense_h_to_4h.weight": "model-00060-of-00081.safetensors",
453
- "transformer.h.59.self_attention.dense.weight": "model-00060-of-00081.safetensors",
454
- "transformer.h.59.self_attention.query_key_value.weight": "model-00060-of-00081.safetensors",
455
- "transformer.h.6.ln_attn.bias": "model-00008-of-00081.safetensors",
456
- "transformer.h.6.ln_attn.weight": "model-00008-of-00081.safetensors",
457
- "transformer.h.6.ln_mlp.bias": "model-00008-of-00081.safetensors",
458
- "transformer.h.6.ln_mlp.weight": "model-00008-of-00081.safetensors",
459
- "transformer.h.6.mlp.dense_4h_to_h.weight": "model-00008-of-00081.safetensors",
460
- "transformer.h.6.mlp.dense_h_to_4h.weight": "model-00007-of-00081.safetensors",
461
- "transformer.h.6.self_attention.dense.weight": "model-00007-of-00081.safetensors",
462
- "transformer.h.6.self_attention.query_key_value.weight": "model-00007-of-00081.safetensors",
463
- "transformer.h.60.ln_attn.bias": "model-00062-of-00081.safetensors",
464
- "transformer.h.60.ln_attn.weight": "model-00062-of-00081.safetensors",
465
- "transformer.h.60.ln_mlp.bias": "model-00062-of-00081.safetensors",
466
- "transformer.h.60.ln_mlp.weight": "model-00062-of-00081.safetensors",
467
- "transformer.h.60.mlp.dense_4h_to_h.weight": "model-00062-of-00081.safetensors",
468
- "transformer.h.60.mlp.dense_h_to_4h.weight": "model-00061-of-00081.safetensors",
469
- "transformer.h.60.self_attention.dense.weight": "model-00061-of-00081.safetensors",
470
- "transformer.h.60.self_attention.query_key_value.weight": "model-00061-of-00081.safetensors",
471
- "transformer.h.61.ln_attn.bias": "model-00063-of-00081.safetensors",
472
- "transformer.h.61.ln_attn.weight": "model-00063-of-00081.safetensors",
473
- "transformer.h.61.ln_mlp.bias": "model-00063-of-00081.safetensors",
474
- "transformer.h.61.ln_mlp.weight": "model-00063-of-00081.safetensors",
475
- "transformer.h.61.mlp.dense_4h_to_h.weight": "model-00063-of-00081.safetensors",
476
- "transformer.h.61.mlp.dense_h_to_4h.weight": "model-00062-of-00081.safetensors",
477
- "transformer.h.61.self_attention.dense.weight": "model-00062-of-00081.safetensors",
478
- "transformer.h.61.self_attention.query_key_value.weight": "model-00062-of-00081.safetensors",
479
- "transformer.h.62.ln_attn.bias": "model-00064-of-00081.safetensors",
480
- "transformer.h.62.ln_attn.weight": "model-00064-of-00081.safetensors",
481
- "transformer.h.62.ln_mlp.bias": "model-00064-of-00081.safetensors",
482
- "transformer.h.62.ln_mlp.weight": "model-00064-of-00081.safetensors",
483
- "transformer.h.62.mlp.dense_4h_to_h.weight": "model-00064-of-00081.safetensors",
484
- "transformer.h.62.mlp.dense_h_to_4h.weight": "model-00063-of-00081.safetensors",
485
- "transformer.h.62.self_attention.dense.weight": "model-00063-of-00081.safetensors",
486
- "transformer.h.62.self_attention.query_key_value.weight": "model-00063-of-00081.safetensors",
487
- "transformer.h.63.ln_attn.bias": "model-00065-of-00081.safetensors",
488
- "transformer.h.63.ln_attn.weight": "model-00065-of-00081.safetensors",
489
- "transformer.h.63.ln_mlp.bias": "model-00065-of-00081.safetensors",
490
- "transformer.h.63.ln_mlp.weight": "model-00065-of-00081.safetensors",
491
- "transformer.h.63.mlp.dense_4h_to_h.weight": "model-00065-of-00081.safetensors",
492
- "transformer.h.63.mlp.dense_h_to_4h.weight": "model-00064-of-00081.safetensors",
493
- "transformer.h.63.self_attention.dense.weight": "model-00064-of-00081.safetensors",
494
- "transformer.h.63.self_attention.query_key_value.weight": "model-00064-of-00081.safetensors",
495
- "transformer.h.64.ln_attn.bias": "model-00066-of-00081.safetensors",
496
- "transformer.h.64.ln_attn.weight": "model-00066-of-00081.safetensors",
497
- "transformer.h.64.ln_mlp.bias": "model-00066-of-00081.safetensors",
498
- "transformer.h.64.ln_mlp.weight": "model-00066-of-00081.safetensors",
499
- "transformer.h.64.mlp.dense_4h_to_h.weight": "model-00066-of-00081.safetensors",
500
- "transformer.h.64.mlp.dense_h_to_4h.weight": "model-00065-of-00081.safetensors",
501
- "transformer.h.64.self_attention.dense.weight": "model-00065-of-00081.safetensors",
502
- "transformer.h.64.self_attention.query_key_value.weight": "model-00065-of-00081.safetensors",
503
- "transformer.h.65.ln_attn.bias": "model-00067-of-00081.safetensors",
504
- "transformer.h.65.ln_attn.weight": "model-00067-of-00081.safetensors",
505
- "transformer.h.65.ln_mlp.bias": "model-00067-of-00081.safetensors",
506
- "transformer.h.65.ln_mlp.weight": "model-00067-of-00081.safetensors",
507
- "transformer.h.65.mlp.dense_4h_to_h.weight": "model-00067-of-00081.safetensors",
508
- "transformer.h.65.mlp.dense_h_to_4h.weight": "model-00066-of-00081.safetensors",
509
- "transformer.h.65.self_attention.dense.weight": "model-00066-of-00081.safetensors",
510
- "transformer.h.65.self_attention.query_key_value.weight": "model-00066-of-00081.safetensors",
511
- "transformer.h.66.ln_attn.bias": "model-00068-of-00081.safetensors",
512
- "transformer.h.66.ln_attn.weight": "model-00068-of-00081.safetensors",
513
- "transformer.h.66.ln_mlp.bias": "model-00068-of-00081.safetensors",
514
- "transformer.h.66.ln_mlp.weight": "model-00068-of-00081.safetensors",
515
- "transformer.h.66.mlp.dense_4h_to_h.weight": "model-00068-of-00081.safetensors",
516
- "transformer.h.66.mlp.dense_h_to_4h.weight": "model-00067-of-00081.safetensors",
517
- "transformer.h.66.self_attention.dense.weight": "model-00067-of-00081.safetensors",
518
- "transformer.h.66.self_attention.query_key_value.weight": "model-00067-of-00081.safetensors",
519
- "transformer.h.67.ln_attn.bias": "model-00069-of-00081.safetensors",
520
- "transformer.h.67.ln_attn.weight": "model-00069-of-00081.safetensors",
521
- "transformer.h.67.ln_mlp.bias": "model-00069-of-00081.safetensors",
522
- "transformer.h.67.ln_mlp.weight": "model-00069-of-00081.safetensors",
523
- "transformer.h.67.mlp.dense_4h_to_h.weight": "model-00069-of-00081.safetensors",
524
- "transformer.h.67.mlp.dense_h_to_4h.weight": "model-00068-of-00081.safetensors",
525
- "transformer.h.67.self_attention.dense.weight": "model-00068-of-00081.safetensors",
526
- "transformer.h.67.self_attention.query_key_value.weight": "model-00068-of-00081.safetensors",
527
- "transformer.h.68.ln_attn.bias": "model-00070-of-00081.safetensors",
528
- "transformer.h.68.ln_attn.weight": "model-00070-of-00081.safetensors",
529
- "transformer.h.68.ln_mlp.bias": "model-00070-of-00081.safetensors",
530
- "transformer.h.68.ln_mlp.weight": "model-00070-of-00081.safetensors",
531
- "transformer.h.68.mlp.dense_4h_to_h.weight": "model-00070-of-00081.safetensors",
532
- "transformer.h.68.mlp.dense_h_to_4h.weight": "model-00069-of-00081.safetensors",
533
- "transformer.h.68.self_attention.dense.weight": "model-00069-of-00081.safetensors",
534
- "transformer.h.68.self_attention.query_key_value.weight": "model-00069-of-00081.safetensors",
535
- "transformer.h.69.ln_attn.bias": "model-00071-of-00081.safetensors",
536
- "transformer.h.69.ln_attn.weight": "model-00071-of-00081.safetensors",
537
- "transformer.h.69.ln_mlp.bias": "model-00071-of-00081.safetensors",
538
- "transformer.h.69.ln_mlp.weight": "model-00071-of-00081.safetensors",
539
- "transformer.h.69.mlp.dense_4h_to_h.weight": "model-00071-of-00081.safetensors",
540
- "transformer.h.69.mlp.dense_h_to_4h.weight": "model-00070-of-00081.safetensors",
541
- "transformer.h.69.self_attention.dense.weight": "model-00070-of-00081.safetensors",
542
- "transformer.h.69.self_attention.query_key_value.weight": "model-00070-of-00081.safetensors",
543
- "transformer.h.7.ln_attn.bias": "model-00009-of-00081.safetensors",
544
- "transformer.h.7.ln_attn.weight": "model-00009-of-00081.safetensors",
545
- "transformer.h.7.ln_mlp.bias": "model-00009-of-00081.safetensors",
546
- "transformer.h.7.ln_mlp.weight": "model-00009-of-00081.safetensors",
547
- "transformer.h.7.mlp.dense_4h_to_h.weight": "model-00009-of-00081.safetensors",
548
- "transformer.h.7.mlp.dense_h_to_4h.weight": "model-00008-of-00081.safetensors",
549
- "transformer.h.7.self_attention.dense.weight": "model-00008-of-00081.safetensors",
550
- "transformer.h.7.self_attention.query_key_value.weight": "model-00008-of-00081.safetensors",
551
- "transformer.h.70.ln_attn.bias": "model-00072-of-00081.safetensors",
552
- "transformer.h.70.ln_attn.weight": "model-00072-of-00081.safetensors",
553
- "transformer.h.70.ln_mlp.bias": "model-00072-of-00081.safetensors",
554
- "transformer.h.70.ln_mlp.weight": "model-00072-of-00081.safetensors",
555
- "transformer.h.70.mlp.dense_4h_to_h.weight": "model-00072-of-00081.safetensors",
556
- "transformer.h.70.mlp.dense_h_to_4h.weight": "model-00071-of-00081.safetensors",
557
- "transformer.h.70.self_attention.dense.weight": "model-00071-of-00081.safetensors",
558
- "transformer.h.70.self_attention.query_key_value.weight": "model-00071-of-00081.safetensors",
559
- "transformer.h.71.ln_attn.bias": "model-00073-of-00081.safetensors",
560
- "transformer.h.71.ln_attn.weight": "model-00073-of-00081.safetensors",
561
- "transformer.h.71.ln_mlp.bias": "model-00073-of-00081.safetensors",
562
- "transformer.h.71.ln_mlp.weight": "model-00073-of-00081.safetensors",
563
- "transformer.h.71.mlp.dense_4h_to_h.weight": "model-00073-of-00081.safetensors",
564
- "transformer.h.71.mlp.dense_h_to_4h.weight": "model-00072-of-00081.safetensors",
565
- "transformer.h.71.self_attention.dense.weight": "model-00072-of-00081.safetensors",
566
- "transformer.h.71.self_attention.query_key_value.weight": "model-00072-of-00081.safetensors",
567
- "transformer.h.72.ln_attn.bias": "model-00074-of-00081.safetensors",
568
- "transformer.h.72.ln_attn.weight": "model-00074-of-00081.safetensors",
569
- "transformer.h.72.ln_mlp.bias": "model-00074-of-00081.safetensors",
570
- "transformer.h.72.ln_mlp.weight": "model-00074-of-00081.safetensors",
571
- "transformer.h.72.mlp.dense_4h_to_h.weight": "model-00074-of-00081.safetensors",
572
- "transformer.h.72.mlp.dense_h_to_4h.weight": "model-00073-of-00081.safetensors",
573
- "transformer.h.72.self_attention.dense.weight": "model-00073-of-00081.safetensors",
574
- "transformer.h.72.self_attention.query_key_value.weight": "model-00073-of-00081.safetensors",
575
- "transformer.h.73.ln_attn.bias": "model-00075-of-00081.safetensors",
576
- "transformer.h.73.ln_attn.weight": "model-00075-of-00081.safetensors",
577
- "transformer.h.73.ln_mlp.bias": "model-00075-of-00081.safetensors",
578
- "transformer.h.73.ln_mlp.weight": "model-00075-of-00081.safetensors",
579
- "transformer.h.73.mlp.dense_4h_to_h.weight": "model-00075-of-00081.safetensors",
580
- "transformer.h.73.mlp.dense_h_to_4h.weight": "model-00074-of-00081.safetensors",
581
- "transformer.h.73.self_attention.dense.weight": "model-00074-of-00081.safetensors",
582
- "transformer.h.73.self_attention.query_key_value.weight": "model-00074-of-00081.safetensors",
583
- "transformer.h.74.ln_attn.bias": "model-00076-of-00081.safetensors",
584
- "transformer.h.74.ln_attn.weight": "model-00076-of-00081.safetensors",
585
- "transformer.h.74.ln_mlp.bias": "model-00076-of-00081.safetensors",
586
- "transformer.h.74.ln_mlp.weight": "model-00076-of-00081.safetensors",
587
- "transformer.h.74.mlp.dense_4h_to_h.weight": "model-00076-of-00081.safetensors",
588
- "transformer.h.74.mlp.dense_h_to_4h.weight": "model-00075-of-00081.safetensors",
589
- "transformer.h.74.self_attention.dense.weight": "model-00075-of-00081.safetensors",
590
- "transformer.h.74.self_attention.query_key_value.weight": "model-00075-of-00081.safetensors",
591
- "transformer.h.75.ln_attn.bias": "model-00077-of-00081.safetensors",
592
- "transformer.h.75.ln_attn.weight": "model-00077-of-00081.safetensors",
593
- "transformer.h.75.ln_mlp.bias": "model-00077-of-00081.safetensors",
594
- "transformer.h.75.ln_mlp.weight": "model-00077-of-00081.safetensors",
595
- "transformer.h.75.mlp.dense_4h_to_h.weight": "model-00077-of-00081.safetensors",
596
- "transformer.h.75.mlp.dense_h_to_4h.weight": "model-00076-of-00081.safetensors",
597
- "transformer.h.75.self_attention.dense.weight": "model-00076-of-00081.safetensors",
598
- "transformer.h.75.self_attention.query_key_value.weight": "model-00076-of-00081.safetensors",
599
- "transformer.h.76.ln_attn.bias": "model-00078-of-00081.safetensors",
600
- "transformer.h.76.ln_attn.weight": "model-00078-of-00081.safetensors",
601
- "transformer.h.76.ln_mlp.bias": "model-00078-of-00081.safetensors",
602
- "transformer.h.76.ln_mlp.weight": "model-00078-of-00081.safetensors",
603
- "transformer.h.76.mlp.dense_4h_to_h.weight": "model-00078-of-00081.safetensors",
604
- "transformer.h.76.mlp.dense_h_to_4h.weight": "model-00077-of-00081.safetensors",
605
- "transformer.h.76.self_attention.dense.weight": "model-00077-of-00081.safetensors",
606
- "transformer.h.76.self_attention.query_key_value.weight": "model-00077-of-00081.safetensors",
607
- "transformer.h.77.ln_attn.bias": "model-00079-of-00081.safetensors",
608
- "transformer.h.77.ln_attn.weight": "model-00079-of-00081.safetensors",
609
- "transformer.h.77.ln_mlp.bias": "model-00079-of-00081.safetensors",
610
- "transformer.h.77.ln_mlp.weight": "model-00079-of-00081.safetensors",
611
- "transformer.h.77.mlp.dense_4h_to_h.weight": "model-00079-of-00081.safetensors",
612
- "transformer.h.77.mlp.dense_h_to_4h.weight": "model-00078-of-00081.safetensors",
613
- "transformer.h.77.self_attention.dense.weight": "model-00078-of-00081.safetensors",
614
- "transformer.h.77.self_attention.query_key_value.weight": "model-00078-of-00081.safetensors",
615
- "transformer.h.78.ln_attn.bias": "model-00080-of-00081.safetensors",
616
- "transformer.h.78.ln_attn.weight": "model-00080-of-00081.safetensors",
617
- "transformer.h.78.ln_mlp.bias": "model-00080-of-00081.safetensors",
618
- "transformer.h.78.ln_mlp.weight": "model-00080-of-00081.safetensors",
619
- "transformer.h.78.mlp.dense_4h_to_h.weight": "model-00080-of-00081.safetensors",
620
- "transformer.h.78.mlp.dense_h_to_4h.weight": "model-00079-of-00081.safetensors",
621
- "transformer.h.78.self_attention.dense.weight": "model-00079-of-00081.safetensors",
622
- "transformer.h.78.self_attention.query_key_value.weight": "model-00079-of-00081.safetensors",
623
- "transformer.h.79.ln_attn.bias": "model-00081-of-00081.safetensors",
624
- "transformer.h.79.ln_attn.weight": "model-00081-of-00081.safetensors",
625
- "transformer.h.79.ln_mlp.bias": "model-00081-of-00081.safetensors",
626
- "transformer.h.79.ln_mlp.weight": "model-00081-of-00081.safetensors",
627
- "transformer.h.79.mlp.dense_4h_to_h.weight": "model-00081-of-00081.safetensors",
628
- "transformer.h.79.mlp.dense_h_to_4h.weight": "model-00080-of-00081.safetensors",
629
- "transformer.h.79.self_attention.dense.weight": "model-00080-of-00081.safetensors",
630
- "transformer.h.79.self_attention.query_key_value.weight": "model-00080-of-00081.safetensors",
631
- "transformer.h.8.ln_attn.bias": "model-00010-of-00081.safetensors",
632
- "transformer.h.8.ln_attn.weight": "model-00010-of-00081.safetensors",
633
- "transformer.h.8.ln_mlp.bias": "model-00010-of-00081.safetensors",
634
- "transformer.h.8.ln_mlp.weight": "model-00010-of-00081.safetensors",
635
- "transformer.h.8.mlp.dense_4h_to_h.weight": "model-00010-of-00081.safetensors",
636
- "transformer.h.8.mlp.dense_h_to_4h.weight": "model-00009-of-00081.safetensors",
637
- "transformer.h.8.self_attention.dense.weight": "model-00009-of-00081.safetensors",
638
- "transformer.h.8.self_attention.query_key_value.weight": "model-00009-of-00081.safetensors",
639
- "transformer.h.9.ln_attn.bias": "model-00011-of-00081.safetensors",
640
- "transformer.h.9.ln_attn.weight": "model-00011-of-00081.safetensors",
641
- "transformer.h.9.ln_mlp.bias": "model-00011-of-00081.safetensors",
642
- "transformer.h.9.ln_mlp.weight": "model-00011-of-00081.safetensors",
643
- "transformer.h.9.mlp.dense_4h_to_h.weight": "model-00011-of-00081.safetensors",
644
- "transformer.h.9.mlp.dense_h_to_4h.weight": "model-00010-of-00081.safetensors",
645
- "transformer.h.9.self_attention.dense.weight": "model-00010-of-00081.safetensors",
646
- "transformer.h.9.self_attention.query_key_value.weight": "model-00010-of-00081.safetensors",
647
- "transformer.ln_f.bias": "model-00081-of-00081.safetensors",
648
- "transformer.ln_f.weight": "model-00081-of-00081.safetensors",
649
- "transformer.word_embeddings.weight": "model-00001-of-00081.safetensors"
650
- }
651
- }