BCenti commited on
Commit
42c626f
1 Parent(s): 9f26773

Upload 285 files

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
.gitattributes CHANGED
@@ -32,3 +32,173 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
32
  *.zip filter=lfs diff=lfs merge=lfs -text
33
  *.zst filter=lfs diff=lfs merge=lfs -text
34
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
32
  *.zip filter=lfs diff=lfs merge=lfs -text
33
  *.zst filter=lfs diff=lfs merge=lfs -text
34
  *tfevents* filter=lfs diff=lfs merge=lfs -text
35
+ h.0.attn.bias filter=lfs diff=lfs merge=lfs -text
36
+ onnx__MatMul_10003 filter=lfs diff=lfs merge=lfs -text
37
+ onnx__MatMul_10004 filter=lfs diff=lfs merge=lfs -text
38
+ onnx__MatMul_10005 filter=lfs diff=lfs merge=lfs -text
39
+ onnx__MatMul_10006 filter=lfs diff=lfs merge=lfs -text
40
+ onnx__MatMul_10007 filter=lfs diff=lfs merge=lfs -text
41
+ onnx__MatMul_10008 filter=lfs diff=lfs merge=lfs -text
42
+ onnx__MatMul_10033 filter=lfs diff=lfs merge=lfs -text
43
+ onnx__MatMul_10034 filter=lfs diff=lfs merge=lfs -text
44
+ onnx__MatMul_10035 filter=lfs diff=lfs merge=lfs -text
45
+ onnx__MatMul_10036 filter=lfs diff=lfs merge=lfs -text
46
+ onnx__MatMul_10037 filter=lfs diff=lfs merge=lfs -text
47
+ onnx__MatMul_10038 filter=lfs diff=lfs merge=lfs -text
48
+ onnx__MatMul_10063 filter=lfs diff=lfs merge=lfs -text
49
+ onnx__MatMul_10064 filter=lfs diff=lfs merge=lfs -text
50
+ onnx__MatMul_10065 filter=lfs diff=lfs merge=lfs -text
51
+ onnx__MatMul_10066 filter=lfs diff=lfs merge=lfs -text
52
+ onnx__MatMul_10067 filter=lfs diff=lfs merge=lfs -text
53
+ onnx__MatMul_10068 filter=lfs diff=lfs merge=lfs -text
54
+ onnx__MatMul_10093 filter=lfs diff=lfs merge=lfs -text
55
+ onnx__MatMul_10094 filter=lfs diff=lfs merge=lfs -text
56
+ onnx__MatMul_10095 filter=lfs diff=lfs merge=lfs -text
57
+ onnx__MatMul_10096 filter=lfs diff=lfs merge=lfs -text
58
+ onnx__MatMul_10097 filter=lfs diff=lfs merge=lfs -text
59
+ onnx__MatMul_10098 filter=lfs diff=lfs merge=lfs -text
60
+ onnx__MatMul_10123 filter=lfs diff=lfs merge=lfs -text
61
+ onnx__MatMul_10124 filter=lfs diff=lfs merge=lfs -text
62
+ onnx__MatMul_10125 filter=lfs diff=lfs merge=lfs -text
63
+ onnx__MatMul_10126 filter=lfs diff=lfs merge=lfs -text
64
+ onnx__MatMul_10127 filter=lfs diff=lfs merge=lfs -text
65
+ onnx__MatMul_10128 filter=lfs diff=lfs merge=lfs -text
66
+ onnx__MatMul_10153 filter=lfs diff=lfs merge=lfs -text
67
+ onnx__MatMul_10154 filter=lfs diff=lfs merge=lfs -text
68
+ onnx__MatMul_10155 filter=lfs diff=lfs merge=lfs -text
69
+ onnx__MatMul_10156 filter=lfs diff=lfs merge=lfs -text
70
+ onnx__MatMul_10157 filter=lfs diff=lfs merge=lfs -text
71
+ onnx__MatMul_10158 filter=lfs diff=lfs merge=lfs -text
72
+ onnx__MatMul_10183 filter=lfs diff=lfs merge=lfs -text
73
+ onnx__MatMul_10184 filter=lfs diff=lfs merge=lfs -text
74
+ onnx__MatMul_10185 filter=lfs diff=lfs merge=lfs -text
75
+ onnx__MatMul_10186 filter=lfs diff=lfs merge=lfs -text
76
+ onnx__MatMul_10187 filter=lfs diff=lfs merge=lfs -text
77
+ onnx__MatMul_10188 filter=lfs diff=lfs merge=lfs -text
78
+ onnx__MatMul_10213 filter=lfs diff=lfs merge=lfs -text
79
+ onnx__MatMul_10214 filter=lfs diff=lfs merge=lfs -text
80
+ onnx__MatMul_10215 filter=lfs diff=lfs merge=lfs -text
81
+ onnx__MatMul_10216 filter=lfs diff=lfs merge=lfs -text
82
+ onnx__MatMul_10217 filter=lfs diff=lfs merge=lfs -text
83
+ onnx__MatMul_10218 filter=lfs diff=lfs merge=lfs -text
84
+ onnx__MatMul_10243 filter=lfs diff=lfs merge=lfs -text
85
+ onnx__MatMul_10244 filter=lfs diff=lfs merge=lfs -text
86
+ onnx__MatMul_10245 filter=lfs diff=lfs merge=lfs -text
87
+ onnx__MatMul_10246 filter=lfs diff=lfs merge=lfs -text
88
+ onnx__MatMul_10247 filter=lfs diff=lfs merge=lfs -text
89
+ onnx__MatMul_10248 filter=lfs diff=lfs merge=lfs -text
90
+ onnx__MatMul_10273 filter=lfs diff=lfs merge=lfs -text
91
+ onnx__MatMul_10274 filter=lfs diff=lfs merge=lfs -text
92
+ onnx__MatMul_10275 filter=lfs diff=lfs merge=lfs -text
93
+ onnx__MatMul_10276 filter=lfs diff=lfs merge=lfs -text
94
+ onnx__MatMul_10277 filter=lfs diff=lfs merge=lfs -text
95
+ onnx__MatMul_10278 filter=lfs diff=lfs merge=lfs -text
96
+ onnx__MatMul_10303 filter=lfs diff=lfs merge=lfs -text
97
+ onnx__MatMul_10304 filter=lfs diff=lfs merge=lfs -text
98
+ onnx__MatMul_10305 filter=lfs diff=lfs merge=lfs -text
99
+ onnx__MatMul_10306 filter=lfs diff=lfs merge=lfs -text
100
+ onnx__MatMul_10307 filter=lfs diff=lfs merge=lfs -text
101
+ onnx__MatMul_10308 filter=lfs diff=lfs merge=lfs -text
102
+ onnx__MatMul_10333 filter=lfs diff=lfs merge=lfs -text
103
+ onnx__MatMul_10334 filter=lfs diff=lfs merge=lfs -text
104
+ onnx__MatMul_10335 filter=lfs diff=lfs merge=lfs -text
105
+ onnx__MatMul_10336 filter=lfs diff=lfs merge=lfs -text
106
+ onnx__MatMul_10337 filter=lfs diff=lfs merge=lfs -text
107
+ onnx__MatMul_10338 filter=lfs diff=lfs merge=lfs -text
108
+ onnx__MatMul_10363 filter=lfs diff=lfs merge=lfs -text
109
+ onnx__MatMul_10364 filter=lfs diff=lfs merge=lfs -text
110
+ onnx__MatMul_10365 filter=lfs diff=lfs merge=lfs -text
111
+ onnx__MatMul_10366 filter=lfs diff=lfs merge=lfs -text
112
+ onnx__MatMul_10367 filter=lfs diff=lfs merge=lfs -text
113
+ onnx__MatMul_10368 filter=lfs diff=lfs merge=lfs -text
114
+ onnx__MatMul_10393 filter=lfs diff=lfs merge=lfs -text
115
+ onnx__MatMul_10394 filter=lfs diff=lfs merge=lfs -text
116
+ onnx__MatMul_10395 filter=lfs diff=lfs merge=lfs -text
117
+ onnx__MatMul_10396 filter=lfs diff=lfs merge=lfs -text
118
+ onnx__MatMul_10397 filter=lfs diff=lfs merge=lfs -text
119
+ onnx__MatMul_10398 filter=lfs diff=lfs merge=lfs -text
120
+ onnx__MatMul_10423 filter=lfs diff=lfs merge=lfs -text
121
+ onnx__MatMul_10424 filter=lfs diff=lfs merge=lfs -text
122
+ onnx__MatMul_10425 filter=lfs diff=lfs merge=lfs -text
123
+ onnx__MatMul_10426 filter=lfs diff=lfs merge=lfs -text
124
+ onnx__MatMul_10427 filter=lfs diff=lfs merge=lfs -text
125
+ onnx__MatMul_10428 filter=lfs diff=lfs merge=lfs -text
126
+ onnx__MatMul_10453 filter=lfs diff=lfs merge=lfs -text
127
+ onnx__MatMul_10454 filter=lfs diff=lfs merge=lfs -text
128
+ onnx__MatMul_10455 filter=lfs diff=lfs merge=lfs -text
129
+ onnx__MatMul_10456 filter=lfs diff=lfs merge=lfs -text
130
+ onnx__MatMul_10457 filter=lfs diff=lfs merge=lfs -text
131
+ onnx__MatMul_10458 filter=lfs diff=lfs merge=lfs -text
132
+ onnx__MatMul_10483 filter=lfs diff=lfs merge=lfs -text
133
+ onnx__MatMul_10484 filter=lfs diff=lfs merge=lfs -text
134
+ onnx__MatMul_10485 filter=lfs diff=lfs merge=lfs -text
135
+ onnx__MatMul_10486 filter=lfs diff=lfs merge=lfs -text
136
+ onnx__MatMul_10487 filter=lfs diff=lfs merge=lfs -text
137
+ onnx__MatMul_10488 filter=lfs diff=lfs merge=lfs -text
138
+ onnx__MatMul_10513 filter=lfs diff=lfs merge=lfs -text
139
+ onnx__MatMul_10514 filter=lfs diff=lfs merge=lfs -text
140
+ onnx__MatMul_10515 filter=lfs diff=lfs merge=lfs -text
141
+ onnx__MatMul_10516 filter=lfs diff=lfs merge=lfs -text
142
+ onnx__MatMul_10517 filter=lfs diff=lfs merge=lfs -text
143
+ onnx__MatMul_10518 filter=lfs diff=lfs merge=lfs -text
144
+ onnx__MatMul_10543 filter=lfs diff=lfs merge=lfs -text
145
+ onnx__MatMul_10544 filter=lfs diff=lfs merge=lfs -text
146
+ onnx__MatMul_10545 filter=lfs diff=lfs merge=lfs -text
147
+ onnx__MatMul_9706 filter=lfs diff=lfs merge=lfs -text
148
+ onnx__MatMul_9707 filter=lfs diff=lfs merge=lfs -text
149
+ onnx__MatMul_9708 filter=lfs diff=lfs merge=lfs -text
150
+ onnx__MatMul_9733 filter=lfs diff=lfs merge=lfs -text
151
+ onnx__MatMul_9734 filter=lfs diff=lfs merge=lfs -text
152
+ onnx__MatMul_9735 filter=lfs diff=lfs merge=lfs -text
153
+ onnx__MatMul_9736 filter=lfs diff=lfs merge=lfs -text
154
+ onnx__MatMul_9737 filter=lfs diff=lfs merge=lfs -text
155
+ onnx__MatMul_9738 filter=lfs diff=lfs merge=lfs -text
156
+ onnx__MatMul_9763 filter=lfs diff=lfs merge=lfs -text
157
+ onnx__MatMul_9764 filter=lfs diff=lfs merge=lfs -text
158
+ onnx__MatMul_9765 filter=lfs diff=lfs merge=lfs -text
159
+ onnx__MatMul_9766 filter=lfs diff=lfs merge=lfs -text
160
+ onnx__MatMul_9767 filter=lfs diff=lfs merge=lfs -text
161
+ onnx__MatMul_9768 filter=lfs diff=lfs merge=lfs -text
162
+ onnx__MatMul_9793 filter=lfs diff=lfs merge=lfs -text
163
+ onnx__MatMul_9794 filter=lfs diff=lfs merge=lfs -text
164
+ onnx__MatMul_9795 filter=lfs diff=lfs merge=lfs -text
165
+ onnx__MatMul_9796 filter=lfs diff=lfs merge=lfs -text
166
+ onnx__MatMul_9797 filter=lfs diff=lfs merge=lfs -text
167
+ onnx__MatMul_9798 filter=lfs diff=lfs merge=lfs -text
168
+ onnx__MatMul_9823 filter=lfs diff=lfs merge=lfs -text
169
+ onnx__MatMul_9824 filter=lfs diff=lfs merge=lfs -text
170
+ onnx__MatMul_9825 filter=lfs diff=lfs merge=lfs -text
171
+ onnx__MatMul_9826 filter=lfs diff=lfs merge=lfs -text
172
+ onnx__MatMul_9827 filter=lfs diff=lfs merge=lfs -text
173
+ onnx__MatMul_9828 filter=lfs diff=lfs merge=lfs -text
174
+ onnx__MatMul_9853 filter=lfs diff=lfs merge=lfs -text
175
+ onnx__MatMul_9854 filter=lfs diff=lfs merge=lfs -text
176
+ onnx__MatMul_9855 filter=lfs diff=lfs merge=lfs -text
177
+ onnx__MatMul_9856 filter=lfs diff=lfs merge=lfs -text
178
+ onnx__MatMul_9857 filter=lfs diff=lfs merge=lfs -text
179
+ onnx__MatMul_9858 filter=lfs diff=lfs merge=lfs -text
180
+ onnx__MatMul_9883 filter=lfs diff=lfs merge=lfs -text
181
+ onnx__MatMul_9884 filter=lfs diff=lfs merge=lfs -text
182
+ onnx__MatMul_9885 filter=lfs diff=lfs merge=lfs -text
183
+ onnx__MatMul_9886 filter=lfs diff=lfs merge=lfs -text
184
+ onnx__MatMul_9887 filter=lfs diff=lfs merge=lfs -text
185
+ onnx__MatMul_9888 filter=lfs diff=lfs merge=lfs -text
186
+ onnx__MatMul_9913 filter=lfs diff=lfs merge=lfs -text
187
+ onnx__MatMul_9914 filter=lfs diff=lfs merge=lfs -text
188
+ onnx__MatMul_9915 filter=lfs diff=lfs merge=lfs -text
189
+ onnx__MatMul_9916 filter=lfs diff=lfs merge=lfs -text
190
+ onnx__MatMul_9917 filter=lfs diff=lfs merge=lfs -text
191
+ onnx__MatMul_9918 filter=lfs diff=lfs merge=lfs -text
192
+ onnx__MatMul_9943 filter=lfs diff=lfs merge=lfs -text
193
+ onnx__MatMul_9944 filter=lfs diff=lfs merge=lfs -text
194
+ onnx__MatMul_9945 filter=lfs diff=lfs merge=lfs -text
195
+ onnx__MatMul_9946 filter=lfs diff=lfs merge=lfs -text
196
+ onnx__MatMul_9947 filter=lfs diff=lfs merge=lfs -text
197
+ onnx__MatMul_9948 filter=lfs diff=lfs merge=lfs -text
198
+ onnx__MatMul_9973 filter=lfs diff=lfs merge=lfs -text
199
+ onnx__MatMul_9974 filter=lfs diff=lfs merge=lfs -text
200
+ onnx__MatMul_9975 filter=lfs diff=lfs merge=lfs -text
201
+ onnx__MatMul_9976 filter=lfs diff=lfs merge=lfs -text
202
+ onnx__MatMul_9977 filter=lfs diff=lfs merge=lfs -text
203
+ onnx__MatMul_9978 filter=lfs diff=lfs merge=lfs -text
204
+ wte.weight filter=lfs diff=lfs merge=lfs -text
h.0.attn.bias ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:83dd5f1a5a07912d2e7dbb73b75edeb2ce4c1f1d6ec48144a887bd4a4f72325b
3
+ size 4194304
h.0.ln_1.bias ADDED
Binary file (16.4 kB). View file
 
h.0.ln_1.weight ADDED
Binary file (16.4 kB). View file
 
h.0.mlp.fc_in.bias ADDED
Binary file (65.5 kB). View file
 
h.0.mlp.fc_out.bias ADDED
Binary file (16.4 kB). View file
 
h.1.ln_1.bias ADDED
Binary file (16.4 kB). View file
 
h.1.ln_1.weight ADDED
Binary file (16.4 kB). View file
 
h.1.mlp.fc_in.bias ADDED
Binary file (65.5 kB). View file
 
h.1.mlp.fc_out.bias ADDED
Binary file (16.4 kB). View file
 
h.10.ln_1.bias ADDED
Binary file (16.4 kB). View file
 
h.10.ln_1.weight ADDED
Binary file (16.4 kB). View file
 
h.10.mlp.fc_in.bias ADDED
Binary file (65.5 kB). View file
 
h.10.mlp.fc_out.bias ADDED
Binary file (16.4 kB). View file
 
h.11.ln_1.bias ADDED
Binary file (16.4 kB). View file
 
h.11.ln_1.weight ADDED
Binary file (16.4 kB). View file
 
h.11.mlp.fc_in.bias ADDED
Binary file (65.5 kB). View file
 
h.11.mlp.fc_out.bias ADDED
Binary file (16.4 kB). View file
 
h.12.ln_1.bias ADDED
Binary file (16.4 kB). View file
 
h.12.ln_1.weight ADDED
Binary file (16.4 kB). View file
 
h.12.mlp.fc_in.bias ADDED
Binary file (65.5 kB). View file
 
h.12.mlp.fc_out.bias ADDED
Binary file (16.4 kB). View file
 
h.13.ln_1.bias ADDED
Binary file (16.4 kB). View file
 
h.13.ln_1.weight ADDED
Binary file (16.4 kB). View file
 
h.13.mlp.fc_in.bias ADDED
Binary file (65.5 kB). View file
 
h.13.mlp.fc_out.bias ADDED
Binary file (16.4 kB). View file
 
h.14.ln_1.bias ADDED
Binary file (16.4 kB). View file
 
h.14.ln_1.weight ADDED
Binary file (16.4 kB). View file
 
h.14.mlp.fc_in.bias ADDED
Binary file (65.5 kB). View file
 
h.14.mlp.fc_out.bias ADDED
Binary file (16.4 kB). View file
 
h.15.ln_1.bias ADDED
Binary file (16.4 kB). View file
 
h.15.ln_1.weight ADDED
Binary file (16.4 kB). View file
 
h.15.mlp.fc_in.bias ADDED
Binary file (65.5 kB). View file
 
h.15.mlp.fc_out.bias ADDED
Binary file (16.4 kB). View file
 
h.16.ln_1.bias ADDED
Binary file (16.4 kB). View file
 
h.16.ln_1.weight ADDED
Binary file (16.4 kB). View file
 
h.16.mlp.fc_in.bias ADDED
Binary file (65.5 kB). View file
 
h.16.mlp.fc_out.bias ADDED
Binary file (16.4 kB). View file
 
h.17.ln_1.bias ADDED
Binary file (16.4 kB). View file
 
h.17.ln_1.weight ADDED
Binary file (16.4 kB). View file
 
h.17.mlp.fc_in.bias ADDED
Binary file (65.5 kB). View file
 
h.17.mlp.fc_out.bias ADDED
Binary file (16.4 kB). View file
 
h.18.ln_1.bias ADDED
Binary file (16.4 kB). View file
 
h.18.ln_1.weight ADDED
Binary file (16.4 kB). View file
 
h.18.mlp.fc_in.bias ADDED
Binary file (65.5 kB). View file
 
h.18.mlp.fc_out.bias ADDED
Binary file (16.4 kB). View file
 
h.19.ln_1.bias ADDED
Binary file (16.4 kB). View file
 
h.19.ln_1.weight ADDED
Binary file (16.4 kB). View file
 
h.19.mlp.fc_in.bias ADDED
Binary file (65.5 kB). View file
 
h.19.mlp.fc_out.bias ADDED
Binary file (16.4 kB). View file