Hugging Face's logo

cerebras
/
Cerebras-GPT-Intermediate

Text Generation
PyTorch
English
causal-lm
Model card Files Community
Yishi Xu commited on
Commit
ab92fcd
•
1 Parent(s): 700743b

added CerebrasGPT-1.3B_muP

Browse files
Files changed (15) hide show
  1. CerebrasGPT-1.3B_muP/checkpoint_14598/config.json +28 -0
  2. CerebrasGPT-1.3B_muP/checkpoint_14598/pytorch_model-00001-of-00001.bin +3 -0
  3. CerebrasGPT-1.3B_muP/checkpoint_14598/pytorch_model.bin.index.json +300 -0
  4. CerebrasGPT-1.3B_muP/checkpoint_21897/config.json +28 -0
  5. CerebrasGPT-1.3B_muP/checkpoint_21897/pytorch_model-00001-of-00001.bin +3 -0
  6. CerebrasGPT-1.3B_muP/checkpoint_21897/pytorch_model.bin.index.json +300 -0
  7. CerebrasGPT-1.3B_muP/checkpoint_24334/config.json +28 -0
  8. CerebrasGPT-1.3B_muP/checkpoint_24334/pytorch_model-00001-of-00001.bin +3 -0
  9. CerebrasGPT-1.3B_muP/checkpoint_24334/pytorch_model.bin.index.json +300 -0
  10. CerebrasGPT-1.3B_muP/checkpoint_4866/config.json +28 -0
  11. CerebrasGPT-1.3B_muP/checkpoint_4866/pytorch_model-00001-of-00001.bin +3 -0
  12. CerebrasGPT-1.3B_muP/checkpoint_4866/pytorch_model.bin.index.json +300 -0
  13. CerebrasGPT-1.3B_muP/checkpoint_9732/config.json +28 -0
  14. CerebrasGPT-1.3B_muP/checkpoint_9732/pytorch_model-00001-of-00001.bin +3 -0
  15. CerebrasGPT-1.3B_muP/checkpoint_9732/pytorch_model.bin.index.json +300 -0
CerebrasGPT-1.3B_muP/checkpoint_14598/config.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "attn_pdrop": 0.0,
3
+ "scale_attn_weights": true,
4
+ "resid_pdrop": 0.0,
5
+ "mup_embeddings_scale": 10,
6
+ "n_inner": 8192,
7
+ "n_embd": 2048,
8
+ "layer_norm_epsilon": 1e-05,
9
+ "n_positions": 2048,
10
+ "activation_function": "gelu",
11
+ "n_head": 16,
12
+ "n_layer": 24,
13
+ "mup_output_alpha": 1.0,
14
+ "mup_width_scale": 0.125,
15
+ "position_embedding_type": "learned",
16
+ "mup_scale_qk_dot_by_d": true,
17
+ "tie_word_embeddings": true,
18
+ "vocab_size": 50257,
19
+ "alibi_scaling": null,
20
+ "embd_pdrop": 0.0,
21
+ "model_type": "btlm",
22
+ "use_cache": true,
23
+ "auto_map": {
24
+ "AutoConfig": "cerebras/btlm-3b-8k-base--configuration_btlm.BTLMConfig",
25
+ "AutoModel": "cerebras/btlm-3b-8k-base--modeling_btlm.BTLMModel",
26
+ "AutoModelForCausalLM": "cerebras/btlm-3b-8k-base--modeling_btlm.BTLMLMHeadModel"
27
+ }
28
+ }
CerebrasGPT-1.3B_muP/checkpoint_14598/pytorch_model-00001-of-00001.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aa76d08d6c89e3e5cdc829c5c986b15085870445cbd215f3cb44f6b3b66160e7
3
+ size 5674695665
CerebrasGPT-1.3B_muP/checkpoint_14598/pytorch_model.bin.index.json ADDED
@@ -0,0 +1,300 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 5674598400
4
+ },
5
+ "weight_map": {
6
+ "transformer.wte.weight": "pytorch_model-00001-of-00001.bin",
7
+ "lm_head.weight": "pytorch_model-00001-of-00001.bin",
8
+ "transformer.wpe.weight": "pytorch_model-00001-of-00001.bin",
9
+ "transformer.ln_f.weight": "pytorch_model-00001-of-00001.bin",
10
+ "transformer.ln_f.bias": "pytorch_model-00001-of-00001.bin",
11
+ "transformer.h.0.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
12
+ "transformer.h.0.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
13
+ "transformer.h.0.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
14
+ "transformer.h.0.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
15
+ "transformer.h.0.ln_1.weight": "pytorch_model-00001-of-00001.bin",
16
+ "transformer.h.0.ln_1.bias": "pytorch_model-00001-of-00001.bin",
17
+ "transformer.h.0.ln_2.weight": "pytorch_model-00001-of-00001.bin",
18
+ "transformer.h.0.ln_2.bias": "pytorch_model-00001-of-00001.bin",
19
+ "transformer.h.0.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
20
+ "transformer.h.0.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
21
+ "transformer.h.0.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
22
+ "transformer.h.0.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
23
+ "transformer.h.1.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
24
+ "transformer.h.1.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
25
+ "transformer.h.1.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
26
+ "transformer.h.1.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
27
+ "transformer.h.1.ln_1.weight": "pytorch_model-00001-of-00001.bin",
28
+ "transformer.h.1.ln_1.bias": "pytorch_model-00001-of-00001.bin",
29
+ "transformer.h.1.ln_2.weight": "pytorch_model-00001-of-00001.bin",
30
+ "transformer.h.1.ln_2.bias": "pytorch_model-00001-of-00001.bin",
31
+ "transformer.h.1.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
32
+ "transformer.h.1.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
33
+ "transformer.h.1.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
34
+ "transformer.h.1.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
35
+ "transformer.h.2.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
36
+ "transformer.h.2.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
37
+ "transformer.h.2.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
38
+ "transformer.h.2.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
39
+ "transformer.h.2.ln_1.weight": "pytorch_model-00001-of-00001.bin",
40
+ "transformer.h.2.ln_1.bias": "pytorch_model-00001-of-00001.bin",
41
+ "transformer.h.2.ln_2.weight": "pytorch_model-00001-of-00001.bin",
42
+ "transformer.h.2.ln_2.bias": "pytorch_model-00001-of-00001.bin",
43
+ "transformer.h.2.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
44
+ "transformer.h.2.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
45
+ "transformer.h.2.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
46
+ "transformer.h.2.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
47
+ "transformer.h.3.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
48
+ "transformer.h.3.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
49
+ "transformer.h.3.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
50
+ "transformer.h.3.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
51
+ "transformer.h.3.ln_1.weight": "pytorch_model-00001-of-00001.bin",
52
+ "transformer.h.3.ln_1.bias": "pytorch_model-00001-of-00001.bin",
53
+ "transformer.h.3.ln_2.weight": "pytorch_model-00001-of-00001.bin",
54
+ "transformer.h.3.ln_2.bias": "pytorch_model-00001-of-00001.bin",
55
+ "transformer.h.3.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
56
+ "transformer.h.3.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
57
+ "transformer.h.3.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
58
+ "transformer.h.3.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
59
+ "transformer.h.4.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
60
+ "transformer.h.4.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
61
+ "transformer.h.4.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
62
+ "transformer.h.4.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
63
+ "transformer.h.4.ln_1.weight": "pytorch_model-00001-of-00001.bin",
64
+ "transformer.h.4.ln_1.bias": "pytorch_model-00001-of-00001.bin",
65
+ "transformer.h.4.ln_2.weight": "pytorch_model-00001-of-00001.bin",
66
+ "transformer.h.4.ln_2.bias": "pytorch_model-00001-of-00001.bin",
67
+ "transformer.h.4.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
68
+ "transformer.h.4.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
69
+ "transformer.h.4.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
70
+ "transformer.h.4.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
71
+ "transformer.h.5.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
72
+ "transformer.h.5.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
73
+ "transformer.h.5.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
74
+ "transformer.h.5.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
75
+ "transformer.h.5.ln_1.weight": "pytorch_model-00001-of-00001.bin",
76
+ "transformer.h.5.ln_1.bias": "pytorch_model-00001-of-00001.bin",
77
+ "transformer.h.5.ln_2.weight": "pytorch_model-00001-of-00001.bin",
78
+ "transformer.h.5.ln_2.bias": "pytorch_model-00001-of-00001.bin",
79
+ "transformer.h.5.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
80
+ "transformer.h.5.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
81
+ "transformer.h.5.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
82
+ "transformer.h.5.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
83
+ "transformer.h.6.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
84
+ "transformer.h.6.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
85
+ "transformer.h.6.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
86
+ "transformer.h.6.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
87
+ "transformer.h.6.ln_1.weight": "pytorch_model-00001-of-00001.bin",
88
+ "transformer.h.6.ln_1.bias": "pytorch_model-00001-of-00001.bin",
89
+ "transformer.h.6.ln_2.weight": "pytorch_model-00001-of-00001.bin",
90
+ "transformer.h.6.ln_2.bias": "pytorch_model-00001-of-00001.bin",
91
+ "transformer.h.6.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
92
+ "transformer.h.6.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
93
+ "transformer.h.6.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
94
+ "transformer.h.6.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
95
+ "transformer.h.7.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
96
+ "transformer.h.7.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
97
+ "transformer.h.7.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
98
+ "transformer.h.7.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
99
+ "transformer.h.7.ln_1.weight": "pytorch_model-00001-of-00001.bin",
100
+ "transformer.h.7.ln_1.bias": "pytorch_model-00001-of-00001.bin",
101
+ "transformer.h.7.ln_2.weight": "pytorch_model-00001-of-00001.bin",
102
+ "transformer.h.7.ln_2.bias": "pytorch_model-00001-of-00001.bin",
103
+ "transformer.h.7.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
104
+ "transformer.h.7.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
105
+ "transformer.h.7.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
106
+ "transformer.h.7.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
107
+ "transformer.h.8.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
108
+ "transformer.h.8.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
109
+ "transformer.h.8.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
110
+ "transformer.h.8.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
111
+ "transformer.h.8.ln_1.weight": "pytorch_model-00001-of-00001.bin",
112
+ "transformer.h.8.ln_1.bias": "pytorch_model-00001-of-00001.bin",
113
+ "transformer.h.8.ln_2.weight": "pytorch_model-00001-of-00001.bin",
114
+ "transformer.h.8.ln_2.bias": "pytorch_model-00001-of-00001.bin",
115
+ "transformer.h.8.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
116
+ "transformer.h.8.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
117
+ "transformer.h.8.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
118
+ "transformer.h.8.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
119
+ "transformer.h.9.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
120
+ "transformer.h.9.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
121
+ "transformer.h.9.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
122
+ "transformer.h.9.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
123
+ "transformer.h.9.ln_1.weight": "pytorch_model-00001-of-00001.bin",
124
+ "transformer.h.9.ln_1.bias": "pytorch_model-00001-of-00001.bin",
125
+ "transformer.h.9.ln_2.weight": "pytorch_model-00001-of-00001.bin",
126
+ "transformer.h.9.ln_2.bias": "pytorch_model-00001-of-00001.bin",
127
+ "transformer.h.9.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
128
+ "transformer.h.9.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
129
+ "transformer.h.9.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
130
+ "transformer.h.9.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
131
+ "transformer.h.10.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
132
+ "transformer.h.10.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
133
+ "transformer.h.10.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
134
+ "transformer.h.10.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
135
+ "transformer.h.10.ln_1.weight": "pytorch_model-00001-of-00001.bin",
136
+ "transformer.h.10.ln_1.bias": "pytorch_model-00001-of-00001.bin",
137
+ "transformer.h.10.ln_2.weight": "pytorch_model-00001-of-00001.bin",
138
+ "transformer.h.10.ln_2.bias": "pytorch_model-00001-of-00001.bin",
139
+ "transformer.h.10.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
140
+ "transformer.h.10.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
141
+ "transformer.h.10.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
142
+ "transformer.h.10.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
143
+ "transformer.h.11.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
144
+ "transformer.h.11.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
145
+ "transformer.h.11.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
146
+ "transformer.h.11.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
147
+ "transformer.h.11.ln_1.weight": "pytorch_model-00001-of-00001.bin",
148
+ "transformer.h.11.ln_1.bias": "pytorch_model-00001-of-00001.bin",
149
+ "transformer.h.11.ln_2.weight": "pytorch_model-00001-of-00001.bin",
150
+ "transformer.h.11.ln_2.bias": "pytorch_model-00001-of-00001.bin",
151
+ "transformer.h.11.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
152
+ "transformer.h.11.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
153
+ "transformer.h.11.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
154
+ "transformer.h.11.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
155
+ "transformer.h.12.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
156
+ "transformer.h.12.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
157
+ "transformer.h.12.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
158
+ "transformer.h.12.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
159
+ "transformer.h.12.ln_1.weight": "pytorch_model-00001-of-00001.bin",
160
+ "transformer.h.12.ln_1.bias": "pytorch_model-00001-of-00001.bin",
161
+ "transformer.h.12.ln_2.weight": "pytorch_model-00001-of-00001.bin",
162
+ "transformer.h.12.ln_2.bias": "pytorch_model-00001-of-00001.bin",
163
+ "transformer.h.12.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
164
+ "transformer.h.12.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
165
+ "transformer.h.12.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
166
+ "transformer.h.12.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
167
+ "transformer.h.13.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
168
+ "transformer.h.13.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
169
+ "transformer.h.13.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
170
+ "transformer.h.13.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
171
+ "transformer.h.13.ln_1.weight": "pytorch_model-00001-of-00001.bin",
172
+ "transformer.h.13.ln_1.bias": "pytorch_model-00001-of-00001.bin",
173
+ "transformer.h.13.ln_2.weight": "pytorch_model-00001-of-00001.bin",
174
+ "transformer.h.13.ln_2.bias": "pytorch_model-00001-of-00001.bin",
175
+ "transformer.h.13.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
176
+ "transformer.h.13.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
177
+ "transformer.h.13.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
178
+ "transformer.h.13.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
179
+ "transformer.h.14.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
180
+ "transformer.h.14.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
181
+ "transformer.h.14.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
182
+ "transformer.h.14.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
183
+ "transformer.h.14.ln_1.weight": "pytorch_model-00001-of-00001.bin",
184
+ "transformer.h.14.ln_1.bias": "pytorch_model-00001-of-00001.bin",
185
+ "transformer.h.14.ln_2.weight": "pytorch_model-00001-of-00001.bin",
186
+ "transformer.h.14.ln_2.bias": "pytorch_model-00001-of-00001.bin",
187
+ "transformer.h.14.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
188
+ "transformer.h.14.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
189
+ "transformer.h.14.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
190
+ "transformer.h.14.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
191
+ "transformer.h.15.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
192
+ "transformer.h.15.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
193
+ "transformer.h.15.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
194
+ "transformer.h.15.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
195
+ "transformer.h.15.ln_1.weight": "pytorch_model-00001-of-00001.bin",
196
+ "transformer.h.15.ln_1.bias": "pytorch_model-00001-of-00001.bin",
197
+ "transformer.h.15.ln_2.weight": "pytorch_model-00001-of-00001.bin",
198
+ "transformer.h.15.ln_2.bias": "pytorch_model-00001-of-00001.bin",
199
+ "transformer.h.15.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
200
+ "transformer.h.15.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
201
+ "transformer.h.15.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
202
+ "transformer.h.15.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
203
+ "transformer.h.16.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
204
+ "transformer.h.16.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
205
+ "transformer.h.16.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
206
+ "transformer.h.16.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
207
+ "transformer.h.16.ln_1.weight": "pytorch_model-00001-of-00001.bin",
208
+ "transformer.h.16.ln_1.bias": "pytorch_model-00001-of-00001.bin",
209
+ "transformer.h.16.ln_2.weight": "pytorch_model-00001-of-00001.bin",
210
+ "transformer.h.16.ln_2.bias": "pytorch_model-00001-of-00001.bin",
211
+ "transformer.h.16.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
212
+ "transformer.h.16.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
213
+ "transformer.h.16.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
214
+ "transformer.h.16.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
215
+ "transformer.h.17.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
216
+ "transformer.h.17.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
217
+ "transformer.h.17.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
218
+ "transformer.h.17.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
219
+ "transformer.h.17.ln_1.weight": "pytorch_model-00001-of-00001.bin",
220
+ "transformer.h.17.ln_1.bias": "pytorch_model-00001-of-00001.bin",
221
+ "transformer.h.17.ln_2.weight": "pytorch_model-00001-of-00001.bin",
222
+ "transformer.h.17.ln_2.bias": "pytorch_model-00001-of-00001.bin",
223
+ "transformer.h.17.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
224
+ "transformer.h.17.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
225
+ "transformer.h.17.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
226
+ "transformer.h.17.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
227
+ "transformer.h.18.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
228
+ "transformer.h.18.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
229
+ "transformer.h.18.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
230
+ "transformer.h.18.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
231
+ "transformer.h.18.ln_1.weight": "pytorch_model-00001-of-00001.bin",
232
+ "transformer.h.18.ln_1.bias": "pytorch_model-00001-of-00001.bin",
233
+ "transformer.h.18.ln_2.weight": "pytorch_model-00001-of-00001.bin",
234
+ "transformer.h.18.ln_2.bias": "pytorch_model-00001-of-00001.bin",
235
+ "transformer.h.18.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
236
+ "transformer.h.18.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
237
+ "transformer.h.18.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
238
+ "transformer.h.18.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
239
+ "transformer.h.19.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
240
+ "transformer.h.19.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
241
+ "transformer.h.19.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
242
+ "transformer.h.19.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
243
+ "transformer.h.19.ln_1.weight": "pytorch_model-00001-of-00001.bin",
244
+ "transformer.h.19.ln_1.bias": "pytorch_model-00001-of-00001.bin",
245
+ "transformer.h.19.ln_2.weight": "pytorch_model-00001-of-00001.bin",
246
+ "transformer.h.19.ln_2.bias": "pytorch_model-00001-of-00001.bin",
247
+ "transformer.h.19.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
248
+ "transformer.h.19.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
249
+ "transformer.h.19.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
250
+ "transformer.h.19.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
251
+ "transformer.h.20.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
252
+ "transformer.h.20.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
253
+ "transformer.h.20.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
254
+ "transformer.h.20.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
255
+ "transformer.h.20.ln_1.weight": "pytorch_model-00001-of-00001.bin",
256
+ "transformer.h.20.ln_1.bias": "pytorch_model-00001-of-00001.bin",
257
+ "transformer.h.20.ln_2.weight": "pytorch_model-00001-of-00001.bin",
258
+ "transformer.h.20.ln_2.bias": "pytorch_model-00001-of-00001.bin",
259
+ "transformer.h.20.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
260
+ "transformer.h.20.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
261
+ "transformer.h.20.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
262
+ "transformer.h.20.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
263
+ "transformer.h.21.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
264
+ "transformer.h.21.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
265
+ "transformer.h.21.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
266
+ "transformer.h.21.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
267
+ "transformer.h.21.ln_1.weight": "pytorch_model-00001-of-00001.bin",
268
+ "transformer.h.21.ln_1.bias": "pytorch_model-00001-of-00001.bin",
269
+ "transformer.h.21.ln_2.weight": "pytorch_model-00001-of-00001.bin",
270
+ "transformer.h.21.ln_2.bias": "pytorch_model-00001-of-00001.bin",
271
+ "transformer.h.21.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
272
+ "transformer.h.21.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
273
+ "transformer.h.21.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
274
+ "transformer.h.21.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
275
+ "transformer.h.22.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
276
+ "transformer.h.22.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
277
+ "transformer.h.22.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
278
+ "transformer.h.22.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
279
+ "transformer.h.22.ln_1.weight": "pytorch_model-00001-of-00001.bin",
280
+ "transformer.h.22.ln_1.bias": "pytorch_model-00001-of-00001.bin",
281
+ "transformer.h.22.ln_2.weight": "pytorch_model-00001-of-00001.bin",
282
+ "transformer.h.22.ln_2.bias": "pytorch_model-00001-of-00001.bin",
283
+ "transformer.h.22.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
284
+ "transformer.h.22.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
285
+ "transformer.h.22.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
286
+ "transformer.h.22.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
287
+ "transformer.h.23.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
288
+ "transformer.h.23.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
289
+ "transformer.h.23.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
290
+ "transformer.h.23.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
291
+ "transformer.h.23.ln_1.weight": "pytorch_model-00001-of-00001.bin",
292
+ "transformer.h.23.ln_1.bias": "pytorch_model-00001-of-00001.bin",
293
+ "transformer.h.23.ln_2.weight": "pytorch_model-00001-of-00001.bin",
294
+ "transformer.h.23.ln_2.bias": "pytorch_model-00001-of-00001.bin",
295
+ "transformer.h.23.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
296
+ "transformer.h.23.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
297
+ "transformer.h.23.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
298
+ "transformer.h.23.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin"
299
+ }
300
+ }
CerebrasGPT-1.3B_muP/checkpoint_21897/config.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "attn_pdrop": 0.0,
3
+ "scale_attn_weights": true,
4
+ "resid_pdrop": 0.0,
5
+ "mup_embeddings_scale": 10,
6
+ "n_inner": 8192,
7
+ "n_embd": 2048,
8
+ "layer_norm_epsilon": 1e-05,
9
+ "n_positions": 2048,
10
+ "activation_function": "gelu",
11
+ "n_head": 16,
12
+ "n_layer": 24,
13
+ "mup_output_alpha": 1.0,
14
+ "mup_width_scale": 0.125,
15
+ "position_embedding_type": "learned",
16
+ "mup_scale_qk_dot_by_d": true,
17
+ "tie_word_embeddings": true,
18
+ "vocab_size": 50257,
19
+ "alibi_scaling": null,
20
+ "embd_pdrop": 0.0,
21
+ "model_type": "btlm",
22
+ "use_cache": true,
23
+ "auto_map": {
24
+ "AutoConfig": "cerebras/btlm-3b-8k-base--configuration_btlm.BTLMConfig",
25
+ "AutoModel": "cerebras/btlm-3b-8k-base--modeling_btlm.BTLMModel",
26
+ "AutoModelForCausalLM": "cerebras/btlm-3b-8k-base--modeling_btlm.BTLMLMHeadModel"
27
+ }
28
+ }
CerebrasGPT-1.3B_muP/checkpoint_21897/pytorch_model-00001-of-00001.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:68830a5a58d713b9d510bc32a768f0259ebedb5c7bc9525527ba79ce2b1fe90f
3
+ size 5674695665
CerebrasGPT-1.3B_muP/checkpoint_21897/pytorch_model.bin.index.json ADDED
@@ -0,0 +1,300 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 5674598400
4
+ },
5
+ "weight_map": {
6
+ "transformer.wte.weight": "pytorch_model-00001-of-00001.bin",
7
+ "lm_head.weight": "pytorch_model-00001-of-00001.bin",
8
+ "transformer.wpe.weight": "pytorch_model-00001-of-00001.bin",
9
+ "transformer.ln_f.weight": "pytorch_model-00001-of-00001.bin",
10
+ "transformer.ln_f.bias": "pytorch_model-00001-of-00001.bin",
11
+ "transformer.h.0.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
12
+ "transformer.h.0.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
13
+ "transformer.h.0.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
14
+ "transformer.h.0.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
15
+ "transformer.h.0.ln_1.weight": "pytorch_model-00001-of-00001.bin",
16
+ "transformer.h.0.ln_1.bias": "pytorch_model-00001-of-00001.bin",
17
+ "transformer.h.0.ln_2.weight": "pytorch_model-00001-of-00001.bin",
18
+ "transformer.h.0.ln_2.bias": "pytorch_model-00001-of-00001.bin",
19
+ "transformer.h.0.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
20
+ "transformer.h.0.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
21
+ "transformer.h.0.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
22
+ "transformer.h.0.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
23
+ "transformer.h.1.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
24
+ "transformer.h.1.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
25
+ "transformer.h.1.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
26
+ "transformer.h.1.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
27
+ "transformer.h.1.ln_1.weight": "pytorch_model-00001-of-00001.bin",
28
+ "transformer.h.1.ln_1.bias": "pytorch_model-00001-of-00001.bin",
29
+ "transformer.h.1.ln_2.weight": "pytorch_model-00001-of-00001.bin",
30
+ "transformer.h.1.ln_2.bias": "pytorch_model-00001-of-00001.bin",
31
+ "transformer.h.1.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
32
+ "transformer.h.1.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
33
+ "transformer.h.1.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
34
+ "transformer.h.1.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
35
+ "transformer.h.2.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
36
+ "transformer.h.2.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
37
+ "transformer.h.2.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
38
+ "transformer.h.2.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
39
+ "transformer.h.2.ln_1.weight": "pytorch_model-00001-of-00001.bin",
40
+ "transformer.h.2.ln_1.bias": "pytorch_model-00001-of-00001.bin",
41
+ "transformer.h.2.ln_2.weight": "pytorch_model-00001-of-00001.bin",
42
+ "transformer.h.2.ln_2.bias": "pytorch_model-00001-of-00001.bin",
43
+ "transformer.h.2.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
44
+ "transformer.h.2.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
45
+ "transformer.h.2.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
46
+ "transformer.h.2.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
47
+ "transformer.h.3.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
48
+ "transformer.h.3.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
49
+ "transformer.h.3.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
50
+ "transformer.h.3.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
51
+ "transformer.h.3.ln_1.weight": "pytorch_model-00001-of-00001.bin",
52
+ "transformer.h.3.ln_1.bias": "pytorch_model-00001-of-00001.bin",
53
+ "transformer.h.3.ln_2.weight": "pytorch_model-00001-of-00001.bin",
54
+ "transformer.h.3.ln_2.bias": "pytorch_model-00001-of-00001.bin",
55
+ "transformer.h.3.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
56
+ "transformer.h.3.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
57
+ "transformer.h.3.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
58
+ "transformer.h.3.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
59
+ "transformer.h.4.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
60
+ "transformer.h.4.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
61
+ "transformer.h.4.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
62
+ "transformer.h.4.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
63
+ "transformer.h.4.ln_1.weight": "pytorch_model-00001-of-00001.bin",
64
+ "transformer.h.4.ln_1.bias": "pytorch_model-00001-of-00001.bin",
65
+ "transformer.h.4.ln_2.weight": "pytorch_model-00001-of-00001.bin",
66
+ "transformer.h.4.ln_2.bias": "pytorch_model-00001-of-00001.bin",
67
+ "transformer.h.4.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
68
+ "transformer.h.4.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
69
+ "transformer.h.4.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
70
+ "transformer.h.4.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
71
+ "transformer.h.5.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
72
+ "transformer.h.5.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
73
+ "transformer.h.5.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
74
+ "transformer.h.5.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
75
+ "transformer.h.5.ln_1.weight": "pytorch_model-00001-of-00001.bin",
76
+ "transformer.h.5.ln_1.bias": "pytorch_model-00001-of-00001.bin",
77
+ "transformer.h.5.ln_2.weight": "pytorch_model-00001-of-00001.bin",
78
+ "transformer.h.5.ln_2.bias": "pytorch_model-00001-of-00001.bin",
79
+ "transformer.h.5.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
80
+ "transformer.h.5.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
81
+ "transformer.h.5.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
82
+ "transformer.h.5.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
83
+ "transformer.h.6.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
84
+ "transformer.h.6.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
85
+ "transformer.h.6.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
86
+ "transformer.h.6.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
87
+ "transformer.h.6.ln_1.weight": "pytorch_model-00001-of-00001.bin",
88
+ "transformer.h.6.ln_1.bias": "pytorch_model-00001-of-00001.bin",
89
+ "transformer.h.6.ln_2.weight": "pytorch_model-00001-of-00001.bin",
90
+ "transformer.h.6.ln_2.bias": "pytorch_model-00001-of-00001.bin",
91
+ "transformer.h.6.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
92
+ "transformer.h.6.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
93
+ "transformer.h.6.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
94
+ "transformer.h.6.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
95
+ "transformer.h.7.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
96
+ "transformer.h.7.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
97
+ "transformer.h.7.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
98
+ "transformer.h.7.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
99
+ "transformer.h.7.ln_1.weight": "pytorch_model-00001-of-00001.bin",
100
+ "transformer.h.7.ln_1.bias": "pytorch_model-00001-of-00001.bin",
101
+ "transformer.h.7.ln_2.weight": "pytorch_model-00001-of-00001.bin",
102
+ "transformer.h.7.ln_2.bias": "pytorch_model-00001-of-00001.bin",
103
+ "transformer.h.7.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
104
+ "transformer.h.7.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
105
+ "transformer.h.7.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
106
+ "transformer.h.7.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
107
+ "transformer.h.8.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
108
+ "transformer.h.8.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
109
+ "transformer.h.8.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
110
+ "transformer.h.8.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
111
+ "transformer.h.8.ln_1.weight": "pytorch_model-00001-of-00001.bin",
112
+ "transformer.h.8.ln_1.bias": "pytorch_model-00001-of-00001.bin",
113
+ "transformer.h.8.ln_2.weight": "pytorch_model-00001-of-00001.bin",
114
+ "transformer.h.8.ln_2.bias": "pytorch_model-00001-of-00001.bin",
115
+ "transformer.h.8.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
116
+ "transformer.h.8.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
117
+ "transformer.h.8.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
118
+ "transformer.h.8.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
119
+ "transformer.h.9.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
120
+ "transformer.h.9.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
121
+ "transformer.h.9.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
122
+ "transformer.h.9.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
123
+ "transformer.h.9.ln_1.weight": "pytorch_model-00001-of-00001.bin",
124
+ "transformer.h.9.ln_1.bias": "pytorch_model-00001-of-00001.bin",
125
+ "transformer.h.9.ln_2.weight": "pytorch_model-00001-of-00001.bin",
126
+ "transformer.h.9.ln_2.bias": "pytorch_model-00001-of-00001.bin",
127
+ "transformer.h.9.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
128
+ "transformer.h.9.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
129
+ "transformer.h.9.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
130
+ "transformer.h.9.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
131
+ "transformer.h.10.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
132
+ "transformer.h.10.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
133
+ "transformer.h.10.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
134
+ "transformer.h.10.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
135
+ "transformer.h.10.ln_1.weight": "pytorch_model-00001-of-00001.bin",
136
+ "transformer.h.10.ln_1.bias": "pytorch_model-00001-of-00001.bin",
137
+ "transformer.h.10.ln_2.weight": "pytorch_model-00001-of-00001.bin",
138
+ "transformer.h.10.ln_2.bias": "pytorch_model-00001-of-00001.bin",
139
+ "transformer.h.10.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
140
+ "transformer.h.10.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
141
+ "transformer.h.10.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
142
+ "transformer.h.10.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
143
+ "transformer.h.11.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
144
+ "transformer.h.11.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
145
+ "transformer.h.11.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
146
+ "transformer.h.11.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
147
+ "transformer.h.11.ln_1.weight": "pytorch_model-00001-of-00001.bin",
148
+ "transformer.h.11.ln_1.bias": "pytorch_model-00001-of-00001.bin",
149
+ "transformer.h.11.ln_2.weight": "pytorch_model-00001-of-00001.bin",
150
+ "transformer.h.11.ln_2.bias": "pytorch_model-00001-of-00001.bin",
151
+ "transformer.h.11.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
152
+ "transformer.h.11.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
153
+ "transformer.h.11.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
154
+ "transformer.h.11.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
155
+ "transformer.h.12.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
156
+ "transformer.h.12.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
157
+ "transformer.h.12.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
158
+ "transformer.h.12.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
159
+ "transformer.h.12.ln_1.weight": "pytorch_model-00001-of-00001.bin",
160
+ "transformer.h.12.ln_1.bias": "pytorch_model-00001-of-00001.bin",
161
+ "transformer.h.12.ln_2.weight": "pytorch_model-00001-of-00001.bin",
162
+ "transformer.h.12.ln_2.bias": "pytorch_model-00001-of-00001.bin",
163
+ "transformer.h.12.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
164
+ "transformer.h.12.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
165
+ "transformer.h.12.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
166
+ "transformer.h.12.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
167
+ "transformer.h.13.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
168
+ "transformer.h.13.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
169
+ "transformer.h.13.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
170
+ "transformer.h.13.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
171
+ "transformer.h.13.ln_1.weight": "pytorch_model-00001-of-00001.bin",
172
+ "transformer.h.13.ln_1.bias": "pytorch_model-00001-of-00001.bin",
173
+ "transformer.h.13.ln_2.weight": "pytorch_model-00001-of-00001.bin",
174
+ "transformer.h.13.ln_2.bias": "pytorch_model-00001-of-00001.bin",
175
+ "transformer.h.13.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
176
+ "transformer.h.13.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
177
+ "transformer.h.13.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
178
+ "transformer.h.13.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
179
+ "transformer.h.14.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
180
+ "transformer.h.14.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
181
+ "transformer.h.14.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
182
+ "transformer.h.14.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
183
+ "transformer.h.14.ln_1.weight": "pytorch_model-00001-of-00001.bin",
184
+ "transformer.h.14.ln_1.bias": "pytorch_model-00001-of-00001.bin",
185
+ "transformer.h.14.ln_2.weight": "pytorch_model-00001-of-00001.bin",
186
+ "transformer.h.14.ln_2.bias": "pytorch_model-00001-of-00001.bin",
187
+ "transformer.h.14.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
188
+ "transformer.h.14.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
189
+ "transformer.h.14.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
190
+ "transformer.h.14.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
191
+ "transformer.h.15.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
192
+ "transformer.h.15.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
193
+ "transformer.h.15.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
194
+ "transformer.h.15.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
195
+ "transformer.h.15.ln_1.weight": "pytorch_model-00001-of-00001.bin",
196
+ "transformer.h.15.ln_1.bias": "pytorch_model-00001-of-00001.bin",
197
+ "transformer.h.15.ln_2.weight": "pytorch_model-00001-of-00001.bin",
198
+ "transformer.h.15.ln_2.bias": "pytorch_model-00001-of-00001.bin",
199
+ "transformer.h.15.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
200
+ "transformer.h.15.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
201
+ "transformer.h.15.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
202
+ "transformer.h.15.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
203
+ "transformer.h.16.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
204
+ "transformer.h.16.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
205
+ "transformer.h.16.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
206
+ "transformer.h.16.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
207
+ "transformer.h.16.ln_1.weight": "pytorch_model-00001-of-00001.bin",
208
+ "transformer.h.16.ln_1.bias": "pytorch_model-00001-of-00001.bin",
209
+ "transformer.h.16.ln_2.weight": "pytorch_model-00001-of-00001.bin",
210
+ "transformer.h.16.ln_2.bias": "pytorch_model-00001-of-00001.bin",
211
+ "transformer.h.16.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
212
+ "transformer.h.16.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
213
+ "transformer.h.16.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
214
+ "transformer.h.16.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
215
+ "transformer.h.17.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
216
+ "transformer.h.17.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
217
+ "transformer.h.17.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
218
+ "transformer.h.17.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
219
+ "transformer.h.17.ln_1.weight": "pytorch_model-00001-of-00001.bin",
220
+ "transformer.h.17.ln_1.bias": "pytorch_model-00001-of-00001.bin",
221
+ "transformer.h.17.ln_2.weight": "pytorch_model-00001-of-00001.bin",
222
+ "transformer.h.17.ln_2.bias": "pytorch_model-00001-of-00001.bin",
223
+ "transformer.h.17.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
224
+ "transformer.h.17.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
225
+ "transformer.h.17.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
226
+ "transformer.h.17.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
227
+ "transformer.h.18.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
228
+ "transformer.h.18.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
229
+ "transformer.h.18.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
230
+ "transformer.h.18.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
231
+ "transformer.h.18.ln_1.weight": "pytorch_model-00001-of-00001.bin",
232
+ "transformer.h.18.ln_1.bias": "pytorch_model-00001-of-00001.bin",
233
+ "transformer.h.18.ln_2.weight": "pytorch_model-00001-of-00001.bin",
234
+ "transformer.h.18.ln_2.bias": "pytorch_model-00001-of-00001.bin",
235
+ "transformer.h.18.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
236
+ "transformer.h.18.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
237
+ "transformer.h.18.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
238
+ "transformer.h.18.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
239
+ "transformer.h.19.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
240
+ "transformer.h.19.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
241
+ "transformer.h.19.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
242
+ "transformer.h.19.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
243
+ "transformer.h.19.ln_1.weight": "pytorch_model-00001-of-00001.bin",
244
+ "transformer.h.19.ln_1.bias": "pytorch_model-00001-of-00001.bin",
245
+ "transformer.h.19.ln_2.weight": "pytorch_model-00001-of-00001.bin",
246
+ "transformer.h.19.ln_2.bias": "pytorch_model-00001-of-00001.bin",
247
+ "transformer.h.19.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
248
+ "transformer.h.19.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
249
+ "transformer.h.19.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
250
+ "transformer.h.19.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
251
+ "transformer.h.20.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
252
+ "transformer.h.20.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
253
+ "transformer.h.20.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
254
+ "transformer.h.20.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
255
+ "transformer.h.20.ln_1.weight": "pytorch_model-00001-of-00001.bin",
256
+ "transformer.h.20.ln_1.bias": "pytorch_model-00001-of-00001.bin",
257
+ "transformer.h.20.ln_2.weight": "pytorch_model-00001-of-00001.bin",
258
+ "transformer.h.20.ln_2.bias": "pytorch_model-00001-of-00001.bin",
259
+ "transformer.h.20.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
260
+ "transformer.h.20.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
261
+ "transformer.h.20.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
262
+ "transformer.h.20.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
263
+ "transformer.h.21.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
264
+ "transformer.h.21.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
265
+ "transformer.h.21.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
266
+ "transformer.h.21.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
267
+ "transformer.h.21.ln_1.weight": "pytorch_model-00001-of-00001.bin",
268
+ "transformer.h.21.ln_1.bias": "pytorch_model-00001-of-00001.bin",
269
+ "transformer.h.21.ln_2.weight": "pytorch_model-00001-of-00001.bin",
270
+ "transformer.h.21.ln_2.bias": "pytorch_model-00001-of-00001.bin",
271
+ "transformer.h.21.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
272
+ "transformer.h.21.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
273
+ "transformer.h.21.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
274
+ "transformer.h.21.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
275
+ "transformer.h.22.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
276
+ "transformer.h.22.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
277
+ "transformer.h.22.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
278
+ "transformer.h.22.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
279
+ "transformer.h.22.ln_1.weight": "pytorch_model-00001-of-00001.bin",
280
+ "transformer.h.22.ln_1.bias": "pytorch_model-00001-of-00001.bin",
281
+ "transformer.h.22.ln_2.weight": "pytorch_model-00001-of-00001.bin",
282
+ "transformer.h.22.ln_2.bias": "pytorch_model-00001-of-00001.bin",
283
+ "transformer.h.22.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
284
+ "transformer.h.22.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
285
+ "transformer.h.22.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
286
+ "transformer.h.22.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
287
+ "transformer.h.23.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
288
+ "transformer.h.23.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
289
+ "transformer.h.23.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
290
+ "transformer.h.23.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
291
+ "transformer.h.23.ln_1.weight": "pytorch_model-00001-of-00001.bin",
292
+ "transformer.h.23.ln_1.bias": "pytorch_model-00001-of-00001.bin",
293
+ "transformer.h.23.ln_2.weight": "pytorch_model-00001-of-00001.bin",
294
+ "transformer.h.23.ln_2.bias": "pytorch_model-00001-of-00001.bin",
295
+ "transformer.h.23.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
296
+ "transformer.h.23.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
297
+ "transformer.h.23.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
298
+ "transformer.h.23.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin"
299
+ }
300
+ }
CerebrasGPT-1.3B_muP/checkpoint_24334/config.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "attn_pdrop": 0.0,
3
+ "scale_attn_weights": true,
4
+ "resid_pdrop": 0.0,
5
+ "mup_embeddings_scale": 10,
6
+ "n_inner": 8192,
7
+ "n_embd": 2048,
8
+ "layer_norm_epsilon": 1e-05,
9
+ "n_positions": 2048,
10
+ "activation_function": "gelu",
11
+ "n_head": 16,
12
+ "n_layer": 24,
13
+ "mup_output_alpha": 1.0,
14
+ "mup_width_scale": 0.125,
15
+ "position_embedding_type": "learned",
16
+ "mup_scale_qk_dot_by_d": true,
17
+ "tie_word_embeddings": true,
18
+ "vocab_size": 50257,
19
+ "alibi_scaling": null,
20
+ "embd_pdrop": 0.0,
21
+ "model_type": "btlm",
22
+ "use_cache": true,
23
+ "auto_map": {
24
+ "AutoConfig": "cerebras/btlm-3b-8k-base--configuration_btlm.BTLMConfig",
25
+ "AutoModel": "cerebras/btlm-3b-8k-base--modeling_btlm.BTLMModel",
26
+ "AutoModelForCausalLM": "cerebras/btlm-3b-8k-base--modeling_btlm.BTLMLMHeadModel"
27
+ }
28
+ }
CerebrasGPT-1.3B_muP/checkpoint_24334/pytorch_model-00001-of-00001.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:26fab85af4320c2e1c2dfc80b10be768ff2cdb05415bddc54e963bd634be5ac7
3
+ size 5674695665
CerebrasGPT-1.3B_muP/checkpoint_24334/pytorch_model.bin.index.json ADDED
@@ -0,0 +1,300 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 5674598400
4
+ },
5
+ "weight_map": {
6
+ "transformer.wte.weight": "pytorch_model-00001-of-00001.bin",
7
+ "lm_head.weight": "pytorch_model-00001-of-00001.bin",
8
+ "transformer.wpe.weight": "pytorch_model-00001-of-00001.bin",
9
+ "transformer.ln_f.weight": "pytorch_model-00001-of-00001.bin",
10
+ "transformer.ln_f.bias": "pytorch_model-00001-of-00001.bin",
11
+ "transformer.h.0.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
12
+ "transformer.h.0.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
13
+ "transformer.h.0.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
14
+ "transformer.h.0.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
15
+ "transformer.h.0.ln_1.weight": "pytorch_model-00001-of-00001.bin",
16
+ "transformer.h.0.ln_1.bias": "pytorch_model-00001-of-00001.bin",
17
+ "transformer.h.0.ln_2.weight": "pytorch_model-00001-of-00001.bin",
18
+ "transformer.h.0.ln_2.bias": "pytorch_model-00001-of-00001.bin",
19
+ "transformer.h.0.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
20
+ "transformer.h.0.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
21
+ "transformer.h.0.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
22
+ "transformer.h.0.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
23
+ "transformer.h.1.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
24
+ "transformer.h.1.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
25
+ "transformer.h.1.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
26
+ "transformer.h.1.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
27
+ "transformer.h.1.ln_1.weight": "pytorch_model-00001-of-00001.bin",
28
+ "transformer.h.1.ln_1.bias": "pytorch_model-00001-of-00001.bin",
29
+ "transformer.h.1.ln_2.weight": "pytorch_model-00001-of-00001.bin",
30
+ "transformer.h.1.ln_2.bias": "pytorch_model-00001-of-00001.bin",
31
+ "transformer.h.1.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
32
+ "transformer.h.1.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
33
+ "transformer.h.1.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
34
+ "transformer.h.1.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
35
+ "transformer.h.2.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
36
+ "transformer.h.2.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
37
+ "transformer.h.2.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
38
+ "transformer.h.2.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
39
+ "transformer.h.2.ln_1.weight": "pytorch_model-00001-of-00001.bin",
40
+ "transformer.h.2.ln_1.bias": "pytorch_model-00001-of-00001.bin",
41
+ "transformer.h.2.ln_2.weight": "pytorch_model-00001-of-00001.bin",
42
+ "transformer.h.2.ln_2.bias": "pytorch_model-00001-of-00001.bin",
43
+ "transformer.h.2.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
44
+ "transformer.h.2.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
45
+ "transformer.h.2.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
46
+ "transformer.h.2.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
47
+ "transformer.h.3.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
48
+ "transformer.h.3.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
49
+ "transformer.h.3.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
50
+ "transformer.h.3.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
51
+ "transformer.h.3.ln_1.weight": "pytorch_model-00001-of-00001.bin",
52
+ "transformer.h.3.ln_1.bias": "pytorch_model-00001-of-00001.bin",
53
+ "transformer.h.3.ln_2.weight": "pytorch_model-00001-of-00001.bin",
54
+ "transformer.h.3.ln_2.bias": "pytorch_model-00001-of-00001.bin",
55
+ "transformer.h.3.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
56
+ "transformer.h.3.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
57
+ "transformer.h.3.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
58
+ "transformer.h.3.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
59
+ "transformer.h.4.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
60
+ "transformer.h.4.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
61
+ "transformer.h.4.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
62
+ "transformer.h.4.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
63
+ "transformer.h.4.ln_1.weight": "pytorch_model-00001-of-00001.bin",
64
+ "transformer.h.4.ln_1.bias": "pytorch_model-00001-of-00001.bin",
65
+ "transformer.h.4.ln_2.weight": "pytorch_model-00001-of-00001.bin",
66
+ "transformer.h.4.ln_2.bias": "pytorch_model-00001-of-00001.bin",
67
+ "transformer.h.4.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
68
+ "transformer.h.4.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
69
+ "transformer.h.4.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
70
+ "transformer.h.4.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
71
+ "transformer.h.5.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
72
+ "transformer.h.5.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
73
+ "transformer.h.5.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
74
+ "transformer.h.5.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
75
+ "transformer.h.5.ln_1.weight": "pytorch_model-00001-of-00001.bin",
76
+ "transformer.h.5.ln_1.bias": "pytorch_model-00001-of-00001.bin",
77
+ "transformer.h.5.ln_2.weight": "pytorch_model-00001-of-00001.bin",
78
+ "transformer.h.5.ln_2.bias": "pytorch_model-00001-of-00001.bin",
79
+ "transformer.h.5.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
80
+ "transformer.h.5.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
81
+ "transformer.h.5.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
82
+ "transformer.h.5.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
83
+ "transformer.h.6.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
84
+ "transformer.h.6.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
85
+ "transformer.h.6.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
86
+ "transformer.h.6.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
87
+ "transformer.h.6.ln_1.weight": "pytorch_model-00001-of-00001.bin",
88
+ "transformer.h.6.ln_1.bias": "pytorch_model-00001-of-00001.bin",
89
+ "transformer.h.6.ln_2.weight": "pytorch_model-00001-of-00001.bin",
90
+ "transformer.h.6.ln_2.bias": "pytorch_model-00001-of-00001.bin",
91
+ "transformer.h.6.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
92
+ "transformer.h.6.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
93
+ "transformer.h.6.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
94
+ "transformer.h.6.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
95
+ "transformer.h.7.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
96
+ "transformer.h.7.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
97
+ "transformer.h.7.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
98
+ "transformer.h.7.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
99
+ "transformer.h.7.ln_1.weight": "pytorch_model-00001-of-00001.bin",
100
+ "transformer.h.7.ln_1.bias": "pytorch_model-00001-of-00001.bin",
101
+ "transformer.h.7.ln_2.weight": "pytorch_model-00001-of-00001.bin",
102
+ "transformer.h.7.ln_2.bias": "pytorch_model-00001-of-00001.bin",
103
+ "transformer.h.7.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
104
+ "transformer.h.7.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
105
+ "transformer.h.7.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
106
+ "transformer.h.7.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
107
+ "transformer.h.8.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
108
+ "transformer.h.8.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
109
+ "transformer.h.8.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
110
+ "transformer.h.8.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
111
+ "transformer.h.8.ln_1.weight": "pytorch_model-00001-of-00001.bin",
112
+ "transformer.h.8.ln_1.bias": "pytorch_model-00001-of-00001.bin",
113
+ "transformer.h.8.ln_2.weight": "pytorch_model-00001-of-00001.bin",
114
+ "transformer.h.8.ln_2.bias": "pytorch_model-00001-of-00001.bin",
115
+ "transformer.h.8.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
116
+ "transformer.h.8.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
117
+ "transformer.h.8.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
118
+ "transformer.h.8.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
119
+ "transformer.h.9.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
120
+ "transformer.h.9.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
121
+ "transformer.h.9.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
122
+ "transformer.h.9.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
123
+ "transformer.h.9.ln_1.weight": "pytorch_model-00001-of-00001.bin",
124
+ "transformer.h.9.ln_1.bias": "pytorch_model-00001-of-00001.bin",
125
+ "transformer.h.9.ln_2.weight": "pytorch_model-00001-of-00001.bin",
126
+ "transformer.h.9.ln_2.bias": "pytorch_model-00001-of-00001.bin",
127
+ "transformer.h.9.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
128
+ "transformer.h.9.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
129
+ "transformer.h.9.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
130
+ "transformer.h.9.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
131
+ "transformer.h.10.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
132
+ "transformer.h.10.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
133
+ "transformer.h.10.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
134
+ "transformer.h.10.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
135
+ "transformer.h.10.ln_1.weight": "pytorch_model-00001-of-00001.bin",
136
+ "transformer.h.10.ln_1.bias": "pytorch_model-00001-of-00001.bin",
137
+ "transformer.h.10.ln_2.weight": "pytorch_model-00001-of-00001.bin",
138
+ "transformer.h.10.ln_2.bias": "pytorch_model-00001-of-00001.bin",
139
+ "transformer.h.10.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
140
+ "transformer.h.10.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
141
+ "transformer.h.10.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
142
+ "transformer.h.10.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
143
+ "transformer.h.11.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
144
+ "transformer.h.11.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
145
+ "transformer.h.11.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
146
+ "transformer.h.11.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
147
+ "transformer.h.11.ln_1.weight": "pytorch_model-00001-of-00001.bin",
148
+ "transformer.h.11.ln_1.bias": "pytorch_model-00001-of-00001.bin",
149
+ "transformer.h.11.ln_2.weight": "pytorch_model-00001-of-00001.bin",
150
+ "transformer.h.11.ln_2.bias": "pytorch_model-00001-of-00001.bin",
151
+ "transformer.h.11.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
152
+ "transformer.h.11.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
153
+ "transformer.h.11.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
154
+ "transformer.h.11.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
155
+ "transformer.h.12.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
156
+ "transformer.h.12.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
157
+ "transformer.h.12.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
158
+ "transformer.h.12.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
159
+ "transformer.h.12.ln_1.weight": "pytorch_model-00001-of-00001.bin",
160
+ "transformer.h.12.ln_1.bias": "pytorch_model-00001-of-00001.bin",
161
+ "transformer.h.12.ln_2.weight": "pytorch_model-00001-of-00001.bin",
162
+ "transformer.h.12.ln_2.bias": "pytorch_model-00001-of-00001.bin",
163
+ "transformer.h.12.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
164
+ "transformer.h.12.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
165
+ "transformer.h.12.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
166
+ "transformer.h.12.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
167
+ "transformer.h.13.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
168
+ "transformer.h.13.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
169
+ "transformer.h.13.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
170
+ "transformer.h.13.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
171
+ "transformer.h.13.ln_1.weight": "pytorch_model-00001-of-00001.bin",
172
+ "transformer.h.13.ln_1.bias": "pytorch_model-00001-of-00001.bin",
173
+ "transformer.h.13.ln_2.weight": "pytorch_model-00001-of-00001.bin",
174
+ "transformer.h.13.ln_2.bias": "pytorch_model-00001-of-00001.bin",
175
+ "transformer.h.13.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
176
+ "transformer.h.13.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
177
+ "transformer.h.13.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
178
+ "transformer.h.13.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
179
+ "transformer.h.14.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
180
+ "transformer.h.14.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
181
+ "transformer.h.14.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
182
+ "transformer.h.14.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
183
+ "transformer.h.14.ln_1.weight": "pytorch_model-00001-of-00001.bin",
184
+ "transformer.h.14.ln_1.bias": "pytorch_model-00001-of-00001.bin",
185
+ "transformer.h.14.ln_2.weight": "pytorch_model-00001-of-00001.bin",
186
+ "transformer.h.14.ln_2.bias": "pytorch_model-00001-of-00001.bin",
187
+ "transformer.h.14.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
188
+ "transformer.h.14.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
189
+ "transformer.h.14.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
190
+ "transformer.h.14.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
191
+ "transformer.h.15.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
192
+ "transformer.h.15.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
193
+ "transformer.h.15.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
194
+ "transformer.h.15.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
195
+ "transformer.h.15.ln_1.weight": "pytorch_model-00001-of-00001.bin",
196
+ "transformer.h.15.ln_1.bias": "pytorch_model-00001-of-00001.bin",
197
+ "transformer.h.15.ln_2.weight": "pytorch_model-00001-of-00001.bin",
198
+ "transformer.h.15.ln_2.bias": "pytorch_model-00001-of-00001.bin",
199
+ "transformer.h.15.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
200
+ "transformer.h.15.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
201
+ "transformer.h.15.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
202
+ "transformer.h.15.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
203
+ "transformer.h.16.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
204
+ "transformer.h.16.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
205
+ "transformer.h.16.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
206
+ "transformer.h.16.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
207
+ "transformer.h.16.ln_1.weight": "pytorch_model-00001-of-00001.bin",
208
+ "transformer.h.16.ln_1.bias": "pytorch_model-00001-of-00001.bin",
209
+ "transformer.h.16.ln_2.weight": "pytorch_model-00001-of-00001.bin",
210
+ "transformer.h.16.ln_2.bias": "pytorch_model-00001-of-00001.bin",
211
+ "transformer.h.16.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
212
+ "transformer.h.16.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
213
+ "transformer.h.16.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
214
+ "transformer.h.16.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
215
+ "transformer.h.17.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
216
+ "transformer.h.17.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
217
+ "transformer.h.17.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
218
+ "transformer.h.17.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
219
+ "transformer.h.17.ln_1.weight": "pytorch_model-00001-of-00001.bin",
220
+ "transformer.h.17.ln_1.bias": "pytorch_model-00001-of-00001.bin",
221
+ "transformer.h.17.ln_2.weight": "pytorch_model-00001-of-00001.bin",
222
+ "transformer.h.17.ln_2.bias": "pytorch_model-00001-of-00001.bin",
223
+ "transformer.h.17.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
224
+ "transformer.h.17.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
225
+ "transformer.h.17.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
226
+ "transformer.h.17.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
227
+ "transformer.h.18.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
228
+ "transformer.h.18.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
229
+ "transformer.h.18.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
230
+ "transformer.h.18.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
231
+ "transformer.h.18.ln_1.weight": "pytorch_model-00001-of-00001.bin",
232
+ "transformer.h.18.ln_1.bias": "pytorch_model-00001-of-00001.bin",
233
+ "transformer.h.18.ln_2.weight": "pytorch_model-00001-of-00001.bin",
234
+ "transformer.h.18.ln_2.bias": "pytorch_model-00001-of-00001.bin",
235
+ "transformer.h.18.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
236
+ "transformer.h.18.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
237
+ "transformer.h.18.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
238
+ "transformer.h.18.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
239
+ "transformer.h.19.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
240
+ "transformer.h.19.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
241
+ "transformer.h.19.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
242
+ "transformer.h.19.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
243
+ "transformer.h.19.ln_1.weight": "pytorch_model-00001-of-00001.bin",
244
+ "transformer.h.19.ln_1.bias": "pytorch_model-00001-of-00001.bin",
245
+ "transformer.h.19.ln_2.weight": "pytorch_model-00001-of-00001.bin",
246
+ "transformer.h.19.ln_2.bias": "pytorch_model-00001-of-00001.bin",
247
+ "transformer.h.19.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
248
+ "transformer.h.19.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
249
+ "transformer.h.19.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
250
+ "transformer.h.19.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
251
+ "transformer.h.20.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
252
+ "transformer.h.20.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
253
+ "transformer.h.20.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
254
+ "transformer.h.20.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
255
+ "transformer.h.20.ln_1.weight": "pytorch_model-00001-of-00001.bin",
256
+ "transformer.h.20.ln_1.bias": "pytorch_model-00001-of-00001.bin",
257
+ "transformer.h.20.ln_2.weight": "pytorch_model-00001-of-00001.bin",
258
+ "transformer.h.20.ln_2.bias": "pytorch_model-00001-of-00001.bin",
259
+ "transformer.h.20.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
260
+ "transformer.h.20.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
261
+ "transformer.h.20.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
262
+ "transformer.h.20.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
263
+ "transformer.h.21.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
264
+ "transformer.h.21.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
265
+ "transformer.h.21.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
266
+ "transformer.h.21.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
267
+ "transformer.h.21.ln_1.weight": "pytorch_model-00001-of-00001.bin",
268
+ "transformer.h.21.ln_1.bias": "pytorch_model-00001-of-00001.bin",
269
+ "transformer.h.21.ln_2.weight": "pytorch_model-00001-of-00001.bin",
270
+ "transformer.h.21.ln_2.bias": "pytorch_model-00001-of-00001.bin",
271
+ "transformer.h.21.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
272
+ "transformer.h.21.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
273
+ "transformer.h.21.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
274
+ "transformer.h.21.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
275
+ "transformer.h.22.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
276
+ "transformer.h.22.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
277
+ "transformer.h.22.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
278
+ "transformer.h.22.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
279
+ "transformer.h.22.ln_1.weight": "pytorch_model-00001-of-00001.bin",
280
+ "transformer.h.22.ln_1.bias": "pytorch_model-00001-of-00001.bin",
281
+ "transformer.h.22.ln_2.weight": "pytorch_model-00001-of-00001.bin",
282
+ "transformer.h.22.ln_2.bias": "pytorch_model-00001-of-00001.bin",
283
+ "transformer.h.22.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
284
+ "transformer.h.22.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
285
+ "transformer.h.22.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
286
+ "transformer.h.22.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
287
+ "transformer.h.23.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
288
+ "transformer.h.23.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
289
+ "transformer.h.23.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
290
+ "transformer.h.23.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
291
+ "transformer.h.23.ln_1.weight": "pytorch_model-00001-of-00001.bin",
292
+ "transformer.h.23.ln_1.bias": "pytorch_model-00001-of-00001.bin",
293
+ "transformer.h.23.ln_2.weight": "pytorch_model-00001-of-00001.bin",
294
+ "transformer.h.23.ln_2.bias": "pytorch_model-00001-of-00001.bin",
295
+ "transformer.h.23.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
296
+ "transformer.h.23.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
297
+ "transformer.h.23.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
298
+ "transformer.h.23.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin"
299
+ }
300
+ }
CerebrasGPT-1.3B_muP/checkpoint_4866/config.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "attn_pdrop": 0.0,
3
+ "scale_attn_weights": true,
4
+ "resid_pdrop": 0.0,
5
+ "n_inner": 8192,
6
+ "n_embd": 2048,
7
+ "layer_norm_epsilon": 1e-05,
8
+ "n_positions": 2048,
9
+ "mup_embeddings_scale": 10,
10
+ "activation_function": "gelu",
11
+ "n_head": 16,
12
+ "n_layer": 24,
13
+ "mup_output_alpha": 1.0,
14
+ "mup_width_scale": 0.125,
15
+ "position_embedding_type": "learned",
16
+ "mup_scale_qk_dot_by_d": true,
17
+ "tie_word_embeddings": true,
18
+ "vocab_size": 50257,
19
+ "alibi_scaling": null,
20
+ "embd_pdrop": 0.0,
21
+ "model_type": "btlm",
22
+ "use_cache": true,
23
+ "auto_map": {
24
+ "AutoConfig": "cerebras/btlm-3b-8k-base--configuration_btlm.BTLMConfig",
25
+ "AutoModel": "cerebras/btlm-3b-8k-base--modeling_btlm.BTLMModel",
26
+ "AutoModelForCausalLM": "cerebras/btlm-3b-8k-base--modeling_btlm.BTLMLMHeadModel"
27
+ }
28
+ }
CerebrasGPT-1.3B_muP/checkpoint_4866/pytorch_model-00001-of-00001.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c514e0d7850c54af2f8bbc001d955855eb7ad85150836dd8c7ab7da7ba4b31a1
3
+ size 5674695665
CerebrasGPT-1.3B_muP/checkpoint_4866/pytorch_model.bin.index.json ADDED
@@ -0,0 +1,300 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 5674598400
4
+ },
5
+ "weight_map": {
6
+ "transformer.wte.weight": "pytorch_model-00001-of-00001.bin",
7
+ "lm_head.weight": "pytorch_model-00001-of-00001.bin",
8
+ "transformer.wpe.weight": "pytorch_model-00001-of-00001.bin",
9
+ "transformer.ln_f.weight": "pytorch_model-00001-of-00001.bin",
10
+ "transformer.ln_f.bias": "pytorch_model-00001-of-00001.bin",
11
+ "transformer.h.0.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
12
+ "transformer.h.0.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
13
+ "transformer.h.0.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
14
+ "transformer.h.0.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
15
+ "transformer.h.0.ln_1.weight": "pytorch_model-00001-of-00001.bin",
16
+ "transformer.h.0.ln_1.bias": "pytorch_model-00001-of-00001.bin",
17
+ "transformer.h.0.ln_2.weight": "pytorch_model-00001-of-00001.bin",
18
+ "transformer.h.0.ln_2.bias": "pytorch_model-00001-of-00001.bin",
19
+ "transformer.h.0.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
20
+ "transformer.h.0.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
21
+ "transformer.h.0.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
22
+ "transformer.h.0.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
23
+ "transformer.h.1.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
24
+ "transformer.h.1.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
25
+ "transformer.h.1.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
26
+ "transformer.h.1.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
27
+ "transformer.h.1.ln_1.weight": "pytorch_model-00001-of-00001.bin",
28
+ "transformer.h.1.ln_1.bias": "pytorch_model-00001-of-00001.bin",
29
+ "transformer.h.1.ln_2.weight": "pytorch_model-00001-of-00001.bin",
30
+ "transformer.h.1.ln_2.bias": "pytorch_model-00001-of-00001.bin",
31
+ "transformer.h.1.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
32
+ "transformer.h.1.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
33
+ "transformer.h.1.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
34
+ "transformer.h.1.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
35
+ "transformer.h.2.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
36
+ "transformer.h.2.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
37
+ "transformer.h.2.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
38
+ "transformer.h.2.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
39
+ "transformer.h.2.ln_1.weight": "pytorch_model-00001-of-00001.bin",
40
+ "transformer.h.2.ln_1.bias": "pytorch_model-00001-of-00001.bin",
41
+ "transformer.h.2.ln_2.weight": "pytorch_model-00001-of-00001.bin",
42
+ "transformer.h.2.ln_2.bias": "pytorch_model-00001-of-00001.bin",
43
+ "transformer.h.2.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
44
+ "transformer.h.2.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
45
+ "transformer.h.2.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
46
+ "transformer.h.2.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
47
+ "transformer.h.3.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
48
+ "transformer.h.3.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
49
+ "transformer.h.3.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
50
+ "transformer.h.3.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
51
+ "transformer.h.3.ln_1.weight": "pytorch_model-00001-of-00001.bin",
52
+ "transformer.h.3.ln_1.bias": "pytorch_model-00001-of-00001.bin",
53
+ "transformer.h.3.ln_2.weight": "pytorch_model-00001-of-00001.bin",
54
+ "transformer.h.3.ln_2.bias": "pytorch_model-00001-of-00001.bin",
55
+ "transformer.h.3.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
56
+ "transformer.h.3.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
57
+ "transformer.h.3.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
58
+ "transformer.h.3.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
59
+ "transformer.h.4.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
60
+ "transformer.h.4.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
61
+ "transformer.h.4.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
62
+ "transformer.h.4.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
63
+ "transformer.h.4.ln_1.weight": "pytorch_model-00001-of-00001.bin",
64
+ "transformer.h.4.ln_1.bias": "pytorch_model-00001-of-00001.bin",
65
+ "transformer.h.4.ln_2.weight": "pytorch_model-00001-of-00001.bin",
66
+ "transformer.h.4.ln_2.bias": "pytorch_model-00001-of-00001.bin",
67
+ "transformer.h.4.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
68
+ "transformer.h.4.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
69
+ "transformer.h.4.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
70
+ "transformer.h.4.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
71
+ "transformer.h.5.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
72
+ "transformer.h.5.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
73
+ "transformer.h.5.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
74
+ "transformer.h.5.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
75
+ "transformer.h.5.ln_1.weight": "pytorch_model-00001-of-00001.bin",
76
+ "transformer.h.5.ln_1.bias": "pytorch_model-00001-of-00001.bin",
77
+ "transformer.h.5.ln_2.weight": "pytorch_model-00001-of-00001.bin",
78
+ "transformer.h.5.ln_2.bias": "pytorch_model-00001-of-00001.bin",
79
+ "transformer.h.5.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
80
+ "transformer.h.5.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
81
+ "transformer.h.5.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
82
+ "transformer.h.5.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
83
+ "transformer.h.6.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
84
+ "transformer.h.6.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
85
+ "transformer.h.6.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
86
+ "transformer.h.6.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
87
+ "transformer.h.6.ln_1.weight": "pytorch_model-00001-of-00001.bin",
88
+ "transformer.h.6.ln_1.bias": "pytorch_model-00001-of-00001.bin",
89
+ "transformer.h.6.ln_2.weight": "pytorch_model-00001-of-00001.bin",
90
+ "transformer.h.6.ln_2.bias": "pytorch_model-00001-of-00001.bin",
91
+ "transformer.h.6.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
92
+ "transformer.h.6.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
93
+ "transformer.h.6.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
94
+ "transformer.h.6.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
95
+ "transformer.h.7.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
96
+ "transformer.h.7.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
97
+ "transformer.h.7.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
98
+ "transformer.h.7.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
99
+ "transformer.h.7.ln_1.weight": "pytorch_model-00001-of-00001.bin",
100
+ "transformer.h.7.ln_1.bias": "pytorch_model-00001-of-00001.bin",
101
+ "transformer.h.7.ln_2.weight": "pytorch_model-00001-of-00001.bin",
102
+ "transformer.h.7.ln_2.bias": "pytorch_model-00001-of-00001.bin",
103
+ "transformer.h.7.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
104
+ "transformer.h.7.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
105
+ "transformer.h.7.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
106
+ "transformer.h.7.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
107
+ "transformer.h.8.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
108
+ "transformer.h.8.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
109
+ "transformer.h.8.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
110
+ "transformer.h.8.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
111
+ "transformer.h.8.ln_1.weight": "pytorch_model-00001-of-00001.bin",
112
+ "transformer.h.8.ln_1.bias": "pytorch_model-00001-of-00001.bin",
113
+ "transformer.h.8.ln_2.weight": "pytorch_model-00001-of-00001.bin",
114
+ "transformer.h.8.ln_2.bias": "pytorch_model-00001-of-00001.bin",
115
+ "transformer.h.8.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
116
+ "transformer.h.8.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
117
+ "transformer.h.8.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
118
+ "transformer.h.8.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
119
+ "transformer.h.9.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
120
+ "transformer.h.9.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
121
+ "transformer.h.9.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
122
+ "transformer.h.9.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
123
+ "transformer.h.9.ln_1.weight": "pytorch_model-00001-of-00001.bin",
124
+ "transformer.h.9.ln_1.bias": "pytorch_model-00001-of-00001.bin",
125
+ "transformer.h.9.ln_2.weight": "pytorch_model-00001-of-00001.bin",
126
+ "transformer.h.9.ln_2.bias": "pytorch_model-00001-of-00001.bin",
127
+ "transformer.h.9.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
128
+ "transformer.h.9.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
129
+ "transformer.h.9.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
130
+ "transformer.h.9.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
131
+ "transformer.h.10.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
132
+ "transformer.h.10.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
133
+ "transformer.h.10.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
134
+ "transformer.h.10.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
135
+ "transformer.h.10.ln_1.weight": "pytorch_model-00001-of-00001.bin",
136
+ "transformer.h.10.ln_1.bias": "pytorch_model-00001-of-00001.bin",
137
+ "transformer.h.10.ln_2.weight": "pytorch_model-00001-of-00001.bin",
138
+ "transformer.h.10.ln_2.bias": "pytorch_model-00001-of-00001.bin",
139
+ "transformer.h.10.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
140
+ "transformer.h.10.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
141
+ "transformer.h.10.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
142
+ "transformer.h.10.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
143
+ "transformer.h.11.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
144
+ "transformer.h.11.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
145
+ "transformer.h.11.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
146
+ "transformer.h.11.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
147
+ "transformer.h.11.ln_1.weight": "pytorch_model-00001-of-00001.bin",
148
+ "transformer.h.11.ln_1.bias": "pytorch_model-00001-of-00001.bin",
149
+ "transformer.h.11.ln_2.weight": "pytorch_model-00001-of-00001.bin",
150
+ "transformer.h.11.ln_2.bias": "pytorch_model-00001-of-00001.bin",
151
+ "transformer.h.11.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
152
+ "transformer.h.11.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
153
+ "transformer.h.11.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
154
+ "transformer.h.11.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
155
+ "transformer.h.12.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
156
+ "transformer.h.12.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
157
+ "transformer.h.12.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
158
+ "transformer.h.12.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
159
+ "transformer.h.12.ln_1.weight": "pytorch_model-00001-of-00001.bin",
160
+ "transformer.h.12.ln_1.bias": "pytorch_model-00001-of-00001.bin",
161
+ "transformer.h.12.ln_2.weight": "pytorch_model-00001-of-00001.bin",
162
+ "transformer.h.12.ln_2.bias": "pytorch_model-00001-of-00001.bin",
163
+ "transformer.h.12.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
164
+ "transformer.h.12.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
165
+ "transformer.h.12.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
166
+ "transformer.h.12.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
167
+ "transformer.h.13.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
168
+ "transformer.h.13.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
169
+ "transformer.h.13.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
170
+ "transformer.h.13.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
171
+ "transformer.h.13.ln_1.weight": "pytorch_model-00001-of-00001.bin",
172
+ "transformer.h.13.ln_1.bias": "pytorch_model-00001-of-00001.bin",
173
+ "transformer.h.13.ln_2.weight": "pytorch_model-00001-of-00001.bin",
174
+ "transformer.h.13.ln_2.bias": "pytorch_model-00001-of-00001.bin",
175
+ "transformer.h.13.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
176
+ "transformer.h.13.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
177
+ "transformer.h.13.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
178
+ "transformer.h.13.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
179
+ "transformer.h.14.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
180
+ "transformer.h.14.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
181
+ "transformer.h.14.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
182
+ "transformer.h.14.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
183
+ "transformer.h.14.ln_1.weight": "pytorch_model-00001-of-00001.bin",
184
+ "transformer.h.14.ln_1.bias": "pytorch_model-00001-of-00001.bin",
185
+ "transformer.h.14.ln_2.weight": "pytorch_model-00001-of-00001.bin",
186
+ "transformer.h.14.ln_2.bias": "pytorch_model-00001-of-00001.bin",
187
+ "transformer.h.14.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
188
+ "transformer.h.14.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
189
+ "transformer.h.14.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
190
+ "transformer.h.14.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
191
+ "transformer.h.15.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
192
+ "transformer.h.15.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
193
+ "transformer.h.15.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
194
+ "transformer.h.15.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
195
+ "transformer.h.15.ln_1.weight": "pytorch_model-00001-of-00001.bin",
196
+ "transformer.h.15.ln_1.bias": "pytorch_model-00001-of-00001.bin",
197
+ "transformer.h.15.ln_2.weight": "pytorch_model-00001-of-00001.bin",
198
+ "transformer.h.15.ln_2.bias": "pytorch_model-00001-of-00001.bin",
199
+ "transformer.h.15.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
200
+ "transformer.h.15.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
201
+ "transformer.h.15.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
202
+ "transformer.h.15.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
203
+ "transformer.h.16.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
204
+ "transformer.h.16.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
205
+ "transformer.h.16.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
206
+ "transformer.h.16.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
207
+ "transformer.h.16.ln_1.weight": "pytorch_model-00001-of-00001.bin",
208
+ "transformer.h.16.ln_1.bias": "pytorch_model-00001-of-00001.bin",
209
+ "transformer.h.16.ln_2.weight": "pytorch_model-00001-of-00001.bin",
210
+ "transformer.h.16.ln_2.bias": "pytorch_model-00001-of-00001.bin",
211
+ "transformer.h.16.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
212
+ "transformer.h.16.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
213
+ "transformer.h.16.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
214
+ "transformer.h.16.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
215
+ "transformer.h.17.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
216
+ "transformer.h.17.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
217
+ "transformer.h.17.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
218
+ "transformer.h.17.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
219
+ "transformer.h.17.ln_1.weight": "pytorch_model-00001-of-00001.bin",
220
+ "transformer.h.17.ln_1.bias": "pytorch_model-00001-of-00001.bin",
221
+ "transformer.h.17.ln_2.weight": "pytorch_model-00001-of-00001.bin",
222
+ "transformer.h.17.ln_2.bias": "pytorch_model-00001-of-00001.bin",
223
+ "transformer.h.17.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
224
+ "transformer.h.17.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
225
+ "transformer.h.17.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
226
+ "transformer.h.17.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
227
+ "transformer.h.18.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
228
+ "transformer.h.18.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
229
+ "transformer.h.18.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
230
+ "transformer.h.18.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
231
+ "transformer.h.18.ln_1.weight": "pytorch_model-00001-of-00001.bin",
232
+ "transformer.h.18.ln_1.bias": "pytorch_model-00001-of-00001.bin",
233
+ "transformer.h.18.ln_2.weight": "pytorch_model-00001-of-00001.bin",
234
+ "transformer.h.18.ln_2.bias": "pytorch_model-00001-of-00001.bin",
235
+ "transformer.h.18.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
236
+ "transformer.h.18.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
237
+ "transformer.h.18.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
238
+ "transformer.h.18.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
239
+ "transformer.h.19.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
240
+ "transformer.h.19.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
241
+ "transformer.h.19.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
242
+ "transformer.h.19.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
243
+ "transformer.h.19.ln_1.weight": "pytorch_model-00001-of-00001.bin",
244
+ "transformer.h.19.ln_1.bias": "pytorch_model-00001-of-00001.bin",
245
+ "transformer.h.19.ln_2.weight": "pytorch_model-00001-of-00001.bin",
246
+ "transformer.h.19.ln_2.bias": "pytorch_model-00001-of-00001.bin",
247
+ "transformer.h.19.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
248
+ "transformer.h.19.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
249
+ "transformer.h.19.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
250
+ "transformer.h.19.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
251
+ "transformer.h.20.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
252
+ "transformer.h.20.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
253
+ "transformer.h.20.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
254
+ "transformer.h.20.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
255
+ "transformer.h.20.ln_1.weight": "pytorch_model-00001-of-00001.bin",
256
+ "transformer.h.20.ln_1.bias": "pytorch_model-00001-of-00001.bin",
257
+ "transformer.h.20.ln_2.weight": "pytorch_model-00001-of-00001.bin",
258
+ "transformer.h.20.ln_2.bias": "pytorch_model-00001-of-00001.bin",
259
+ "transformer.h.20.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
260
+ "transformer.h.20.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
261
+ "transformer.h.20.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
262
+ "transformer.h.20.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
263
+ "transformer.h.21.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
264
+ "transformer.h.21.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
265
+ "transformer.h.21.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
266
+ "transformer.h.21.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
267
+ "transformer.h.21.ln_1.weight": "pytorch_model-00001-of-00001.bin",
268
+ "transformer.h.21.ln_1.bias": "pytorch_model-00001-of-00001.bin",
269
+ "transformer.h.21.ln_2.weight": "pytorch_model-00001-of-00001.bin",
270
+ "transformer.h.21.ln_2.bias": "pytorch_model-00001-of-00001.bin",
271
+ "transformer.h.21.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
272
+ "transformer.h.21.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
273
+ "transformer.h.21.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
274
+ "transformer.h.21.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
275
+ "transformer.h.22.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
276
+ "transformer.h.22.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
277
+ "transformer.h.22.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
278
+ "transformer.h.22.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
279
+ "transformer.h.22.ln_1.weight": "pytorch_model-00001-of-00001.bin",
280
+ "transformer.h.22.ln_1.bias": "pytorch_model-00001-of-00001.bin",
281
+ "transformer.h.22.ln_2.weight": "pytorch_model-00001-of-00001.bin",
282
+ "transformer.h.22.ln_2.bias": "pytorch_model-00001-of-00001.bin",
283
+ "transformer.h.22.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
284
+ "transformer.h.22.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
285
+ "transformer.h.22.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
286
+ "transformer.h.22.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
287
+ "transformer.h.23.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
288
+ "transformer.h.23.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
289
+ "transformer.h.23.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
290
+ "transformer.h.23.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
291
+ "transformer.h.23.ln_1.weight": "pytorch_model-00001-of-00001.bin",
292
+ "transformer.h.23.ln_1.bias": "pytorch_model-00001-of-00001.bin",
293
+ "transformer.h.23.ln_2.weight": "pytorch_model-00001-of-00001.bin",
294
+ "transformer.h.23.ln_2.bias": "pytorch_model-00001-of-00001.bin",
295
+ "transformer.h.23.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
296
+ "transformer.h.23.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
297
+ "transformer.h.23.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
298
+ "transformer.h.23.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin"
299
+ }
300
+ }
CerebrasGPT-1.3B_muP/checkpoint_9732/config.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "attn_pdrop": 0.0,
3
+ "scale_attn_weights": true,
4
+ "resid_pdrop": 0.0,
5
+ "mup_embeddings_scale": 10,
6
+ "n_inner": 8192,
7
+ "n_embd": 2048,
8
+ "layer_norm_epsilon": 1e-05,
9
+ "n_positions": 2048,
10
+ "activation_function": "gelu",
11
+ "n_head": 16,
12
+ "n_layer": 24,
13
+ "mup_output_alpha": 1.0,
14
+ "mup_width_scale": 0.125,
15
+ "position_embedding_type": "learned",
16
+ "mup_scale_qk_dot_by_d": true,
17
+ "tie_word_embeddings": true,
18
+ "vocab_size": 50257,
19
+ "alibi_scaling": null,
20
+ "embd_pdrop": 0.0,
21
+ "model_type": "btlm",
22
+ "use_cache": true,
23
+ "auto_map": {
24
+ "AutoConfig": "cerebras/btlm-3b-8k-base--configuration_btlm.BTLMConfig",
25
+ "AutoModel": "cerebras/btlm-3b-8k-base--modeling_btlm.BTLMModel",
26
+ "AutoModelForCausalLM": "cerebras/btlm-3b-8k-base--modeling_btlm.BTLMLMHeadModel"
27
+ }
28
+ }
CerebrasGPT-1.3B_muP/checkpoint_9732/pytorch_model-00001-of-00001.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f8f60b8283d4bf810a4e271febec084f0d110b71d6d8e864cd44100f94bddf1e
3
+ size 5674695617
CerebrasGPT-1.3B_muP/checkpoint_9732/pytorch_model.bin.index.json ADDED
@@ -0,0 +1,300 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 5674598400
4
+ },
5
+ "weight_map": {
6
+ "transformer.wte.weight": "pytorch_model-00001-of-00001.bin",
7
+ "transformer.wpe.weight": "pytorch_model-00001-of-00001.bin",
8
+ "transformer.h.0.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
9
+ "transformer.h.0.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
10
+ "transformer.h.0.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
11
+ "transformer.h.0.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
12
+ "transformer.h.0.ln_1.weight": "pytorch_model-00001-of-00001.bin",
13
+ "transformer.h.0.ln_1.bias": "pytorch_model-00001-of-00001.bin",
14
+ "transformer.h.0.ln_2.weight": "pytorch_model-00001-of-00001.bin",
15
+ "transformer.h.0.ln_2.bias": "pytorch_model-00001-of-00001.bin",
16
+ "transformer.h.0.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
17
+ "transformer.h.0.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
18
+ "transformer.h.0.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
19
+ "transformer.h.0.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
20
+ "transformer.h.1.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
21
+ "transformer.h.1.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
22
+ "transformer.h.1.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
23
+ "transformer.h.1.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
24
+ "transformer.h.1.ln_1.weight": "pytorch_model-00001-of-00001.bin",
25
+ "transformer.h.1.ln_1.bias": "pytorch_model-00001-of-00001.bin",
26
+ "transformer.h.1.ln_2.weight": "pytorch_model-00001-of-00001.bin",
27
+ "transformer.h.1.ln_2.bias": "pytorch_model-00001-of-00001.bin",
28
+ "transformer.h.1.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
29
+ "transformer.h.1.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
30
+ "transformer.h.1.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
31
+ "transformer.h.1.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
32
+ "transformer.h.2.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
33
+ "transformer.h.2.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
34
+ "transformer.h.2.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
35
+ "transformer.h.2.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
36
+ "transformer.h.2.ln_1.weight": "pytorch_model-00001-of-00001.bin",
37
+ "transformer.h.2.ln_1.bias": "pytorch_model-00001-of-00001.bin",
38
+ "transformer.h.2.ln_2.weight": "pytorch_model-00001-of-00001.bin",
39
+ "transformer.h.2.ln_2.bias": "pytorch_model-00001-of-00001.bin",
40
+ "transformer.h.2.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
41
+ "transformer.h.2.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
42
+ "transformer.h.2.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
43
+ "transformer.h.2.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
44
+ "transformer.h.3.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
45
+ "transformer.h.3.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
46
+ "transformer.h.3.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
47
+ "transformer.h.3.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
48
+ "transformer.h.3.ln_1.weight": "pytorch_model-00001-of-00001.bin",
49
+ "transformer.h.3.ln_1.bias": "pytorch_model-00001-of-00001.bin",
50
+ "transformer.h.3.ln_2.weight": "pytorch_model-00001-of-00001.bin",
51
+ "transformer.h.3.ln_2.bias": "pytorch_model-00001-of-00001.bin",
52
+ "transformer.h.3.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
53
+ "transformer.h.3.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
54
+ "transformer.h.3.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
55
+ "transformer.h.3.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
56
+ "transformer.h.4.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
57
+ "transformer.h.4.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
58
+ "transformer.h.4.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
59
+ "transformer.h.4.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
60
+ "transformer.h.4.ln_1.weight": "pytorch_model-00001-of-00001.bin",
61
+ "transformer.h.4.ln_1.bias": "pytorch_model-00001-of-00001.bin",
62
+ "transformer.h.4.ln_2.weight": "pytorch_model-00001-of-00001.bin",
63
+ "transformer.h.4.ln_2.bias": "pytorch_model-00001-of-00001.bin",
64
+ "transformer.h.4.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
65
+ "transformer.h.4.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
66
+ "transformer.h.4.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
67
+ "transformer.h.4.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
68
+ "transformer.h.5.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
69
+ "transformer.h.5.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
70
+ "transformer.h.5.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
71
+ "transformer.h.5.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
72
+ "transformer.h.5.ln_1.weight": "pytorch_model-00001-of-00001.bin",
73
+ "transformer.h.5.ln_1.bias": "pytorch_model-00001-of-00001.bin",
74
+ "transformer.h.5.ln_2.weight": "pytorch_model-00001-of-00001.bin",
75
+ "transformer.h.5.ln_2.bias": "pytorch_model-00001-of-00001.bin",
76
+ "transformer.h.5.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
77
+ "transformer.h.5.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
78
+ "transformer.h.5.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
79
+ "transformer.h.5.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
80
+ "transformer.h.6.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
81
+ "transformer.h.6.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
82
+ "transformer.h.6.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
83
+ "transformer.h.6.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
84
+ "transformer.h.6.ln_1.weight": "pytorch_model-00001-of-00001.bin",
85
+ "transformer.h.6.ln_1.bias": "pytorch_model-00001-of-00001.bin",
86
+ "transformer.h.6.ln_2.weight": "pytorch_model-00001-of-00001.bin",
87
+ "transformer.h.6.ln_2.bias": "pytorch_model-00001-of-00001.bin",
88
+ "transformer.h.6.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
89
+ "transformer.h.6.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
90
+ "transformer.h.6.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
91
+ "transformer.h.6.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
92
+ "transformer.h.7.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
93
+ "transformer.h.7.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
94
+ "transformer.h.7.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
95
+ "transformer.h.7.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
96
+ "transformer.h.7.ln_1.weight": "pytorch_model-00001-of-00001.bin",
97
+ "transformer.h.7.ln_1.bias": "pytorch_model-00001-of-00001.bin",
98
+ "transformer.h.7.ln_2.weight": "pytorch_model-00001-of-00001.bin",
99
+ "transformer.h.7.ln_2.bias": "pytorch_model-00001-of-00001.bin",
100
+ "transformer.h.7.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
101
+ "transformer.h.7.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
102
+ "transformer.h.7.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
103
+ "transformer.h.7.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
104
+ "transformer.h.8.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
105
+ "transformer.h.8.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
106
+ "transformer.h.8.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
107
+ "transformer.h.8.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
108
+ "transformer.h.8.ln_1.weight": "pytorch_model-00001-of-00001.bin",
109
+ "transformer.h.8.ln_1.bias": "pytorch_model-00001-of-00001.bin",
110
+ "transformer.h.8.ln_2.weight": "pytorch_model-00001-of-00001.bin",
111
+ "transformer.h.8.ln_2.bias": "pytorch_model-00001-of-00001.bin",
112
+ "transformer.h.8.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
113
+ "transformer.h.8.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
114
+ "transformer.h.8.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
115
+ "transformer.h.8.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
116
+ "transformer.h.9.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
117
+ "transformer.h.9.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
118
+ "transformer.h.9.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
119
+ "transformer.h.9.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
120
+ "transformer.h.9.ln_1.weight": "pytorch_model-00001-of-00001.bin",
121
+ "transformer.h.9.ln_1.bias": "pytorch_model-00001-of-00001.bin",
122
+ "transformer.h.9.ln_2.weight": "pytorch_model-00001-of-00001.bin",
123
+ "transformer.h.9.ln_2.bias": "pytorch_model-00001-of-00001.bin",
124
+ "transformer.h.9.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
125
+ "transformer.h.9.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
126
+ "transformer.h.9.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
127
+ "transformer.h.9.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
128
+ "transformer.h.10.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
129
+ "transformer.h.10.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
130
+ "transformer.h.10.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
131
+ "transformer.h.10.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
132
+ "transformer.h.10.ln_1.weight": "pytorch_model-00001-of-00001.bin",
133
+ "transformer.h.10.ln_1.bias": "pytorch_model-00001-of-00001.bin",
134
+ "transformer.h.10.ln_2.weight": "pytorch_model-00001-of-00001.bin",
135
+ "transformer.h.10.ln_2.bias": "pytorch_model-00001-of-00001.bin",
136
+ "transformer.h.10.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
137
+ "transformer.h.10.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
138
+ "transformer.h.10.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
139
+ "transformer.h.10.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
140
+ "transformer.h.11.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
141
+ "transformer.h.11.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
142
+ "transformer.h.11.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
143
+ "transformer.h.11.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
144
+ "transformer.h.11.ln_1.weight": "pytorch_model-00001-of-00001.bin",
145
+ "transformer.h.11.ln_1.bias": "pytorch_model-00001-of-00001.bin",
146
+ "transformer.h.11.ln_2.weight": "pytorch_model-00001-of-00001.bin",
147
+ "transformer.h.11.ln_2.bias": "pytorch_model-00001-of-00001.bin",
148
+ "transformer.h.11.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
149
+ "transformer.h.11.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
150
+ "transformer.h.11.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
151
+ "transformer.h.11.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
152
+ "transformer.h.12.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
153
+ "transformer.h.12.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
154
+ "transformer.h.12.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
155
+ "transformer.h.12.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
156
+ "transformer.h.12.ln_1.weight": "pytorch_model-00001-of-00001.bin",
157
+ "transformer.h.12.ln_1.bias": "pytorch_model-00001-of-00001.bin",
158
+ "transformer.h.12.ln_2.weight": "pytorch_model-00001-of-00001.bin",
159
+ "transformer.h.12.ln_2.bias": "pytorch_model-00001-of-00001.bin",
160
+ "transformer.h.12.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
161
+ "transformer.h.12.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
162
+ "transformer.h.12.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
163
+ "transformer.h.12.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
164
+ "transformer.h.13.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
165
+ "transformer.h.13.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
166
+ "transformer.h.13.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
167
+ "transformer.h.13.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
168
+ "transformer.h.13.ln_1.weight": "pytorch_model-00001-of-00001.bin",
169
+ "transformer.h.13.ln_1.bias": "pytorch_model-00001-of-00001.bin",
170
+ "transformer.h.13.ln_2.weight": "pytorch_model-00001-of-00001.bin",
171
+ "transformer.h.13.ln_2.bias": "pytorch_model-00001-of-00001.bin",
172
+ "transformer.h.13.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
173
+ "transformer.h.13.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
174
+ "transformer.h.13.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
175
+ "transformer.h.13.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
176
+ "transformer.h.14.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
177
+ "transformer.h.14.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
178
+ "transformer.h.14.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
179
+ "transformer.h.14.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
180
+ "transformer.h.14.ln_1.weight": "pytorch_model-00001-of-00001.bin",
181
+ "transformer.h.14.ln_1.bias": "pytorch_model-00001-of-00001.bin",
182
+ "transformer.h.14.ln_2.weight": "pytorch_model-00001-of-00001.bin",
183
+ "transformer.h.14.ln_2.bias": "pytorch_model-00001-of-00001.bin",
184
+ "transformer.h.14.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
185
+ "transformer.h.14.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
186
+ "transformer.h.14.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
187
+ "transformer.h.14.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
188
+ "transformer.h.15.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
189
+ "transformer.h.15.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
190
+ "transformer.h.15.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
191
+ "transformer.h.15.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
192
+ "transformer.h.15.ln_1.weight": "pytorch_model-00001-of-00001.bin",
193
+ "transformer.h.15.ln_1.bias": "pytorch_model-00001-of-00001.bin",
194
+ "transformer.h.15.ln_2.weight": "pytorch_model-00001-of-00001.bin",
195
+ "transformer.h.15.ln_2.bias": "pytorch_model-00001-of-00001.bin",
196
+ "transformer.h.15.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
197
+ "transformer.h.15.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
198
+ "transformer.h.15.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
199
+ "transformer.h.15.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
200
+ "transformer.h.16.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
201
+ "transformer.h.16.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
202
+ "transformer.h.16.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
203
+ "transformer.h.16.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
204
+ "transformer.h.16.ln_1.weight": "pytorch_model-00001-of-00001.bin",
205
+ "transformer.h.16.ln_1.bias": "pytorch_model-00001-of-00001.bin",
206
+ "transformer.h.16.ln_2.weight": "pytorch_model-00001-of-00001.bin",
207
+ "transformer.h.16.ln_2.bias": "pytorch_model-00001-of-00001.bin",
208
+ "transformer.h.16.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
209
+ "transformer.h.16.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
210
+ "transformer.h.16.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
211
+ "transformer.h.16.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
212
+ "transformer.h.17.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
213
+ "transformer.h.17.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
214
+ "transformer.h.17.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
215
+ "transformer.h.17.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
216
+ "transformer.h.17.ln_1.weight": "pytorch_model-00001-of-00001.bin",
217
+ "transformer.h.17.ln_1.bias": "pytorch_model-00001-of-00001.bin",
218
+ "transformer.h.17.ln_2.weight": "pytorch_model-00001-of-00001.bin",
219
+ "transformer.h.17.ln_2.bias": "pytorch_model-00001-of-00001.bin",
220
+ "transformer.h.17.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
221
+ "transformer.h.17.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
222
+ "transformer.h.17.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
223
+ "transformer.h.17.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
224
+ "transformer.h.18.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
225
+ "transformer.h.18.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
226
+ "transformer.h.18.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
227
+ "transformer.h.18.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
228
+ "transformer.h.18.ln_1.weight": "pytorch_model-00001-of-00001.bin",
229
+ "transformer.h.18.ln_1.bias": "pytorch_model-00001-of-00001.bin",
230
+ "transformer.h.18.ln_2.weight": "pytorch_model-00001-of-00001.bin",
231
+ "transformer.h.18.ln_2.bias": "pytorch_model-00001-of-00001.bin",
232
+ "transformer.h.18.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
233
+ "transformer.h.18.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
234
+ "transformer.h.18.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
235
+ "transformer.h.18.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
236
+ "transformer.h.19.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
237
+ "transformer.h.19.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
238
+ "transformer.h.19.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
239
+ "transformer.h.19.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
240
+ "transformer.h.19.ln_1.weight": "pytorch_model-00001-of-00001.bin",
241
+ "transformer.h.19.ln_1.bias": "pytorch_model-00001-of-00001.bin",
242
+ "transformer.h.19.ln_2.weight": "pytorch_model-00001-of-00001.bin",
243
+ "transformer.h.19.ln_2.bias": "pytorch_model-00001-of-00001.bin",
244
+ "transformer.h.19.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
245
+ "transformer.h.19.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
246
+ "transformer.h.19.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
247
+ "transformer.h.19.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
248
+ "transformer.h.20.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
249
+ "transformer.h.20.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
250
+ "transformer.h.20.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
251
+ "transformer.h.20.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
252
+ "transformer.h.20.ln_1.weight": "pytorch_model-00001-of-00001.bin",
253
+ "transformer.h.20.ln_1.bias": "pytorch_model-00001-of-00001.bin",
254
+ "transformer.h.20.ln_2.weight": "pytorch_model-00001-of-00001.bin",
255
+ "transformer.h.20.ln_2.bias": "pytorch_model-00001-of-00001.bin",
256
+ "transformer.h.20.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
257
+ "transformer.h.20.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
258
+ "transformer.h.20.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
259
+ "transformer.h.20.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
260
+ "transformer.h.21.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
261
+ "transformer.h.21.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
262
+ "transformer.h.21.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
263
+ "transformer.h.21.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
264
+ "transformer.h.21.ln_1.weight": "pytorch_model-00001-of-00001.bin",
265
+ "transformer.h.21.ln_1.bias": "pytorch_model-00001-of-00001.bin",
266
+ "transformer.h.21.ln_2.weight": "pytorch_model-00001-of-00001.bin",
267
+ "transformer.h.21.ln_2.bias": "pytorch_model-00001-of-00001.bin",
268
+ "transformer.h.21.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
269
+ "transformer.h.21.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
270
+ "transformer.h.21.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
271
+ "transformer.h.21.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
272
+ "transformer.h.22.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
273
+ "transformer.h.22.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
274
+ "transformer.h.22.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
275
+ "transformer.h.22.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
276
+ "transformer.h.22.ln_1.weight": "pytorch_model-00001-of-00001.bin",
277
+ "transformer.h.22.ln_1.bias": "pytorch_model-00001-of-00001.bin",
278
+ "transformer.h.22.ln_2.weight": "pytorch_model-00001-of-00001.bin",
279
+ "transformer.h.22.ln_2.bias": "pytorch_model-00001-of-00001.bin",
280
+ "transformer.h.22.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
281
+ "transformer.h.22.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
282
+ "transformer.h.22.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
283
+ "transformer.h.22.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
284
+ "transformer.h.23.attn.c_attn.weight": "pytorch_model-00001-of-00001.bin",
285
+ "transformer.h.23.attn.c_attn.bias": "pytorch_model-00001-of-00001.bin",
286
+ "transformer.h.23.attn.c_proj.weight": "pytorch_model-00001-of-00001.bin",
287
+ "transformer.h.23.attn.c_proj.bias": "pytorch_model-00001-of-00001.bin",
288
+ "transformer.h.23.ln_1.weight": "pytorch_model-00001-of-00001.bin",
289
+ "transformer.h.23.ln_1.bias": "pytorch_model-00001-of-00001.bin",
290
+ "transformer.h.23.ln_2.weight": "pytorch_model-00001-of-00001.bin",
291
+ "transformer.h.23.ln_2.bias": "pytorch_model-00001-of-00001.bin",
292
+ "transformer.h.23.mlp.c_fc.weight": "pytorch_model-00001-of-00001.bin",
293
+ "transformer.h.23.mlp.c_fc.bias": "pytorch_model-00001-of-00001.bin",
294
+ "transformer.h.23.mlp.c_proj.weight": "pytorch_model-00001-of-00001.bin",
295
+ "transformer.h.23.mlp.c_proj.bias": "pytorch_model-00001-of-00001.bin",
296
+ "transformer.ln_f.weight": "pytorch_model-00001-of-00001.bin",
297
+ "transformer.ln_f.bias": "pytorch_model-00001-of-00001.bin",
298
+ "lm_head.weight": "pytorch_model-00001-of-00001.bin"
299
+ }
300
+ }