Hugging Face's logo

GeoV
/
GeoV-9b

Text Generation
Transformers
PyTorch
English
geov
causal-lm
custom_code
Model card Files Community
vpj commited on
Commit
91c8a3a
•
1 Parent(s): a269360

init

Browse files
Files changed (16) hide show
  1. config.json +21 -0
  2. generation_config.json +6 -0
  3. pytorch_model-00001-of-00010.bin +3 -0
  4. pytorch_model-00002-of-00010.bin +3 -0
  5. pytorch_model-00003-of-00010.bin +3 -0
  6. pytorch_model-00004-of-00010.bin +3 -0
  7. pytorch_model-00005-of-00010.bin +3 -0
  8. pytorch_model-00006-of-00010.bin +3 -0
  9. pytorch_model-00007-of-00010.bin +3 -0
  10. pytorch_model-00008-of-00010.bin +3 -0
  11. pytorch_model-00009-of-00010.bin +3 -0
  12. pytorch_model-00010-of-00010.bin +3 -0
  13. pytorch_model.bin.index.json +396 -0
  14. special_tokens_map.json +5 -0
  15. spiece.model +3 -0
  16. tokenizer_config.json +8 -0
config.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "GeoVForCausalLM"
4
+ ],
5
+ "bos_token_id": 0,
6
+ "eos_token_id": 2,
7
+ "hidden_size": 5120,
8
+ "intermediate_size": 20480,
9
+ "layer_norm_eps": 0.0001,
10
+ "max_position_embeddings": 2049,
11
+ "model_type": "geov",
12
+ "num_attention_heads": 40,
13
+ "num_hidden_layers": 32,
14
+ "rotary_emb_base": 10000,
15
+ "tie_word_embeddings": false,
16
+ "torch_dtype": "bfloat16",
17
+ "transformers_version": "4.28.0.dev0",
18
+ "use_cache": true,
19
+ "use_extra_biases_ffn": false,
20
+ "vocab_size": 65536
21
+ }
generation_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 0,
4
+ "eos_token_id": 2,
5
+ "transformers_version": "4.28.0.dev0"
6
+ }
pytorch_model-00001-of-00010.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1ede87f9ae81fd8b16128157f47e4fe71ea05344eee98f6fb65426bf577c4e1c
3
+ size 1942283571
pytorch_model-00002-of-00010.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ca2fc549434f3943b062cfbdff60b42481456739b47d16fc1afded3e5fe1111a
3
+ size 1904665657
pytorch_model-00003-of-00010.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8204517aaa30e5896495cbb35b0b809b8495e8cd7a9f9bfca450be9c56fba7eb
3
+ size 2005313137
pytorch_model-00004-of-00010.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d44feee89788d655bbe70be34646f84760abcd7904be062e298ebfd6e393461b
3
+ size 1904655819
pytorch_model-00005-of-00010.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8aabe32e13fd0a1c20d820fc9a681d40011af46599c85598939fd35a87cc7a30
3
+ size 1904665721
pytorch_model-00006-of-00010.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:459fec422b7f3475c53b033e0697772ce94e8673b9b2541ee9209d7c68ab72e6
3
+ size 2005313201
pytorch_model-00007-of-00010.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b67d8f8d8c12182afb9d2dd1944e7baab560cae1ec965193609360705cf6cbbe
3
+ size 1904655819
pytorch_model-00008-of-00010.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f28dc95c8f3c225c7b3501580f0814d94097a87f9beac3dd9057b1af020c9370
3
+ size 1904665721
pytorch_model-00009-of-00010.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0f62ce69598ad11f8891642dae0222bbd8a9c3bb54250cafa3a7543fbbee9e0d
3
+ size 2005313201
pytorch_model-00010-of-00010.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0f4cb73e90561633e9ded31ef557f72681f9b58a459fb7f724a4c891f8a01fdf
3
+ size 1447340142
pytorch_model.bin.index.json ADDED
@@ -0,0 +1,396 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 18811187204.0
4
+ },
5
+ "weight_map": {
6
+ "embed_out.bias": "pytorch_model-00010-of-00010.bin",
7
+ "embed_out.weight": "pytorch_model-00010-of-00010.bin",
8
+ "geov.embed_in.weight": "pytorch_model-00001-of-00010.bin",
9
+ "geov.final_layer_norm.bias": "pytorch_model-00010-of-00010.bin",
10
+ "geov.final_layer_norm.weight": "pytorch_model-00010-of-00010.bin",
11
+ "geov.layers.0.attention.causal_mask": "pytorch_model-00001-of-00010.bin",
12
+ "geov.layers.0.attention.dense.weight": "pytorch_model-00001-of-00010.bin",
13
+ "geov.layers.0.attention.qkv.bias": "pytorch_model-00001-of-00010.bin",
14
+ "geov.layers.0.attention.qkv.weight": "pytorch_model-00001-of-00010.bin",
15
+ "geov.layers.0.attention.rotary_emb.inv_freq": "pytorch_model-00001-of-00010.bin",
16
+ "geov.layers.0.input_layernorm.bias": "pytorch_model-00001-of-00010.bin",
17
+ "geov.layers.0.input_layernorm.weight": "pytorch_model-00001-of-00010.bin",
18
+ "geov.layers.0.mlp.dense_2h_to_h.weight": "pytorch_model-00001-of-00010.bin",
19
+ "geov.layers.0.mlp.dense_h_to_4h.bias": "pytorch_model-00001-of-00010.bin",
20
+ "geov.layers.0.mlp.dense_h_to_4h.weight": "pytorch_model-00001-of-00010.bin",
21
+ "geov.layers.0.post_attention_layernorm.bias": "pytorch_model-00001-of-00010.bin",
22
+ "geov.layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00010.bin",
23
+ "geov.layers.1.attention.causal_mask": "pytorch_model-00001-of-00010.bin",
24
+ "geov.layers.1.attention.dense.weight": "pytorch_model-00001-of-00010.bin",
25
+ "geov.layers.1.attention.qkv.bias": "pytorch_model-00001-of-00010.bin",
26
+ "geov.layers.1.attention.qkv.weight": "pytorch_model-00001-of-00010.bin",
27
+ "geov.layers.1.attention.rotary_emb.inv_freq": "pytorch_model-00001-of-00010.bin",
28
+ "geov.layers.1.input_layernorm.bias": "pytorch_model-00001-of-00010.bin",
29
+ "geov.layers.1.input_layernorm.weight": "pytorch_model-00001-of-00010.bin",
30
+ "geov.layers.1.mlp.dense_2h_to_h.weight": "pytorch_model-00001-of-00010.bin",
31
+ "geov.layers.1.mlp.dense_h_to_4h.bias": "pytorch_model-00001-of-00010.bin",
32
+ "geov.layers.1.mlp.dense_h_to_4h.weight": "pytorch_model-00001-of-00010.bin",
33
+ "geov.layers.1.post_attention_layernorm.bias": "pytorch_model-00001-of-00010.bin",
34
+ "geov.layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00010.bin",
35
+ "geov.layers.10.attention.causal_mask": "pytorch_model-00004-of-00010.bin",
36
+ "geov.layers.10.attention.dense.weight": "pytorch_model-00004-of-00010.bin",
37
+ "geov.layers.10.attention.qkv.bias": "pytorch_model-00004-of-00010.bin",
38
+ "geov.layers.10.attention.qkv.weight": "pytorch_model-00004-of-00010.bin",
39
+ "geov.layers.10.attention.rotary_emb.inv_freq": "pytorch_model-00004-of-00010.bin",
40
+ "geov.layers.10.input_layernorm.bias": "pytorch_model-00004-of-00010.bin",
41
+ "geov.layers.10.input_layernorm.weight": "pytorch_model-00004-of-00010.bin",
42
+ "geov.layers.10.mlp.dense_2h_to_h.weight": "pytorch_model-00004-of-00010.bin",
43
+ "geov.layers.10.mlp.dense_h_to_4h.bias": "pytorch_model-00004-of-00010.bin",
44
+ "geov.layers.10.mlp.dense_h_to_4h.weight": "pytorch_model-00004-of-00010.bin",
45
+ "geov.layers.10.post_attention_layernorm.bias": "pytorch_model-00004-of-00010.bin",
46
+ "geov.layers.10.post_attention_layernorm.weight": "pytorch_model-00004-of-00010.bin",
47
+ "geov.layers.11.attention.causal_mask": "pytorch_model-00004-of-00010.bin",
48
+ "geov.layers.11.attention.dense.weight": "pytorch_model-00004-of-00010.bin",
49
+ "geov.layers.11.attention.qkv.bias": "pytorch_model-00004-of-00010.bin",
50
+ "geov.layers.11.attention.qkv.weight": "pytorch_model-00004-of-00010.bin",
51
+ "geov.layers.11.attention.rotary_emb.inv_freq": "pytorch_model-00004-of-00010.bin",
52
+ "geov.layers.11.input_layernorm.bias": "pytorch_model-00004-of-00010.bin",
53
+ "geov.layers.11.input_layernorm.weight": "pytorch_model-00004-of-00010.bin",
54
+ "geov.layers.11.mlp.dense_2h_to_h.weight": "pytorch_model-00004-of-00010.bin",
55
+ "geov.layers.11.mlp.dense_h_to_4h.bias": "pytorch_model-00004-of-00010.bin",
56
+ "geov.layers.11.mlp.dense_h_to_4h.weight": "pytorch_model-00004-of-00010.bin",
57
+ "geov.layers.11.post_attention_layernorm.bias": "pytorch_model-00004-of-00010.bin",
58
+ "geov.layers.11.post_attention_layernorm.weight": "pytorch_model-00004-of-00010.bin",
59
+ "geov.layers.12.attention.causal_mask": "pytorch_model-00004-of-00010.bin",
60
+ "geov.layers.12.attention.dense.weight": "pytorch_model-00004-of-00010.bin",
61
+ "geov.layers.12.attention.qkv.bias": "pytorch_model-00004-of-00010.bin",
62
+ "geov.layers.12.attention.qkv.weight": "pytorch_model-00004-of-00010.bin",
63
+ "geov.layers.12.attention.rotary_emb.inv_freq": "pytorch_model-00004-of-00010.bin",
64
+ "geov.layers.12.input_layernorm.bias": "pytorch_model-00004-of-00010.bin",
65
+ "geov.layers.12.input_layernorm.weight": "pytorch_model-00004-of-00010.bin",
66
+ "geov.layers.12.mlp.dense_2h_to_h.weight": "pytorch_model-00004-of-00010.bin",
67
+ "geov.layers.12.mlp.dense_h_to_4h.bias": "pytorch_model-00004-of-00010.bin",
68
+ "geov.layers.12.mlp.dense_h_to_4h.weight": "pytorch_model-00004-of-00010.bin",
69
+ "geov.layers.12.post_attention_layernorm.bias": "pytorch_model-00004-of-00010.bin",
70
+ "geov.layers.12.post_attention_layernorm.weight": "pytorch_model-00004-of-00010.bin",
71
+ "geov.layers.13.attention.causal_mask": "pytorch_model-00004-of-00010.bin",
72
+ "geov.layers.13.attention.dense.weight": "pytorch_model-00004-of-00010.bin",
73
+ "geov.layers.13.attention.qkv.bias": "pytorch_model-00004-of-00010.bin",
74
+ "geov.layers.13.attention.qkv.weight": "pytorch_model-00004-of-00010.bin",
75
+ "geov.layers.13.attention.rotary_emb.inv_freq": "pytorch_model-00004-of-00010.bin",
76
+ "geov.layers.13.input_layernorm.bias": "pytorch_model-00004-of-00010.bin",
77
+ "geov.layers.13.input_layernorm.weight": "pytorch_model-00004-of-00010.bin",
78
+ "geov.layers.13.mlp.dense_2h_to_h.weight": "pytorch_model-00005-of-00010.bin",
79
+ "geov.layers.13.mlp.dense_h_to_4h.bias": "pytorch_model-00005-of-00010.bin",
80
+ "geov.layers.13.mlp.dense_h_to_4h.weight": "pytorch_model-00005-of-00010.bin",
81
+ "geov.layers.13.post_attention_layernorm.bias": "pytorch_model-00004-of-00010.bin",
82
+ "geov.layers.13.post_attention_layernorm.weight": "pytorch_model-00004-of-00010.bin",
83
+ "geov.layers.14.attention.causal_mask": "pytorch_model-00005-of-00010.bin",
84
+ "geov.layers.14.attention.dense.weight": "pytorch_model-00005-of-00010.bin",
85
+ "geov.layers.14.attention.qkv.bias": "pytorch_model-00005-of-00010.bin",
86
+ "geov.layers.14.attention.qkv.weight": "pytorch_model-00005-of-00010.bin",
87
+ "geov.layers.14.attention.rotary_emb.inv_freq": "pytorch_model-00005-of-00010.bin",
88
+ "geov.layers.14.input_layernorm.bias": "pytorch_model-00005-of-00010.bin",
89
+ "geov.layers.14.input_layernorm.weight": "pytorch_model-00005-of-00010.bin",
90
+ "geov.layers.14.mlp.dense_2h_to_h.weight": "pytorch_model-00005-of-00010.bin",
91
+ "geov.layers.14.mlp.dense_h_to_4h.bias": "pytorch_model-00005-of-00010.bin",
92
+ "geov.layers.14.mlp.dense_h_to_4h.weight": "pytorch_model-00005-of-00010.bin",
93
+ "geov.layers.14.post_attention_layernorm.bias": "pytorch_model-00005-of-00010.bin",
94
+ "geov.layers.14.post_attention_layernorm.weight": "pytorch_model-00005-of-00010.bin",
95
+ "geov.layers.15.attention.causal_mask": "pytorch_model-00005-of-00010.bin",
96
+ "geov.layers.15.attention.dense.weight": "pytorch_model-00005-of-00010.bin",
97
+ "geov.layers.15.attention.qkv.bias": "pytorch_model-00005-of-00010.bin",
98
+ "geov.layers.15.attention.qkv.weight": "pytorch_model-00005-of-00010.bin",
99
+ "geov.layers.15.attention.rotary_emb.inv_freq": "pytorch_model-00005-of-00010.bin",
100
+ "geov.layers.15.input_layernorm.bias": "pytorch_model-00005-of-00010.bin",
101
+ "geov.layers.15.input_layernorm.weight": "pytorch_model-00005-of-00010.bin",
102
+ "geov.layers.15.mlp.dense_2h_to_h.weight": "pytorch_model-00005-of-00010.bin",
103
+ "geov.layers.15.mlp.dense_h_to_4h.bias": "pytorch_model-00005-of-00010.bin",
104
+ "geov.layers.15.mlp.dense_h_to_4h.weight": "pytorch_model-00005-of-00010.bin",
105
+ "geov.layers.15.post_attention_layernorm.bias": "pytorch_model-00005-of-00010.bin",
106
+ "geov.layers.15.post_attention_layernorm.weight": "pytorch_model-00005-of-00010.bin",
107
+ "geov.layers.16.attention.causal_mask": "pytorch_model-00005-of-00010.bin",
108
+ "geov.layers.16.attention.dense.weight": "pytorch_model-00005-of-00010.bin",
109
+ "geov.layers.16.attention.qkv.bias": "pytorch_model-00005-of-00010.bin",
110
+ "geov.layers.16.attention.qkv.weight": "pytorch_model-00005-of-00010.bin",
111
+ "geov.layers.16.attention.rotary_emb.inv_freq": "pytorch_model-00005-of-00010.bin",
112
+ "geov.layers.16.input_layernorm.bias": "pytorch_model-00005-of-00010.bin",
113
+ "geov.layers.16.input_layernorm.weight": "pytorch_model-00005-of-00010.bin",
114
+ "geov.layers.16.mlp.dense_2h_to_h.weight": "pytorch_model-00005-of-00010.bin",
115
+ "geov.layers.16.mlp.dense_h_to_4h.bias": "pytorch_model-00005-of-00010.bin",
116
+ "geov.layers.16.mlp.dense_h_to_4h.weight": "pytorch_model-00005-of-00010.bin",
117
+ "geov.layers.16.post_attention_layernorm.bias": "pytorch_model-00005-of-00010.bin",
118
+ "geov.layers.16.post_attention_layernorm.weight": "pytorch_model-00005-of-00010.bin",
119
+ "geov.layers.17.attention.causal_mask": "pytorch_model-00005-of-00010.bin",
120
+ "geov.layers.17.attention.dense.weight": "pytorch_model-00006-of-00010.bin",
121
+ "geov.layers.17.attention.qkv.bias": "pytorch_model-00006-of-00010.bin",
122
+ "geov.layers.17.attention.qkv.weight": "pytorch_model-00006-of-00010.bin",
123
+ "geov.layers.17.attention.rotary_emb.inv_freq": "pytorch_model-00005-of-00010.bin",
124
+ "geov.layers.17.input_layernorm.bias": "pytorch_model-00005-of-00010.bin",
125
+ "geov.layers.17.input_layernorm.weight": "pytorch_model-00005-of-00010.bin",
126
+ "geov.layers.17.mlp.dense_2h_to_h.weight": "pytorch_model-00006-of-00010.bin",
127
+ "geov.layers.17.mlp.dense_h_to_4h.bias": "pytorch_model-00006-of-00010.bin",
128
+ "geov.layers.17.mlp.dense_h_to_4h.weight": "pytorch_model-00006-of-00010.bin",
129
+ "geov.layers.17.post_attention_layernorm.bias": "pytorch_model-00005-of-00010.bin",
130
+ "geov.layers.17.post_attention_layernorm.weight": "pytorch_model-00005-of-00010.bin",
131
+ "geov.layers.18.attention.causal_mask": "pytorch_model-00006-of-00010.bin",
132
+ "geov.layers.18.attention.dense.weight": "pytorch_model-00006-of-00010.bin",
133
+ "geov.layers.18.attention.qkv.bias": "pytorch_model-00006-of-00010.bin",
134
+ "geov.layers.18.attention.qkv.weight": "pytorch_model-00006-of-00010.bin",
135
+ "geov.layers.18.attention.rotary_emb.inv_freq": "pytorch_model-00006-of-00010.bin",
136
+ "geov.layers.18.input_layernorm.bias": "pytorch_model-00006-of-00010.bin",
137
+ "geov.layers.18.input_layernorm.weight": "pytorch_model-00006-of-00010.bin",
138
+ "geov.layers.18.mlp.dense_2h_to_h.weight": "pytorch_model-00006-of-00010.bin",
139
+ "geov.layers.18.mlp.dense_h_to_4h.bias": "pytorch_model-00006-of-00010.bin",
140
+ "geov.layers.18.mlp.dense_h_to_4h.weight": "pytorch_model-00006-of-00010.bin",
141
+ "geov.layers.18.post_attention_layernorm.bias": "pytorch_model-00006-of-00010.bin",
142
+ "geov.layers.18.post_attention_layernorm.weight": "pytorch_model-00006-of-00010.bin",
143
+ "geov.layers.19.attention.causal_mask": "pytorch_model-00006-of-00010.bin",
144
+ "geov.layers.19.attention.dense.weight": "pytorch_model-00006-of-00010.bin",
145
+ "geov.layers.19.attention.qkv.bias": "pytorch_model-00006-of-00010.bin",
146
+ "geov.layers.19.attention.qkv.weight": "pytorch_model-00006-of-00010.bin",
147
+ "geov.layers.19.attention.rotary_emb.inv_freq": "pytorch_model-00006-of-00010.bin",
148
+ "geov.layers.19.input_layernorm.bias": "pytorch_model-00006-of-00010.bin",
149
+ "geov.layers.19.input_layernorm.weight": "pytorch_model-00006-of-00010.bin",
150
+ "geov.layers.19.mlp.dense_2h_to_h.weight": "pytorch_model-00006-of-00010.bin",
151
+ "geov.layers.19.mlp.dense_h_to_4h.bias": "pytorch_model-00006-of-00010.bin",
152
+ "geov.layers.19.mlp.dense_h_to_4h.weight": "pytorch_model-00006-of-00010.bin",
153
+ "geov.layers.19.post_attention_layernorm.bias": "pytorch_model-00006-of-00010.bin",
154
+ "geov.layers.19.post_attention_layernorm.weight": "pytorch_model-00006-of-00010.bin",
155
+ "geov.layers.2.attention.causal_mask": "pytorch_model-00001-of-00010.bin",
156
+ "geov.layers.2.attention.dense.weight": "pytorch_model-00001-of-00010.bin",
157
+ "geov.layers.2.attention.qkv.bias": "pytorch_model-00001-of-00010.bin",
158
+ "geov.layers.2.attention.qkv.weight": "pytorch_model-00001-of-00010.bin",
159
+ "geov.layers.2.attention.rotary_emb.inv_freq": "pytorch_model-00001-of-00010.bin",
160
+ "geov.layers.2.input_layernorm.bias": "pytorch_model-00001-of-00010.bin",
161
+ "geov.layers.2.input_layernorm.weight": "pytorch_model-00001-of-00010.bin",
162
+ "geov.layers.2.mlp.dense_2h_to_h.weight": "pytorch_model-00002-of-00010.bin",
163
+ "geov.layers.2.mlp.dense_h_to_4h.bias": "pytorch_model-00002-of-00010.bin",
164
+ "geov.layers.2.mlp.dense_h_to_4h.weight": "pytorch_model-00002-of-00010.bin",
165
+ "geov.layers.2.post_attention_layernorm.bias": "pytorch_model-00001-of-00010.bin",
166
+ "geov.layers.2.post_attention_layernorm.weight": "pytorch_model-00001-of-00010.bin",
167
+ "geov.layers.20.attention.causal_mask": "pytorch_model-00006-of-00010.bin",
168
+ "geov.layers.20.attention.dense.weight": "pytorch_model-00006-of-00010.bin",
169
+ "geov.layers.20.attention.qkv.bias": "pytorch_model-00006-of-00010.bin",
170
+ "geov.layers.20.attention.qkv.weight": "pytorch_model-00006-of-00010.bin",
171
+ "geov.layers.20.attention.rotary_emb.inv_freq": "pytorch_model-00006-of-00010.bin",
172
+ "geov.layers.20.input_layernorm.bias": "pytorch_model-00006-of-00010.bin",
173
+ "geov.layers.20.input_layernorm.weight": "pytorch_model-00006-of-00010.bin",
174
+ "geov.layers.20.mlp.dense_2h_to_h.weight": "pytorch_model-00007-of-00010.bin",
175
+ "geov.layers.20.mlp.dense_h_to_4h.bias": "pytorch_model-00006-of-00010.bin",
176
+ "geov.layers.20.mlp.dense_h_to_4h.weight": "pytorch_model-00006-of-00010.bin",
177
+ "geov.layers.20.post_attention_layernorm.bias": "pytorch_model-00006-of-00010.bin",
178
+ "geov.layers.20.post_attention_layernorm.weight": "pytorch_model-00006-of-00010.bin",
179
+ "geov.layers.21.attention.causal_mask": "pytorch_model-00007-of-00010.bin",
180
+ "geov.layers.21.attention.dense.weight": "pytorch_model-00007-of-00010.bin",
181
+ "geov.layers.21.attention.qkv.bias": "pytorch_model-00007-of-00010.bin",
182
+ "geov.layers.21.attention.qkv.weight": "pytorch_model-00007-of-00010.bin",
183
+ "geov.layers.21.attention.rotary_emb.inv_freq": "pytorch_model-00007-of-00010.bin",
184
+ "geov.layers.21.input_layernorm.bias": "pytorch_model-00007-of-00010.bin",
185
+ "geov.layers.21.input_layernorm.weight": "pytorch_model-00007-of-00010.bin",
186
+ "geov.layers.21.mlp.dense_2h_to_h.weight": "pytorch_model-00007-of-00010.bin",
187
+ "geov.layers.21.mlp.dense_h_to_4h.bias": "pytorch_model-00007-of-00010.bin",
188
+ "geov.layers.21.mlp.dense_h_to_4h.weight": "pytorch_model-00007-of-00010.bin",
189
+ "geov.layers.21.post_attention_layernorm.bias": "pytorch_model-00007-of-00010.bin",
190
+ "geov.layers.21.post_attention_layernorm.weight": "pytorch_model-00007-of-00010.bin",
191
+ "geov.layers.22.attention.causal_mask": "pytorch_model-00007-of-00010.bin",
192
+ "geov.layers.22.attention.dense.weight": "pytorch_model-00007-of-00010.bin",
193
+ "geov.layers.22.attention.qkv.bias": "pytorch_model-00007-of-00010.bin",
194
+ "geov.layers.22.attention.qkv.weight": "pytorch_model-00007-of-00010.bin",
195
+ "geov.layers.22.attention.rotary_emb.inv_freq": "pytorch_model-00007-of-00010.bin",
196
+ "geov.layers.22.input_layernorm.bias": "pytorch_model-00007-of-00010.bin",
197
+ "geov.layers.22.input_layernorm.weight": "pytorch_model-00007-of-00010.bin",
198
+ "geov.layers.22.mlp.dense_2h_to_h.weight": "pytorch_model-00007-of-00010.bin",
199
+ "geov.layers.22.mlp.dense_h_to_4h.bias": "pytorch_model-00007-of-00010.bin",
200
+ "geov.layers.22.mlp.dense_h_to_4h.weight": "pytorch_model-00007-of-00010.bin",
201
+ "geov.layers.22.post_attention_layernorm.bias": "pytorch_model-00007-of-00010.bin",
202
+ "geov.layers.22.post_attention_layernorm.weight": "pytorch_model-00007-of-00010.bin",
203
+ "geov.layers.23.attention.causal_mask": "pytorch_model-00007-of-00010.bin",
204
+ "geov.layers.23.attention.dense.weight": "pytorch_model-00007-of-00010.bin",
205
+ "geov.layers.23.attention.qkv.bias": "pytorch_model-00007-of-00010.bin",
206
+ "geov.layers.23.attention.qkv.weight": "pytorch_model-00007-of-00010.bin",
207
+ "geov.layers.23.attention.rotary_emb.inv_freq": "pytorch_model-00007-of-00010.bin",
208
+ "geov.layers.23.input_layernorm.bias": "pytorch_model-00007-of-00010.bin",
209
+ "geov.layers.23.input_layernorm.weight": "pytorch_model-00007-of-00010.bin",
210
+ "geov.layers.23.mlp.dense_2h_to_h.weight": "pytorch_model-00007-of-00010.bin",
211
+ "geov.layers.23.mlp.dense_h_to_4h.bias": "pytorch_model-00007-of-00010.bin",
212
+ "geov.layers.23.mlp.dense_h_to_4h.weight": "pytorch_model-00007-of-00010.bin",
213
+ "geov.layers.23.post_attention_layernorm.bias": "pytorch_model-00007-of-00010.bin",
214
+ "geov.layers.23.post_attention_layernorm.weight": "pytorch_model-00007-of-00010.bin",
215
+ "geov.layers.24.attention.causal_mask": "pytorch_model-00007-of-00010.bin",
216
+ "geov.layers.24.attention.dense.weight": "pytorch_model-00007-of-00010.bin",
217
+ "geov.layers.24.attention.qkv.bias": "pytorch_model-00007-of-00010.bin",
218
+ "geov.layers.24.attention.qkv.weight": "pytorch_model-00007-of-00010.bin",
219
+ "geov.layers.24.attention.rotary_emb.inv_freq": "pytorch_model-00007-of-00010.bin",
220
+ "geov.layers.24.input_layernorm.bias": "pytorch_model-00007-of-00010.bin",
221
+ "geov.layers.24.input_layernorm.weight": "pytorch_model-00007-of-00010.bin",
222
+ "geov.layers.24.mlp.dense_2h_to_h.weight": "pytorch_model-00008-of-00010.bin",
223
+ "geov.layers.24.mlp.dense_h_to_4h.bias": "pytorch_model-00008-of-00010.bin",
224
+ "geov.layers.24.mlp.dense_h_to_4h.weight": "pytorch_model-00008-of-00010.bin",
225
+ "geov.layers.24.post_attention_layernorm.bias": "pytorch_model-00007-of-00010.bin",
226
+ "geov.layers.24.post_attention_layernorm.weight": "pytorch_model-00007-of-00010.bin",
227
+ "geov.layers.25.attention.causal_mask": "pytorch_model-00008-of-00010.bin",
228
+ "geov.layers.25.attention.dense.weight": "pytorch_model-00008-of-00010.bin",
229
+ "geov.layers.25.attention.qkv.bias": "pytorch_model-00008-of-00010.bin",
230
+ "geov.layers.25.attention.qkv.weight": "pytorch_model-00008-of-00010.bin",
231
+ "geov.layers.25.attention.rotary_emb.inv_freq": "pytorch_model-00008-of-00010.bin",
232
+ "geov.layers.25.input_layernorm.bias": "pytorch_model-00008-of-00010.bin",
233
+ "geov.layers.25.input_layernorm.weight": "pytorch_model-00008-of-00010.bin",
234
+ "geov.layers.25.mlp.dense_2h_to_h.weight": "pytorch_model-00008-of-00010.bin",
235
+ "geov.layers.25.mlp.dense_h_to_4h.bias": "pytorch_model-00008-of-00010.bin",
236
+ "geov.layers.25.mlp.dense_h_to_4h.weight": "pytorch_model-00008-of-00010.bin",
237
+ "geov.layers.25.post_attention_layernorm.bias": "pytorch_model-00008-of-00010.bin",
238
+ "geov.layers.25.post_attention_layernorm.weight": "pytorch_model-00008-of-00010.bin",
239
+ "geov.layers.26.attention.causal_mask": "pytorch_model-00008-of-00010.bin",
240
+ "geov.layers.26.attention.dense.weight": "pytorch_model-00008-of-00010.bin",
241
+ "geov.layers.26.attention.qkv.bias": "pytorch_model-00008-of-00010.bin",
242
+ "geov.layers.26.attention.qkv.weight": "pytorch_model-00008-of-00010.bin",
243
+ "geov.layers.26.attention.rotary_emb.inv_freq": "pytorch_model-00008-of-00010.bin",
244
+ "geov.layers.26.input_layernorm.bias": "pytorch_model-00008-of-00010.bin",
245
+ "geov.layers.26.input_layernorm.weight": "pytorch_model-00008-of-00010.bin",
246
+ "geov.layers.26.mlp.dense_2h_to_h.weight": "pytorch_model-00008-of-00010.bin",
247
+ "geov.layers.26.mlp.dense_h_to_4h.bias": "pytorch_model-00008-of-00010.bin",
248
+ "geov.layers.26.mlp.dense_h_to_4h.weight": "pytorch_model-00008-of-00010.bin",
249
+ "geov.layers.26.post_attention_layernorm.bias": "pytorch_model-00008-of-00010.bin",
250
+ "geov.layers.26.post_attention_layernorm.weight": "pytorch_model-00008-of-00010.bin",
251
+ "geov.layers.27.attention.causal_mask": "pytorch_model-00008-of-00010.bin",
252
+ "geov.layers.27.attention.dense.weight": "pytorch_model-00008-of-00010.bin",
253
+ "geov.layers.27.attention.qkv.bias": "pytorch_model-00008-of-00010.bin",
254
+ "geov.layers.27.attention.qkv.weight": "pytorch_model-00008-of-00010.bin",
255
+ "geov.layers.27.attention.rotary_emb.inv_freq": "pytorch_model-00008-of-00010.bin",
256
+ "geov.layers.27.input_layernorm.bias": "pytorch_model-00008-of-00010.bin",
257
+ "geov.layers.27.input_layernorm.weight": "pytorch_model-00008-of-00010.bin",
258
+ "geov.layers.27.mlp.dense_2h_to_h.weight": "pytorch_model-00008-of-00010.bin",
259
+ "geov.layers.27.mlp.dense_h_to_4h.bias": "pytorch_model-00008-of-00010.bin",
260
+ "geov.layers.27.mlp.dense_h_to_4h.weight": "pytorch_model-00008-of-00010.bin",
261
+ "geov.layers.27.post_attention_layernorm.bias": "pytorch_model-00008-of-00010.bin",
262
+ "geov.layers.27.post_attention_layernorm.weight": "pytorch_model-00008-of-00010.bin",
263
+ "geov.layers.28.attention.causal_mask": "pytorch_model-00008-of-00010.bin",
264
+ "geov.layers.28.attention.dense.weight": "pytorch_model-00009-of-00010.bin",
265
+ "geov.layers.28.attention.qkv.bias": "pytorch_model-00009-of-00010.bin",
266
+ "geov.layers.28.attention.qkv.weight": "pytorch_model-00009-of-00010.bin",
267
+ "geov.layers.28.attention.rotary_emb.inv_freq": "pytorch_model-00008-of-00010.bin",
268
+ "geov.layers.28.input_layernorm.bias": "pytorch_model-00008-of-00010.bin",
269
+ "geov.layers.28.input_layernorm.weight": "pytorch_model-00008-of-00010.bin",
270
+ "geov.layers.28.mlp.dense_2h_to_h.weight": "pytorch_model-00009-of-00010.bin",
271
+ "geov.layers.28.mlp.dense_h_to_4h.bias": "pytorch_model-00009-of-00010.bin",
272
+ "geov.layers.28.mlp.dense_h_to_4h.weight": "pytorch_model-00009-of-00010.bin",
273
+ "geov.layers.28.post_attention_layernorm.bias": "pytorch_model-00008-of-00010.bin",
274
+ "geov.layers.28.post_attention_layernorm.weight": "pytorch_model-00008-of-00010.bin",
275
+ "geov.layers.29.attention.causal_mask": "pytorch_model-00009-of-00010.bin",
276
+ "geov.layers.29.attention.dense.weight": "pytorch_model-00009-of-00010.bin",
277
+ "geov.layers.29.attention.qkv.bias": "pytorch_model-00009-of-00010.bin",
278
+ "geov.layers.29.attention.qkv.weight": "pytorch_model-00009-of-00010.bin",
279
+ "geov.layers.29.attention.rotary_emb.inv_freq": "pytorch_model-00009-of-00010.bin",
280
+ "geov.layers.29.input_layernorm.bias": "pytorch_model-00009-of-00010.bin",
281
+ "geov.layers.29.input_layernorm.weight": "pytorch_model-00009-of-00010.bin",
282
+ "geov.layers.29.mlp.dense_2h_to_h.weight": "pytorch_model-00009-of-00010.bin",
283
+ "geov.layers.29.mlp.dense_h_to_4h.bias": "pytorch_model-00009-of-00010.bin",
284
+ "geov.layers.29.mlp.dense_h_to_4h.weight": "pytorch_model-00009-of-00010.bin",
285
+ "geov.layers.29.post_attention_layernorm.bias": "pytorch_model-00009-of-00010.bin",
286
+ "geov.layers.29.post_attention_layernorm.weight": "pytorch_model-00009-of-00010.bin",
287
+ "geov.layers.3.attention.causal_mask": "pytorch_model-00002-of-00010.bin",
288
+ "geov.layers.3.attention.dense.weight": "pytorch_model-00002-of-00010.bin",
289
+ "geov.layers.3.attention.qkv.bias": "pytorch_model-00002-of-00010.bin",
290
+ "geov.layers.3.attention.qkv.weight": "pytorch_model-00002-of-00010.bin",
291
+ "geov.layers.3.attention.rotary_emb.inv_freq": "pytorch_model-00002-of-00010.bin",
292
+ "geov.layers.3.input_layernorm.bias": "pytorch_model-00002-of-00010.bin",
293
+ "geov.layers.3.input_layernorm.weight": "pytorch_model-00002-of-00010.bin",
294
+ "geov.layers.3.mlp.dense_2h_to_h.weight": "pytorch_model-00002-of-00010.bin",
295
+ "geov.layers.3.mlp.dense_h_to_4h.bias": "pytorch_model-00002-of-00010.bin",
296
+ "geov.layers.3.mlp.dense_h_to_4h.weight": "pytorch_model-00002-of-00010.bin",
297
+ "geov.layers.3.post_attention_layernorm.bias": "pytorch_model-00002-of-00010.bin",
298
+ "geov.layers.3.post_attention_layernorm.weight": "pytorch_model-00002-of-00010.bin",
299
+ "geov.layers.30.attention.causal_mask": "pytorch_model-00009-of-00010.bin",
300
+ "geov.layers.30.attention.dense.weight": "pytorch_model-00009-of-00010.bin",
301
+ "geov.layers.30.attention.qkv.bias": "pytorch_model-00009-of-00010.bin",
302
+ "geov.layers.30.attention.qkv.weight": "pytorch_model-00009-of-00010.bin",
303
+ "geov.layers.30.attention.rotary_emb.inv_freq": "pytorch_model-00009-of-00010.bin",
304
+ "geov.layers.30.input_layernorm.bias": "pytorch_model-00009-of-00010.bin",
305
+ "geov.layers.30.input_layernorm.weight": "pytorch_model-00009-of-00010.bin",
306
+ "geov.layers.30.mlp.dense_2h_to_h.weight": "pytorch_model-00009-of-00010.bin",
307
+ "geov.layers.30.mlp.dense_h_to_4h.bias": "pytorch_model-00009-of-00010.bin",
308
+ "geov.layers.30.mlp.dense_h_to_4h.weight": "pytorch_model-00009-of-00010.bin",
309
+ "geov.layers.30.post_attention_layernorm.bias": "pytorch_model-00009-of-00010.bin",
310
+ "geov.layers.30.post_attention_layernorm.weight": "pytorch_model-00009-of-00010.bin",
311
+ "geov.layers.31.attention.causal_mask": "pytorch_model-00009-of-00010.bin",
312
+ "geov.layers.31.attention.dense.weight": "pytorch_model-00009-of-00010.bin",
313
+ "geov.layers.31.attention.qkv.bias": "pytorch_model-00009-of-00010.bin",
314
+ "geov.layers.31.attention.qkv.weight": "pytorch_model-00009-of-00010.bin",
315
+ "geov.layers.31.attention.rotary_emb.inv_freq": "pytorch_model-00009-of-00010.bin",
316
+ "geov.layers.31.input_layernorm.bias": "pytorch_model-00009-of-00010.bin",
317
+ "geov.layers.31.input_layernorm.weight": "pytorch_model-00009-of-00010.bin",
318
+ "geov.layers.31.mlp.dense_2h_to_h.weight": "pytorch_model-00010-of-00010.bin",
319
+ "geov.layers.31.mlp.dense_h_to_4h.bias": "pytorch_model-00009-of-00010.bin",
320
+ "geov.layers.31.mlp.dense_h_to_4h.weight": "pytorch_model-00009-of-00010.bin",
321
+ "geov.layers.31.post_attention_layernorm.bias": "pytorch_model-00009-of-00010.bin",
322
+ "geov.layers.31.post_attention_layernorm.weight": "pytorch_model-00009-of-00010.bin",
323
+ "geov.layers.4.attention.causal_mask": "pytorch_model-00002-of-00010.bin",
324
+ "geov.layers.4.attention.dense.weight": "pytorch_model-00002-of-00010.bin",
325
+ "geov.layers.4.attention.qkv.bias": "pytorch_model-00002-of-00010.bin",
326
+ "geov.layers.4.attention.qkv.weight": "pytorch_model-00002-of-00010.bin",
327
+ "geov.layers.4.attention.rotary_emb.inv_freq": "pytorch_model-00002-of-00010.bin",
328
+ "geov.layers.4.input_layernorm.bias": "pytorch_model-00002-of-00010.bin",
329
+ "geov.layers.4.input_layernorm.weight": "pytorch_model-00002-of-00010.bin",
330
+ "geov.layers.4.mlp.dense_2h_to_h.weight": "pytorch_model-00002-of-00010.bin",
331
+ "geov.layers.4.mlp.dense_h_to_4h.bias": "pytorch_model-00002-of-00010.bin",
332
+ "geov.layers.4.mlp.dense_h_to_4h.weight": "pytorch_model-00002-of-00010.bin",
333
+ "geov.layers.4.post_attention_layernorm.bias": "pytorch_model-00002-of-00010.bin",
334
+ "geov.layers.4.post_attention_layernorm.weight": "pytorch_model-00002-of-00010.bin",
335
+ "geov.layers.5.attention.causal_mask": "pytorch_model-00002-of-00010.bin",
336
+ "geov.layers.5.attention.dense.weight": "pytorch_model-00002-of-00010.bin",
337
+ "geov.layers.5.attention.qkv.bias": "pytorch_model-00002-of-00010.bin",
338
+ "geov.layers.5.attention.qkv.weight": "pytorch_model-00002-of-00010.bin",
339
+ "geov.layers.5.attention.rotary_emb.inv_freq": "pytorch_model-00002-of-00010.bin",
340
+ "geov.layers.5.input_layernorm.bias": "pytorch_model-00002-of-00010.bin",
341
+ "geov.layers.5.input_layernorm.weight": "pytorch_model-00002-of-00010.bin",
342
+ "geov.layers.5.mlp.dense_2h_to_h.weight": "pytorch_model-00002-of-00010.bin",
343
+ "geov.layers.5.mlp.dense_h_to_4h.bias": "pytorch_model-00002-of-00010.bin",
344
+ "geov.layers.5.mlp.dense_h_to_4h.weight": "pytorch_model-00002-of-00010.bin",
345
+ "geov.layers.5.post_attention_layernorm.bias": "pytorch_model-00002-of-00010.bin",
346
+ "geov.layers.5.post_attention_layernorm.weight": "pytorch_model-00002-of-00010.bin",
347
+ "geov.layers.6.attention.causal_mask": "pytorch_model-00002-of-00010.bin",
348
+ "geov.layers.6.attention.dense.weight": "pytorch_model-00003-of-00010.bin",
349
+ "geov.layers.6.attention.qkv.bias": "pytorch_model-00003-of-00010.bin",
350
+ "geov.layers.6.attention.qkv.weight": "pytorch_model-00003-of-00010.bin",
351
+ "geov.layers.6.attention.rotary_emb.inv_freq": "pytorch_model-00002-of-00010.bin",
352
+ "geov.layers.6.input_layernorm.bias": "pytorch_model-00002-of-00010.bin",
353
+ "geov.layers.6.input_layernorm.weight": "pytorch_model-00002-of-00010.bin",
354
+ "geov.layers.6.mlp.dense_2h_to_h.weight": "pytorch_model-00003-of-00010.bin",
355
+ "geov.layers.6.mlp.dense_h_to_4h.bias": "pytorch_model-00003-of-00010.bin",
356
+ "geov.layers.6.mlp.dense_h_to_4h.weight": "pytorch_model-00003-of-00010.bin",
357
+ "geov.layers.6.post_attention_layernorm.bias": "pytorch_model-00002-of-00010.bin",
358
+ "geov.layers.6.post_attention_layernorm.weight": "pytorch_model-00002-of-00010.bin",
359
+ "geov.layers.7.attention.causal_mask": "pytorch_model-00003-of-00010.bin",
360
+ "geov.layers.7.attention.dense.weight": "pytorch_model-00003-of-00010.bin",
361
+ "geov.layers.7.attention.qkv.bias": "pytorch_model-00003-of-00010.bin",
362
+ "geov.layers.7.attention.qkv.weight": "pytorch_model-00003-of-00010.bin",
363
+ "geov.layers.7.attention.rotary_emb.inv_freq": "pytorch_model-00003-of-00010.bin",
364
+ "geov.layers.7.input_layernorm.bias": "pytorch_model-00003-of-00010.bin",
365
+ "geov.layers.7.input_layernorm.weight": "pytorch_model-00003-of-00010.bin",
366
+ "geov.layers.7.mlp.dense_2h_to_h.weight": "pytorch_model-00003-of-00010.bin",
367
+ "geov.layers.7.mlp.dense_h_to_4h.bias": "pytorch_model-00003-of-00010.bin",
368
+ "geov.layers.7.mlp.dense_h_to_4h.weight": "pytorch_model-00003-of-00010.bin",
369
+ "geov.layers.7.post_attention_layernorm.bias": "pytorch_model-00003-of-00010.bin",
370
+ "geov.layers.7.post_attention_layernorm.weight": "pytorch_model-00003-of-00010.bin",
371
+ "geov.layers.8.attention.causal_mask": "pytorch_model-00003-of-00010.bin",
372
+ "geov.layers.8.attention.dense.weight": "pytorch_model-00003-of-00010.bin",
373
+ "geov.layers.8.attention.qkv.bias": "pytorch_model-00003-of-00010.bin",
374
+ "geov.layers.8.attention.qkv.weight": "pytorch_model-00003-of-00010.bin",
375
+ "geov.layers.8.attention.rotary_emb.inv_freq": "pytorch_model-00003-of-00010.bin",
376
+ "geov.layers.8.input_layernorm.bias": "pytorch_model-00003-of-00010.bin",
377
+ "geov.layers.8.input_layernorm.weight": "pytorch_model-00003-of-00010.bin",
378
+ "geov.layers.8.mlp.dense_2h_to_h.weight": "pytorch_model-00003-of-00010.bin",
379
+ "geov.layers.8.mlp.dense_h_to_4h.bias": "pytorch_model-00003-of-00010.bin",
380
+ "geov.layers.8.mlp.dense_h_to_4h.weight": "pytorch_model-00003-of-00010.bin",
381
+ "geov.layers.8.post_attention_layernorm.bias": "pytorch_model-00003-of-00010.bin",
382
+ "geov.layers.8.post_attention_layernorm.weight": "pytorch_model-00003-of-00010.bin",
383
+ "geov.layers.9.attention.causal_mask": "pytorch_model-00003-of-00010.bin",
384
+ "geov.layers.9.attention.dense.weight": "pytorch_model-00003-of-00010.bin",
385
+ "geov.layers.9.attention.qkv.bias": "pytorch_model-00003-of-00010.bin",
386
+ "geov.layers.9.attention.qkv.weight": "pytorch_model-00003-of-00010.bin",
387
+ "geov.layers.9.attention.rotary_emb.inv_freq": "pytorch_model-00003-of-00010.bin",
388
+ "geov.layers.9.input_layernorm.bias": "pytorch_model-00003-of-00010.bin",
389
+ "geov.layers.9.input_layernorm.weight": "pytorch_model-00003-of-00010.bin",
390
+ "geov.layers.9.mlp.dense_2h_to_h.weight": "pytorch_model-00004-of-00010.bin",
391
+ "geov.layers.9.mlp.dense_h_to_4h.bias": "pytorch_model-00003-of-00010.bin",
392
+ "geov.layers.9.mlp.dense_h_to_4h.weight": "pytorch_model-00003-of-00010.bin",
393
+ "geov.layers.9.post_attention_layernorm.bias": "pytorch_model-00003-of-00010.bin",
394
+ "geov.layers.9.post_attention_layernorm.weight": "pytorch_model-00003-of-00010.bin"
395
+ }
396
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "eos_token": "</s>",
4
+ "unk_token": "<unk>"
5
+ }
spiece.model ADDED
@@ -0,0 +1,3 @@
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:42005475648603fbcb46bf8d08f4ade44c8546c6b37f8048bb5333b7b725ab85
3
+ size 1269630
tokenizer_config.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": "<s>",
3
+ "eos_token": "</s>",
4
+ "model_max_length": 1000000000000000019884624838656,
5
+ "new_line_token_id": 65499,
6
+ "tokenizer_class": "GeoVTokenizer",
7
+ "unk_token": "<unk>"
8
+ }