divergente commited on
Commit
b1fc15b
1 Parent(s): c88025a

Upload 16 files

Browse files
config.json ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "huggyllama/llama-7b",
3
+ "architectures": [
4
+ "LlamaModel"
5
+ ],
6
+ "bos_token_id": 1,
7
+ "eos_token_id": 2,
8
+ "hidden_act": "silu",
9
+ "hidden_size": 4096,
10
+ "initializer_range": 0.02,
11
+ "intermediate_size": 11008,
12
+ "max_position_embeddings": 2048,
13
+ "max_sequence_length": 2048,
14
+ "model_type": "llama",
15
+ "num_attention_heads": 32,
16
+ "num_hidden_layers": 32,
17
+ "num_key_value_heads": 32,
18
+ "pad_token_id": 0,
19
+ "pretraining_tp": 1,
20
+ "rms_norm_eps": 1e-06,
21
+ "rope_scaling": null,
22
+ "tie_word_embeddings": false,
23
+ "torch_dtype": "float16",
24
+ "transformers_version": "4.31.0",
25
+ "use_cache": true,
26
+ "vocab_size": 32000
27
+ }
pytorch_model-00001-of-00014.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5539f5facbfc043c8dc60c690352aa8d5093a5afe0538f31bddf9df73f04124f
3
+ size 981490195
pytorch_model-00002-of-00014.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c4d782d8fc5587faf09c7e357d1bf63b121678397579c6b7aa653a5a7bc5e0e8
3
+ size 966845009
pytorch_model-00003-of-00014.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:46e830c33f091372f4bc5bfec15d33f0a27006994da9ff744c4973c171688913
3
+ size 966828607
pytorch_model-00004-of-00014.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e06fb72dc85303186f6cefd7d68d3ca5521f205fa3924cbd693ff1c277f4deab
3
+ size 989913407
pytorch_model-00005-of-00014.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b56231935c3fe805acac145646bbe55e0b97ed92cb3b31717e4e0b73b2255549
3
+ size 943760273
pytorch_model-00006-of-00014.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1a9833099e10b559d13152092f437d5aeae4751ac3c21db23687b632c5d7adc8
3
+ size 989896411
pytorch_model-00007-of-00014.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e22fb725b5debbb0678bd5d19960da2f3f3e9f86696bde550745990192663be9
3
+ size 966845073
pytorch_model-00008-of-00014.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5cce7f5be733d982b1822244650e147d724ff5a6fde656f2951737b0c1e833f0
3
+ size 966828607
pytorch_model-00009-of-00014.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:91ad105c1970ed3345419699a2a1a4d03846c85cf39022e65117b9dfcc07bc4a
3
+ size 989913407
pytorch_model-00010-of-00014.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:62e0b472e2b68fcdba86cfb0599345e909ee9fc69829f3f5049f2bfe0431d77f
3
+ size 943760273
pytorch_model-00011-of-00014.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1096de7e3af3df40de31faaaf4c6ce015d9a407c010f94ae89ce3a3852907f8
3
+ size 989896411
pytorch_model-00012-of-00014.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:453b25707d912d7fe154866728b4e8fecc01dcddee658c5fff27d3a4ba7742f1
3
+ size 966845073
pytorch_model-00013-of-00014.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:79344d2d2f005cd274a88d13ffa1442955af02b87f19b8548cedb7f8e2ff4027
3
+ size 966828607
pytorch_model-00014-of-00014.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8d3bbf333e649cb48f3f088ca6024a4c3e62d27dfd215c70b761f617a5bc08e4
3
+ size 585151837
pytorch_model.bin.index.json ADDED
@@ -0,0 +1,329 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 13214695424
4
+ },
5
+ "weight_map": {
6
+ "embed_tokens.weight": "pytorch_model-00001-of-00014.bin",
7
+ "layers.0.input_layernorm.weight": "pytorch_model-00001-of-00014.bin",
8
+ "layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00014.bin",
9
+ "layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00014.bin",
10
+ "layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00014.bin",
11
+ "layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00014.bin",
12
+ "layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00014.bin",
13
+ "layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00014.bin",
14
+ "layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00014.bin",
15
+ "layers.0.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00014.bin",
16
+ "layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00014.bin",
17
+ "layers.1.input_layernorm.weight": "pytorch_model-00002-of-00014.bin",
18
+ "layers.1.mlp.down_proj.weight": "pytorch_model-00002-of-00014.bin",
19
+ "layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00014.bin",
20
+ "layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00014.bin",
21
+ "layers.1.post_attention_layernorm.weight": "pytorch_model-00002-of-00014.bin",
22
+ "layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00014.bin",
23
+ "layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00014.bin",
24
+ "layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00014.bin",
25
+ "layers.1.self_attn.rotary_emb.inv_freq": "pytorch_model-00001-of-00014.bin",
26
+ "layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00014.bin",
27
+ "layers.10.input_layernorm.weight": "pytorch_model-00005-of-00014.bin",
28
+ "layers.10.mlp.down_proj.weight": "pytorch_model-00005-of-00014.bin",
29
+ "layers.10.mlp.gate_proj.weight": "pytorch_model-00005-of-00014.bin",
30
+ "layers.10.mlp.up_proj.weight": "pytorch_model-00005-of-00014.bin",
31
+ "layers.10.post_attention_layernorm.weight": "pytorch_model-00005-of-00014.bin",
32
+ "layers.10.self_attn.k_proj.weight": "pytorch_model-00005-of-00014.bin",
33
+ "layers.10.self_attn.o_proj.weight": "pytorch_model-00005-of-00014.bin",
34
+ "layers.10.self_attn.q_proj.weight": "pytorch_model-00005-of-00014.bin",
35
+ "layers.10.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00014.bin",
36
+ "layers.10.self_attn.v_proj.weight": "pytorch_model-00005-of-00014.bin",
37
+ "layers.11.input_layernorm.weight": "pytorch_model-00006-of-00014.bin",
38
+ "layers.11.mlp.down_proj.weight": "pytorch_model-00006-of-00014.bin",
39
+ "layers.11.mlp.gate_proj.weight": "pytorch_model-00006-of-00014.bin",
40
+ "layers.11.mlp.up_proj.weight": "pytorch_model-00006-of-00014.bin",
41
+ "layers.11.post_attention_layernorm.weight": "pytorch_model-00006-of-00014.bin",
42
+ "layers.11.self_attn.k_proj.weight": "pytorch_model-00005-of-00014.bin",
43
+ "layers.11.self_attn.o_proj.weight": "pytorch_model-00005-of-00014.bin",
44
+ "layers.11.self_attn.q_proj.weight": "pytorch_model-00005-of-00014.bin",
45
+ "layers.11.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00014.bin",
46
+ "layers.11.self_attn.v_proj.weight": "pytorch_model-00005-of-00014.bin",
47
+ "layers.12.input_layernorm.weight": "pytorch_model-00006-of-00014.bin",
48
+ "layers.12.mlp.down_proj.weight": "pytorch_model-00006-of-00014.bin",
49
+ "layers.12.mlp.gate_proj.weight": "pytorch_model-00006-of-00014.bin",
50
+ "layers.12.mlp.up_proj.weight": "pytorch_model-00006-of-00014.bin",
51
+ "layers.12.post_attention_layernorm.weight": "pytorch_model-00006-of-00014.bin",
52
+ "layers.12.self_attn.k_proj.weight": "pytorch_model-00006-of-00014.bin",
53
+ "layers.12.self_attn.o_proj.weight": "pytorch_model-00006-of-00014.bin",
54
+ "layers.12.self_attn.q_proj.weight": "pytorch_model-00006-of-00014.bin",
55
+ "layers.12.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00014.bin",
56
+ "layers.12.self_attn.v_proj.weight": "pytorch_model-00006-of-00014.bin",
57
+ "layers.13.input_layernorm.weight": "pytorch_model-00007-of-00014.bin",
58
+ "layers.13.mlp.down_proj.weight": "pytorch_model-00007-of-00014.bin",
59
+ "layers.13.mlp.gate_proj.weight": "pytorch_model-00006-of-00014.bin",
60
+ "layers.13.mlp.up_proj.weight": "pytorch_model-00006-of-00014.bin",
61
+ "layers.13.post_attention_layernorm.weight": "pytorch_model-00007-of-00014.bin",
62
+ "layers.13.self_attn.k_proj.weight": "pytorch_model-00006-of-00014.bin",
63
+ "layers.13.self_attn.o_proj.weight": "pytorch_model-00006-of-00014.bin",
64
+ "layers.13.self_attn.q_proj.weight": "pytorch_model-00006-of-00014.bin",
65
+ "layers.13.self_attn.rotary_emb.inv_freq": "pytorch_model-00006-of-00014.bin",
66
+ "layers.13.self_attn.v_proj.weight": "pytorch_model-00006-of-00014.bin",
67
+ "layers.14.input_layernorm.weight": "pytorch_model-00007-of-00014.bin",
68
+ "layers.14.mlp.down_proj.weight": "pytorch_model-00007-of-00014.bin",
69
+ "layers.14.mlp.gate_proj.weight": "pytorch_model-00007-of-00014.bin",
70
+ "layers.14.mlp.up_proj.weight": "pytorch_model-00007-of-00014.bin",
71
+ "layers.14.post_attention_layernorm.weight": "pytorch_model-00007-of-00014.bin",
72
+ "layers.14.self_attn.k_proj.weight": "pytorch_model-00007-of-00014.bin",
73
+ "layers.14.self_attn.o_proj.weight": "pytorch_model-00007-of-00014.bin",
74
+ "layers.14.self_attn.q_proj.weight": "pytorch_model-00007-of-00014.bin",
75
+ "layers.14.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00014.bin",
76
+ "layers.14.self_attn.v_proj.weight": "pytorch_model-00007-of-00014.bin",
77
+ "layers.15.input_layernorm.weight": "pytorch_model-00007-of-00014.bin",
78
+ "layers.15.mlp.down_proj.weight": "pytorch_model-00007-of-00014.bin",
79
+ "layers.15.mlp.gate_proj.weight": "pytorch_model-00007-of-00014.bin",
80
+ "layers.15.mlp.up_proj.weight": "pytorch_model-00007-of-00014.bin",
81
+ "layers.15.post_attention_layernorm.weight": "pytorch_model-00007-of-00014.bin",
82
+ "layers.15.self_attn.k_proj.weight": "pytorch_model-00007-of-00014.bin",
83
+ "layers.15.self_attn.o_proj.weight": "pytorch_model-00007-of-00014.bin",
84
+ "layers.15.self_attn.q_proj.weight": "pytorch_model-00007-of-00014.bin",
85
+ "layers.15.self_attn.rotary_emb.inv_freq": "pytorch_model-00007-of-00014.bin",
86
+ "layers.15.self_attn.v_proj.weight": "pytorch_model-00007-of-00014.bin",
87
+ "layers.16.input_layernorm.weight": "pytorch_model-00008-of-00014.bin",
88
+ "layers.16.mlp.down_proj.weight": "pytorch_model-00008-of-00014.bin",
89
+ "layers.16.mlp.gate_proj.weight": "pytorch_model-00008-of-00014.bin",
90
+ "layers.16.mlp.up_proj.weight": "pytorch_model-00008-of-00014.bin",
91
+ "layers.16.post_attention_layernorm.weight": "pytorch_model-00008-of-00014.bin",
92
+ "layers.16.self_attn.k_proj.weight": "pytorch_model-00007-of-00014.bin",
93
+ "layers.16.self_attn.o_proj.weight": "pytorch_model-00008-of-00014.bin",
94
+ "layers.16.self_attn.q_proj.weight": "pytorch_model-00007-of-00014.bin",
95
+ "layers.16.self_attn.rotary_emb.inv_freq": "pytorch_model-00008-of-00014.bin",
96
+ "layers.16.self_attn.v_proj.weight": "pytorch_model-00008-of-00014.bin",
97
+ "layers.17.input_layernorm.weight": "pytorch_model-00008-of-00014.bin",
98
+ "layers.17.mlp.down_proj.weight": "pytorch_model-00008-of-00014.bin",
99
+ "layers.17.mlp.gate_proj.weight": "pytorch_model-00008-of-00014.bin",
100
+ "layers.17.mlp.up_proj.weight": "pytorch_model-00008-of-00014.bin",
101
+ "layers.17.post_attention_layernorm.weight": "pytorch_model-00008-of-00014.bin",
102
+ "layers.17.self_attn.k_proj.weight": "pytorch_model-00008-of-00014.bin",
103
+ "layers.17.self_attn.o_proj.weight": "pytorch_model-00008-of-00014.bin",
104
+ "layers.17.self_attn.q_proj.weight": "pytorch_model-00008-of-00014.bin",
105
+ "layers.17.self_attn.rotary_emb.inv_freq": "pytorch_model-00008-of-00014.bin",
106
+ "layers.17.self_attn.v_proj.weight": "pytorch_model-00008-of-00014.bin",
107
+ "layers.18.input_layernorm.weight": "pytorch_model-00009-of-00014.bin",
108
+ "layers.18.mlp.down_proj.weight": "pytorch_model-00009-of-00014.bin",
109
+ "layers.18.mlp.gate_proj.weight": "pytorch_model-00008-of-00014.bin",
110
+ "layers.18.mlp.up_proj.weight": "pytorch_model-00009-of-00014.bin",
111
+ "layers.18.post_attention_layernorm.weight": "pytorch_model-00009-of-00014.bin",
112
+ "layers.18.self_attn.k_proj.weight": "pytorch_model-00008-of-00014.bin",
113
+ "layers.18.self_attn.o_proj.weight": "pytorch_model-00008-of-00014.bin",
114
+ "layers.18.self_attn.q_proj.weight": "pytorch_model-00008-of-00014.bin",
115
+ "layers.18.self_attn.rotary_emb.inv_freq": "pytorch_model-00008-of-00014.bin",
116
+ "layers.18.self_attn.v_proj.weight": "pytorch_model-00008-of-00014.bin",
117
+ "layers.19.input_layernorm.weight": "pytorch_model-00009-of-00014.bin",
118
+ "layers.19.mlp.down_proj.weight": "pytorch_model-00009-of-00014.bin",
119
+ "layers.19.mlp.gate_proj.weight": "pytorch_model-00009-of-00014.bin",
120
+ "layers.19.mlp.up_proj.weight": "pytorch_model-00009-of-00014.bin",
121
+ "layers.19.post_attention_layernorm.weight": "pytorch_model-00009-of-00014.bin",
122
+ "layers.19.self_attn.k_proj.weight": "pytorch_model-00009-of-00014.bin",
123
+ "layers.19.self_attn.o_proj.weight": "pytorch_model-00009-of-00014.bin",
124
+ "layers.19.self_attn.q_proj.weight": "pytorch_model-00009-of-00014.bin",
125
+ "layers.19.self_attn.rotary_emb.inv_freq": "pytorch_model-00009-of-00014.bin",
126
+ "layers.19.self_attn.v_proj.weight": "pytorch_model-00009-of-00014.bin",
127
+ "layers.2.input_layernorm.weight": "pytorch_model-00002-of-00014.bin",
128
+ "layers.2.mlp.down_proj.weight": "pytorch_model-00002-of-00014.bin",
129
+ "layers.2.mlp.gate_proj.weight": "pytorch_model-00002-of-00014.bin",
130
+ "layers.2.mlp.up_proj.weight": "pytorch_model-00002-of-00014.bin",
131
+ "layers.2.post_attention_layernorm.weight": "pytorch_model-00002-of-00014.bin",
132
+ "layers.2.self_attn.k_proj.weight": "pytorch_model-00002-of-00014.bin",
133
+ "layers.2.self_attn.o_proj.weight": "pytorch_model-00002-of-00014.bin",
134
+ "layers.2.self_attn.q_proj.weight": "pytorch_model-00002-of-00014.bin",
135
+ "layers.2.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00014.bin",
136
+ "layers.2.self_attn.v_proj.weight": "pytorch_model-00002-of-00014.bin",
137
+ "layers.20.input_layernorm.weight": "pytorch_model-00009-of-00014.bin",
138
+ "layers.20.mlp.down_proj.weight": "pytorch_model-00009-of-00014.bin",
139
+ "layers.20.mlp.gate_proj.weight": "pytorch_model-00009-of-00014.bin",
140
+ "layers.20.mlp.up_proj.weight": "pytorch_model-00009-of-00014.bin",
141
+ "layers.20.post_attention_layernorm.weight": "pytorch_model-00009-of-00014.bin",
142
+ "layers.20.self_attn.k_proj.weight": "pytorch_model-00009-of-00014.bin",
143
+ "layers.20.self_attn.o_proj.weight": "pytorch_model-00009-of-00014.bin",
144
+ "layers.20.self_attn.q_proj.weight": "pytorch_model-00009-of-00014.bin",
145
+ "layers.20.self_attn.rotary_emb.inv_freq": "pytorch_model-00009-of-00014.bin",
146
+ "layers.20.self_attn.v_proj.weight": "pytorch_model-00009-of-00014.bin",
147
+ "layers.21.input_layernorm.weight": "pytorch_model-00010-of-00014.bin",
148
+ "layers.21.mlp.down_proj.weight": "pytorch_model-00010-of-00014.bin",
149
+ "layers.21.mlp.gate_proj.weight": "pytorch_model-00010-of-00014.bin",
150
+ "layers.21.mlp.up_proj.weight": "pytorch_model-00010-of-00014.bin",
151
+ "layers.21.post_attention_layernorm.weight": "pytorch_model-00010-of-00014.bin",
152
+ "layers.21.self_attn.k_proj.weight": "pytorch_model-00010-of-00014.bin",
153
+ "layers.21.self_attn.o_proj.weight": "pytorch_model-00010-of-00014.bin",
154
+ "layers.21.self_attn.q_proj.weight": "pytorch_model-00010-of-00014.bin",
155
+ "layers.21.self_attn.rotary_emb.inv_freq": "pytorch_model-00010-of-00014.bin",
156
+ "layers.21.self_attn.v_proj.weight": "pytorch_model-00010-of-00014.bin",
157
+ "layers.22.input_layernorm.weight": "pytorch_model-00010-of-00014.bin",
158
+ "layers.22.mlp.down_proj.weight": "pytorch_model-00010-of-00014.bin",
159
+ "layers.22.mlp.gate_proj.weight": "pytorch_model-00010-of-00014.bin",
160
+ "layers.22.mlp.up_proj.weight": "pytorch_model-00010-of-00014.bin",
161
+ "layers.22.post_attention_layernorm.weight": "pytorch_model-00010-of-00014.bin",
162
+ "layers.22.self_attn.k_proj.weight": "pytorch_model-00010-of-00014.bin",
163
+ "layers.22.self_attn.o_proj.weight": "pytorch_model-00010-of-00014.bin",
164
+ "layers.22.self_attn.q_proj.weight": "pytorch_model-00010-of-00014.bin",
165
+ "layers.22.self_attn.rotary_emb.inv_freq": "pytorch_model-00010-of-00014.bin",
166
+ "layers.22.self_attn.v_proj.weight": "pytorch_model-00010-of-00014.bin",
167
+ "layers.23.input_layernorm.weight": "pytorch_model-00011-of-00014.bin",
168
+ "layers.23.mlp.down_proj.weight": "pytorch_model-00011-of-00014.bin",
169
+ "layers.23.mlp.gate_proj.weight": "pytorch_model-00011-of-00014.bin",
170
+ "layers.23.mlp.up_proj.weight": "pytorch_model-00011-of-00014.bin",
171
+ "layers.23.post_attention_layernorm.weight": "pytorch_model-00011-of-00014.bin",
172
+ "layers.23.self_attn.k_proj.weight": "pytorch_model-00010-of-00014.bin",
173
+ "layers.23.self_attn.o_proj.weight": "pytorch_model-00010-of-00014.bin",
174
+ "layers.23.self_attn.q_proj.weight": "pytorch_model-00010-of-00014.bin",
175
+ "layers.23.self_attn.rotary_emb.inv_freq": "pytorch_model-00010-of-00014.bin",
176
+ "layers.23.self_attn.v_proj.weight": "pytorch_model-00010-of-00014.bin",
177
+ "layers.24.input_layernorm.weight": "pytorch_model-00011-of-00014.bin",
178
+ "layers.24.mlp.down_proj.weight": "pytorch_model-00011-of-00014.bin",
179
+ "layers.24.mlp.gate_proj.weight": "pytorch_model-00011-of-00014.bin",
180
+ "layers.24.mlp.up_proj.weight": "pytorch_model-00011-of-00014.bin",
181
+ "layers.24.post_attention_layernorm.weight": "pytorch_model-00011-of-00014.bin",
182
+ "layers.24.self_attn.k_proj.weight": "pytorch_model-00011-of-00014.bin",
183
+ "layers.24.self_attn.o_proj.weight": "pytorch_model-00011-of-00014.bin",
184
+ "layers.24.self_attn.q_proj.weight": "pytorch_model-00011-of-00014.bin",
185
+ "layers.24.self_attn.rotary_emb.inv_freq": "pytorch_model-00011-of-00014.bin",
186
+ "layers.24.self_attn.v_proj.weight": "pytorch_model-00011-of-00014.bin",
187
+ "layers.25.input_layernorm.weight": "pytorch_model-00012-of-00014.bin",
188
+ "layers.25.mlp.down_proj.weight": "pytorch_model-00012-of-00014.bin",
189
+ "layers.25.mlp.gate_proj.weight": "pytorch_model-00011-of-00014.bin",
190
+ "layers.25.mlp.up_proj.weight": "pytorch_model-00011-of-00014.bin",
191
+ "layers.25.post_attention_layernorm.weight": "pytorch_model-00012-of-00014.bin",
192
+ "layers.25.self_attn.k_proj.weight": "pytorch_model-00011-of-00014.bin",
193
+ "layers.25.self_attn.o_proj.weight": "pytorch_model-00011-of-00014.bin",
194
+ "layers.25.self_attn.q_proj.weight": "pytorch_model-00011-of-00014.bin",
195
+ "layers.25.self_attn.rotary_emb.inv_freq": "pytorch_model-00011-of-00014.bin",
196
+ "layers.25.self_attn.v_proj.weight": "pytorch_model-00011-of-00014.bin",
197
+ "layers.26.input_layernorm.weight": "pytorch_model-00012-of-00014.bin",
198
+ "layers.26.mlp.down_proj.weight": "pytorch_model-00012-of-00014.bin",
199
+ "layers.26.mlp.gate_proj.weight": "pytorch_model-00012-of-00014.bin",
200
+ "layers.26.mlp.up_proj.weight": "pytorch_model-00012-of-00014.bin",
201
+ "layers.26.post_attention_layernorm.weight": "pytorch_model-00012-of-00014.bin",
202
+ "layers.26.self_attn.k_proj.weight": "pytorch_model-00012-of-00014.bin",
203
+ "layers.26.self_attn.o_proj.weight": "pytorch_model-00012-of-00014.bin",
204
+ "layers.26.self_attn.q_proj.weight": "pytorch_model-00012-of-00014.bin",
205
+ "layers.26.self_attn.rotary_emb.inv_freq": "pytorch_model-00012-of-00014.bin",
206
+ "layers.26.self_attn.v_proj.weight": "pytorch_model-00012-of-00014.bin",
207
+ "layers.27.input_layernorm.weight": "pytorch_model-00012-of-00014.bin",
208
+ "layers.27.mlp.down_proj.weight": "pytorch_model-00012-of-00014.bin",
209
+ "layers.27.mlp.gate_proj.weight": "pytorch_model-00012-of-00014.bin",
210
+ "layers.27.mlp.up_proj.weight": "pytorch_model-00012-of-00014.bin",
211
+ "layers.27.post_attention_layernorm.weight": "pytorch_model-00012-of-00014.bin",
212
+ "layers.27.self_attn.k_proj.weight": "pytorch_model-00012-of-00014.bin",
213
+ "layers.27.self_attn.o_proj.weight": "pytorch_model-00012-of-00014.bin",
214
+ "layers.27.self_attn.q_proj.weight": "pytorch_model-00012-of-00014.bin",
215
+ "layers.27.self_attn.rotary_emb.inv_freq": "pytorch_model-00012-of-00014.bin",
216
+ "layers.27.self_attn.v_proj.weight": "pytorch_model-00012-of-00014.bin",
217
+ "layers.28.input_layernorm.weight": "pytorch_model-00013-of-00014.bin",
218
+ "layers.28.mlp.down_proj.weight": "pytorch_model-00013-of-00014.bin",
219
+ "layers.28.mlp.gate_proj.weight": "pytorch_model-00013-of-00014.bin",
220
+ "layers.28.mlp.up_proj.weight": "pytorch_model-00013-of-00014.bin",
221
+ "layers.28.post_attention_layernorm.weight": "pytorch_model-00013-of-00014.bin",
222
+ "layers.28.self_attn.k_proj.weight": "pytorch_model-00012-of-00014.bin",
223
+ "layers.28.self_attn.o_proj.weight": "pytorch_model-00013-of-00014.bin",
224
+ "layers.28.self_attn.q_proj.weight": "pytorch_model-00012-of-00014.bin",
225
+ "layers.28.self_attn.rotary_emb.inv_freq": "pytorch_model-00013-of-00014.bin",
226
+ "layers.28.self_attn.v_proj.weight": "pytorch_model-00013-of-00014.bin",
227
+ "layers.29.input_layernorm.weight": "pytorch_model-00013-of-00014.bin",
228
+ "layers.29.mlp.down_proj.weight": "pytorch_model-00013-of-00014.bin",
229
+ "layers.29.mlp.gate_proj.weight": "pytorch_model-00013-of-00014.bin",
230
+ "layers.29.mlp.up_proj.weight": "pytorch_model-00013-of-00014.bin",
231
+ "layers.29.post_attention_layernorm.weight": "pytorch_model-00013-of-00014.bin",
232
+ "layers.29.self_attn.k_proj.weight": "pytorch_model-00013-of-00014.bin",
233
+ "layers.29.self_attn.o_proj.weight": "pytorch_model-00013-of-00014.bin",
234
+ "layers.29.self_attn.q_proj.weight": "pytorch_model-00013-of-00014.bin",
235
+ "layers.29.self_attn.rotary_emb.inv_freq": "pytorch_model-00013-of-00014.bin",
236
+ "layers.29.self_attn.v_proj.weight": "pytorch_model-00013-of-00014.bin",
237
+ "layers.3.input_layernorm.weight": "pytorch_model-00002-of-00014.bin",
238
+ "layers.3.mlp.down_proj.weight": "pytorch_model-00002-of-00014.bin",
239
+ "layers.3.mlp.gate_proj.weight": "pytorch_model-00002-of-00014.bin",
240
+ "layers.3.mlp.up_proj.weight": "pytorch_model-00002-of-00014.bin",
241
+ "layers.3.post_attention_layernorm.weight": "pytorch_model-00002-of-00014.bin",
242
+ "layers.3.self_attn.k_proj.weight": "pytorch_model-00002-of-00014.bin",
243
+ "layers.3.self_attn.o_proj.weight": "pytorch_model-00002-of-00014.bin",
244
+ "layers.3.self_attn.q_proj.weight": "pytorch_model-00002-of-00014.bin",
245
+ "layers.3.self_attn.rotary_emb.inv_freq": "pytorch_model-00002-of-00014.bin",
246
+ "layers.3.self_attn.v_proj.weight": "pytorch_model-00002-of-00014.bin",
247
+ "layers.30.input_layernorm.weight": "pytorch_model-00014-of-00014.bin",
248
+ "layers.30.mlp.down_proj.weight": "pytorch_model-00014-of-00014.bin",
249
+ "layers.30.mlp.gate_proj.weight": "pytorch_model-00013-of-00014.bin",
250
+ "layers.30.mlp.up_proj.weight": "pytorch_model-00014-of-00014.bin",
251
+ "layers.30.post_attention_layernorm.weight": "pytorch_model-00014-of-00014.bin",
252
+ "layers.30.self_attn.k_proj.weight": "pytorch_model-00013-of-00014.bin",
253
+ "layers.30.self_attn.o_proj.weight": "pytorch_model-00013-of-00014.bin",
254
+ "layers.30.self_attn.q_proj.weight": "pytorch_model-00013-of-00014.bin",
255
+ "layers.30.self_attn.rotary_emb.inv_freq": "pytorch_model-00013-of-00014.bin",
256
+ "layers.30.self_attn.v_proj.weight": "pytorch_model-00013-of-00014.bin",
257
+ "layers.31.input_layernorm.weight": "pytorch_model-00014-of-00014.bin",
258
+ "layers.31.mlp.down_proj.weight": "pytorch_model-00014-of-00014.bin",
259
+ "layers.31.mlp.gate_proj.weight": "pytorch_model-00014-of-00014.bin",
260
+ "layers.31.mlp.up_proj.weight": "pytorch_model-00014-of-00014.bin",
261
+ "layers.31.post_attention_layernorm.weight": "pytorch_model-00014-of-00014.bin",
262
+ "layers.31.self_attn.k_proj.weight": "pytorch_model-00014-of-00014.bin",
263
+ "layers.31.self_attn.o_proj.weight": "pytorch_model-00014-of-00014.bin",
264
+ "layers.31.self_attn.q_proj.weight": "pytorch_model-00014-of-00014.bin",
265
+ "layers.31.self_attn.rotary_emb.inv_freq": "pytorch_model-00014-of-00014.bin",
266
+ "layers.31.self_attn.v_proj.weight": "pytorch_model-00014-of-00014.bin",
267
+ "layers.4.input_layernorm.weight": "pytorch_model-00003-of-00014.bin",
268
+ "layers.4.mlp.down_proj.weight": "pytorch_model-00003-of-00014.bin",
269
+ "layers.4.mlp.gate_proj.weight": "pytorch_model-00003-of-00014.bin",
270
+ "layers.4.mlp.up_proj.weight": "pytorch_model-00003-of-00014.bin",
271
+ "layers.4.post_attention_layernorm.weight": "pytorch_model-00003-of-00014.bin",
272
+ "layers.4.self_attn.k_proj.weight": "pytorch_model-00002-of-00014.bin",
273
+ "layers.4.self_attn.o_proj.weight": "pytorch_model-00003-of-00014.bin",
274
+ "layers.4.self_attn.q_proj.weight": "pytorch_model-00002-of-00014.bin",
275
+ "layers.4.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00014.bin",
276
+ "layers.4.self_attn.v_proj.weight": "pytorch_model-00003-of-00014.bin",
277
+ "layers.5.input_layernorm.weight": "pytorch_model-00003-of-00014.bin",
278
+ "layers.5.mlp.down_proj.weight": "pytorch_model-00003-of-00014.bin",
279
+ "layers.5.mlp.gate_proj.weight": "pytorch_model-00003-of-00014.bin",
280
+ "layers.5.mlp.up_proj.weight": "pytorch_model-00003-of-00014.bin",
281
+ "layers.5.post_attention_layernorm.weight": "pytorch_model-00003-of-00014.bin",
282
+ "layers.5.self_attn.k_proj.weight": "pytorch_model-00003-of-00014.bin",
283
+ "layers.5.self_attn.o_proj.weight": "pytorch_model-00003-of-00014.bin",
284
+ "layers.5.self_attn.q_proj.weight": "pytorch_model-00003-of-00014.bin",
285
+ "layers.5.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00014.bin",
286
+ "layers.5.self_attn.v_proj.weight": "pytorch_model-00003-of-00014.bin",
287
+ "layers.6.input_layernorm.weight": "pytorch_model-00004-of-00014.bin",
288
+ "layers.6.mlp.down_proj.weight": "pytorch_model-00004-of-00014.bin",
289
+ "layers.6.mlp.gate_proj.weight": "pytorch_model-00003-of-00014.bin",
290
+ "layers.6.mlp.up_proj.weight": "pytorch_model-00004-of-00014.bin",
291
+ "layers.6.post_attention_layernorm.weight": "pytorch_model-00004-of-00014.bin",
292
+ "layers.6.self_attn.k_proj.weight": "pytorch_model-00003-of-00014.bin",
293
+ "layers.6.self_attn.o_proj.weight": "pytorch_model-00003-of-00014.bin",
294
+ "layers.6.self_attn.q_proj.weight": "pytorch_model-00003-of-00014.bin",
295
+ "layers.6.self_attn.rotary_emb.inv_freq": "pytorch_model-00003-of-00014.bin",
296
+ "layers.6.self_attn.v_proj.weight": "pytorch_model-00003-of-00014.bin",
297
+ "layers.7.input_layernorm.weight": "pytorch_model-00004-of-00014.bin",
298
+ "layers.7.mlp.down_proj.weight": "pytorch_model-00004-of-00014.bin",
299
+ "layers.7.mlp.gate_proj.weight": "pytorch_model-00004-of-00014.bin",
300
+ "layers.7.mlp.up_proj.weight": "pytorch_model-00004-of-00014.bin",
301
+ "layers.7.post_attention_layernorm.weight": "pytorch_model-00004-of-00014.bin",
302
+ "layers.7.self_attn.k_proj.weight": "pytorch_model-00004-of-00014.bin",
303
+ "layers.7.self_attn.o_proj.weight": "pytorch_model-00004-of-00014.bin",
304
+ "layers.7.self_attn.q_proj.weight": "pytorch_model-00004-of-00014.bin",
305
+ "layers.7.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00014.bin",
306
+ "layers.7.self_attn.v_proj.weight": "pytorch_model-00004-of-00014.bin",
307
+ "layers.8.input_layernorm.weight": "pytorch_model-00004-of-00014.bin",
308
+ "layers.8.mlp.down_proj.weight": "pytorch_model-00004-of-00014.bin",
309
+ "layers.8.mlp.gate_proj.weight": "pytorch_model-00004-of-00014.bin",
310
+ "layers.8.mlp.up_proj.weight": "pytorch_model-00004-of-00014.bin",
311
+ "layers.8.post_attention_layernorm.weight": "pytorch_model-00004-of-00014.bin",
312
+ "layers.8.self_attn.k_proj.weight": "pytorch_model-00004-of-00014.bin",
313
+ "layers.8.self_attn.o_proj.weight": "pytorch_model-00004-of-00014.bin",
314
+ "layers.8.self_attn.q_proj.weight": "pytorch_model-00004-of-00014.bin",
315
+ "layers.8.self_attn.rotary_emb.inv_freq": "pytorch_model-00004-of-00014.bin",
316
+ "layers.8.self_attn.v_proj.weight": "pytorch_model-00004-of-00014.bin",
317
+ "layers.9.input_layernorm.weight": "pytorch_model-00005-of-00014.bin",
318
+ "layers.9.mlp.down_proj.weight": "pytorch_model-00005-of-00014.bin",
319
+ "layers.9.mlp.gate_proj.weight": "pytorch_model-00005-of-00014.bin",
320
+ "layers.9.mlp.up_proj.weight": "pytorch_model-00005-of-00014.bin",
321
+ "layers.9.post_attention_layernorm.weight": "pytorch_model-00005-of-00014.bin",
322
+ "layers.9.self_attn.k_proj.weight": "pytorch_model-00005-of-00014.bin",
323
+ "layers.9.self_attn.o_proj.weight": "pytorch_model-00005-of-00014.bin",
324
+ "layers.9.self_attn.q_proj.weight": "pytorch_model-00005-of-00014.bin",
325
+ "layers.9.self_attn.rotary_emb.inv_freq": "pytorch_model-00005-of-00014.bin",
326
+ "layers.9.self_attn.v_proj.weight": "pytorch_model-00005-of-00014.bin",
327
+ "norm.weight": "pytorch_model-00014-of-00014.bin"
328
+ }
329
+ }