Hugging Face's logo

hoon-hee-lee
/
Llama-3.1-8B-Instruct-Nam01

llama
Model card Files Community
Hunhi commited on
Commit
8f5ec26
•
1 Parent(s): e2e85d2

init

Browse files
Files changed (6) hide show
  1. config.json +1 -0
  2. hf_model_0001_2.pt +3 -0
  3. hf_model_0002_2.pt +3 -0
  4. hf_model_0003_2.pt +3 -0
  5. hf_model_0004_2.pt +3 -0
  6. model.index.json +295 -0
config.json ADDED
@@ -0,0 +1 @@
 
 
1
+ {"architectures": ["LlamaForCausalLM"], "attention_bias": false, "attention_dropout": 0.0, "bos_token_id": 128000, "eos_token_id": [128001, 128008, 128009], "hidden_act": "silu", "hidden_size": 4096, "initializer_range": 0.02, "intermediate_size": 14336, "max_position_embeddings": 131072, "mlp_bias": false, "model_type": "llama", "num_attention_heads": 32, "num_hidden_layers": 32, "num_key_value_heads": 8, "pretraining_tp": 1, "rms_norm_eps": 1e-05, "rope_scaling": {"factor": 8.0, "low_freq_factor": 1.0, "high_freq_factor": 4.0, "original_max_position_embeddings": 8192, "rope_type": "llama3"}, "rope_theta": 500000.0, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.42.3", "use_cache": true, "vocab_size": 128256}
hf_model_0001_2.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:355c87bde11ee8ef0bfb62dc6d488cdf464e663a507e3053ba3cde3d4050872c
3
+ size 4976716132
hf_model_0002_2.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:51c2b580902261b917f3e7bf5c863b6eaefc989fccb06826f89fcb6432ab4ee3
3
+ size 4999824778
hf_model_0003_2.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:84ddcca8b33822c36347855e76e97972717b133453a93b43887f51599708e055
3
+ size 4915937346
hf_model_0004_2.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df8cd236a0011f166f5113adf990b71fadad3168749509dc899454a934918697
3
+ size 1168140692
model.index.json ADDED
@@ -0,0 +1,295 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "weight_map": {
3
+ "model.embed_tokens.weight": "hf_model_0001_2.pt",
4
+ "model.layers.0.self_attn.q_proj.weight": "hf_model_0001_2.pt",
5
+ "model.layers.0.self_attn.k_proj.weight": "hf_model_0001_2.pt",
6
+ "model.layers.0.self_attn.v_proj.weight": "hf_model_0001_2.pt",
7
+ "model.layers.0.self_attn.o_proj.weight": "hf_model_0001_2.pt",
8
+ "model.layers.0.mlp.gate_proj.weight": "hf_model_0001_2.pt",
9
+ "model.layers.0.mlp.down_proj.weight": "hf_model_0001_2.pt",
10
+ "model.layers.0.mlp.up_proj.weight": "hf_model_0001_2.pt",
11
+ "model.layers.0.input_layernorm.weight": "hf_model_0001_2.pt",
12
+ "model.layers.0.post_attention_layernorm.weight": "hf_model_0001_2.pt",
13
+ "model.layers.1.self_attn.q_proj.weight": "hf_model_0001_2.pt",
14
+ "model.layers.1.self_attn.k_proj.weight": "hf_model_0001_2.pt",
15
+ "model.layers.1.self_attn.v_proj.weight": "hf_model_0001_2.pt",
16
+ "model.layers.1.self_attn.o_proj.weight": "hf_model_0001_2.pt",
17
+ "model.layers.1.mlp.gate_proj.weight": "hf_model_0001_2.pt",
18
+ "model.layers.1.mlp.down_proj.weight": "hf_model_0001_2.pt",
19
+ "model.layers.1.mlp.up_proj.weight": "hf_model_0001_2.pt",
20
+ "model.layers.1.input_layernorm.weight": "hf_model_0001_2.pt",
21
+ "model.layers.1.post_attention_layernorm.weight": "hf_model_0001_2.pt",
22
+ "model.layers.2.self_attn.q_proj.weight": "hf_model_0001_2.pt",
23
+ "model.layers.2.self_attn.k_proj.weight": "hf_model_0001_2.pt",
24
+ "model.layers.2.self_attn.v_proj.weight": "hf_model_0001_2.pt",
25
+ "model.layers.2.self_attn.o_proj.weight": "hf_model_0001_2.pt",
26
+ "model.layers.2.mlp.gate_proj.weight": "hf_model_0001_2.pt",
27
+ "model.layers.2.mlp.down_proj.weight": "hf_model_0001_2.pt",
28
+ "model.layers.2.mlp.up_proj.weight": "hf_model_0001_2.pt",
29
+ "model.layers.2.input_layernorm.weight": "hf_model_0001_2.pt",
30
+ "model.layers.2.post_attention_layernorm.weight": "hf_model_0001_2.pt",
31
+ "model.layers.3.self_attn.q_proj.weight": "hf_model_0001_2.pt",
32
+ "model.layers.3.self_attn.k_proj.weight": "hf_model_0001_2.pt",
33
+ "model.layers.3.self_attn.v_proj.weight": "hf_model_0001_2.pt",
34
+ "model.layers.3.self_attn.o_proj.weight": "hf_model_0001_2.pt",
35
+ "model.layers.3.mlp.gate_proj.weight": "hf_model_0001_2.pt",
36
+ "model.layers.3.mlp.down_proj.weight": "hf_model_0001_2.pt",
37
+ "model.layers.3.mlp.up_proj.weight": "hf_model_0001_2.pt",
38
+ "model.layers.3.input_layernorm.weight": "hf_model_0001_2.pt",
39
+ "model.layers.3.post_attention_layernorm.weight": "hf_model_0001_2.pt",
40
+ "model.layers.4.self_attn.q_proj.weight": "hf_model_0001_2.pt",
41
+ "model.layers.4.self_attn.k_proj.weight": "hf_model_0001_2.pt",
42
+ "model.layers.4.self_attn.v_proj.weight": "hf_model_0001_2.pt",
43
+ "model.layers.4.self_attn.o_proj.weight": "hf_model_0001_2.pt",
44
+ "model.layers.4.mlp.gate_proj.weight": "hf_model_0001_2.pt",
45
+ "model.layers.4.mlp.down_proj.weight": "hf_model_0001_2.pt",
46
+ "model.layers.4.mlp.up_proj.weight": "hf_model_0001_2.pt",
47
+ "model.layers.4.input_layernorm.weight": "hf_model_0001_2.pt",
48
+ "model.layers.4.post_attention_layernorm.weight": "hf_model_0001_2.pt",
49
+ "model.layers.5.self_attn.q_proj.weight": "hf_model_0001_2.pt",
50
+ "model.layers.5.self_attn.k_proj.weight": "hf_model_0001_2.pt",
51
+ "model.layers.5.self_attn.v_proj.weight": "hf_model_0001_2.pt",
52
+ "model.layers.5.self_attn.o_proj.weight": "hf_model_0001_2.pt",
53
+ "model.layers.5.mlp.gate_proj.weight": "hf_model_0001_2.pt",
54
+ "model.layers.5.mlp.down_proj.weight": "hf_model_0001_2.pt",
55
+ "model.layers.5.mlp.up_proj.weight": "hf_model_0001_2.pt",
56
+ "model.layers.5.input_layernorm.weight": "hf_model_0001_2.pt",
57
+ "model.layers.5.post_attention_layernorm.weight": "hf_model_0001_2.pt",
58
+ "model.layers.6.self_attn.q_proj.weight": "hf_model_0001_2.pt",
59
+ "model.layers.6.self_attn.k_proj.weight": "hf_model_0001_2.pt",
60
+ "model.layers.6.self_attn.v_proj.weight": "hf_model_0001_2.pt",
61
+ "model.layers.6.self_attn.o_proj.weight": "hf_model_0001_2.pt",
62
+ "model.layers.6.mlp.gate_proj.weight": "hf_model_0001_2.pt",
63
+ "model.layers.6.mlp.down_proj.weight": "hf_model_0001_2.pt",
64
+ "model.layers.6.mlp.up_proj.weight": "hf_model_0001_2.pt",
65
+ "model.layers.6.input_layernorm.weight": "hf_model_0001_2.pt",
66
+ "model.layers.6.post_attention_layernorm.weight": "hf_model_0001_2.pt",
67
+ "model.layers.7.self_attn.q_proj.weight": "hf_model_0001_2.pt",
68
+ "model.layers.7.self_attn.k_proj.weight": "hf_model_0001_2.pt",
69
+ "model.layers.7.self_attn.v_proj.weight": "hf_model_0001_2.pt",
70
+ "model.layers.7.self_attn.o_proj.weight": "hf_model_0001_2.pt",
71
+ "model.layers.7.mlp.gate_proj.weight": "hf_model_0001_2.pt",
72
+ "model.layers.7.mlp.down_proj.weight": "hf_model_0001_2.pt",
73
+ "model.layers.7.mlp.up_proj.weight": "hf_model_0001_2.pt",
74
+ "model.layers.7.input_layernorm.weight": "hf_model_0001_2.pt",
75
+ "model.layers.7.post_attention_layernorm.weight": "hf_model_0001_2.pt",
76
+ "model.layers.8.self_attn.q_proj.weight": "hf_model_0001_2.pt",
77
+ "model.layers.8.self_attn.k_proj.weight": "hf_model_0001_2.pt",
78
+ "model.layers.8.self_attn.v_proj.weight": "hf_model_0001_2.pt",
79
+ "model.layers.8.self_attn.o_proj.weight": "hf_model_0001_2.pt",
80
+ "model.layers.8.mlp.gate_proj.weight": "hf_model_0001_2.pt",
81
+ "model.layers.8.mlp.down_proj.weight": "hf_model_0001_2.pt",
82
+ "model.layers.8.mlp.up_proj.weight": "hf_model_0001_2.pt",
83
+ "model.layers.8.input_layernorm.weight": "hf_model_0001_2.pt",
84
+ "model.layers.8.post_attention_layernorm.weight": "hf_model_0001_2.pt",
85
+ "model.layers.9.self_attn.q_proj.weight": "hf_model_0002_2.pt",
86
+ "model.layers.9.self_attn.k_proj.weight": "hf_model_0002_2.pt",
87
+ "model.layers.9.self_attn.v_proj.weight": "hf_model_0002_2.pt",
88
+ "model.layers.9.self_attn.o_proj.weight": "hf_model_0002_2.pt",
89
+ "model.layers.9.mlp.gate_proj.weight": "hf_model_0002_2.pt",
90
+ "model.layers.9.mlp.down_proj.weight": "hf_model_0002_2.pt",
91
+ "model.layers.9.mlp.up_proj.weight": "hf_model_0002_2.pt",
92
+ "model.layers.9.input_layernorm.weight": "hf_model_0002_2.pt",
93
+ "model.layers.9.post_attention_layernorm.weight": "hf_model_0002_2.pt",
94
+ "model.layers.10.self_attn.q_proj.weight": "hf_model_0002_2.pt",
95
+ "model.layers.10.self_attn.k_proj.weight": "hf_model_0002_2.pt",
96
+ "model.layers.10.self_attn.v_proj.weight": "hf_model_0002_2.pt",
97
+ "model.layers.10.self_attn.o_proj.weight": "hf_model_0002_2.pt",
98
+ "model.layers.10.mlp.gate_proj.weight": "hf_model_0002_2.pt",
99
+ "model.layers.10.mlp.down_proj.weight": "hf_model_0002_2.pt",
100
+ "model.layers.10.mlp.up_proj.weight": "hf_model_0002_2.pt",
101
+ "model.layers.10.input_layernorm.weight": "hf_model_0002_2.pt",
102
+ "model.layers.10.post_attention_layernorm.weight": "hf_model_0002_2.pt",
103
+ "model.layers.11.self_attn.q_proj.weight": "hf_model_0002_2.pt",
104
+ "model.layers.11.self_attn.k_proj.weight": "hf_model_0002_2.pt",
105
+ "model.layers.11.self_attn.v_proj.weight": "hf_model_0002_2.pt",
106
+ "model.layers.11.self_attn.o_proj.weight": "hf_model_0002_2.pt",
107
+ "model.layers.11.mlp.gate_proj.weight": "hf_model_0002_2.pt",
108
+ "model.layers.11.mlp.down_proj.weight": "hf_model_0002_2.pt",
109
+ "model.layers.11.mlp.up_proj.weight": "hf_model_0002_2.pt",
110
+ "model.layers.11.input_layernorm.weight": "hf_model_0002_2.pt",
111
+ "model.layers.11.post_attention_layernorm.weight": "hf_model_0002_2.pt",
112
+ "model.layers.12.self_attn.q_proj.weight": "hf_model_0002_2.pt",
113
+ "model.layers.12.self_attn.k_proj.weight": "hf_model_0002_2.pt",
114
+ "model.layers.12.self_attn.v_proj.weight": "hf_model_0002_2.pt",
115
+ "model.layers.12.self_attn.o_proj.weight": "hf_model_0002_2.pt",
116
+ "model.layers.12.mlp.gate_proj.weight": "hf_model_0002_2.pt",
117
+ "model.layers.12.mlp.down_proj.weight": "hf_model_0002_2.pt",
118
+ "model.layers.12.mlp.up_proj.weight": "hf_model_0002_2.pt",
119
+ "model.layers.12.input_layernorm.weight": "hf_model_0002_2.pt",
120
+ "model.layers.12.post_attention_layernorm.weight": "hf_model_0002_2.pt",
121
+ "model.layers.13.self_attn.q_proj.weight": "hf_model_0002_2.pt",
122
+ "model.layers.13.self_attn.k_proj.weight": "hf_model_0002_2.pt",
123
+ "model.layers.13.self_attn.v_proj.weight": "hf_model_0002_2.pt",
124
+ "model.layers.13.self_attn.o_proj.weight": "hf_model_0002_2.pt",
125
+ "model.layers.13.mlp.gate_proj.weight": "hf_model_0002_2.pt",
126
+ "model.layers.13.mlp.down_proj.weight": "hf_model_0002_2.pt",
127
+ "model.layers.13.mlp.up_proj.weight": "hf_model_0002_2.pt",
128
+ "model.layers.13.input_layernorm.weight": "hf_model_0002_2.pt",
129
+ "model.layers.13.post_attention_layernorm.weight": "hf_model_0002_2.pt",
130
+ "model.layers.14.self_attn.q_proj.weight": "hf_model_0002_2.pt",
131
+ "model.layers.14.self_attn.k_proj.weight": "hf_model_0002_2.pt",
132
+ "model.layers.14.self_attn.v_proj.weight": "hf_model_0002_2.pt",
133
+ "model.layers.14.self_attn.o_proj.weight": "hf_model_0002_2.pt",
134
+ "model.layers.14.mlp.gate_proj.weight": "hf_model_0002_2.pt",
135
+ "model.layers.14.mlp.down_proj.weight": "hf_model_0002_2.pt",
136
+ "model.layers.14.mlp.up_proj.weight": "hf_model_0002_2.pt",
137
+ "model.layers.14.input_layernorm.weight": "hf_model_0002_2.pt",
138
+ "model.layers.14.post_attention_layernorm.weight": "hf_model_0002_2.pt",
139
+ "model.layers.15.self_attn.q_proj.weight": "hf_model_0002_2.pt",
140
+ "model.layers.15.self_attn.k_proj.weight": "hf_model_0002_2.pt",
141
+ "model.layers.15.self_attn.v_proj.weight": "hf_model_0002_2.pt",
142
+ "model.layers.15.self_attn.o_proj.weight": "hf_model_0002_2.pt",
143
+ "model.layers.15.mlp.gate_proj.weight": "hf_model_0002_2.pt",
144
+ "model.layers.15.mlp.down_proj.weight": "hf_model_0002_2.pt",
145
+ "model.layers.15.mlp.up_proj.weight": "hf_model_0002_2.pt",
146
+ "model.layers.15.input_layernorm.weight": "hf_model_0002_2.pt",
147
+ "model.layers.15.post_attention_layernorm.weight": "hf_model_0002_2.pt",
148
+ "model.layers.16.self_attn.q_proj.weight": "hf_model_0002_2.pt",
149
+ "model.layers.16.self_attn.k_proj.weight": "hf_model_0002_2.pt",
150
+ "model.layers.16.self_attn.v_proj.weight": "hf_model_0002_2.pt",
151
+ "model.layers.16.self_attn.o_proj.weight": "hf_model_0002_2.pt",
152
+ "model.layers.16.mlp.gate_proj.weight": "hf_model_0002_2.pt",
153
+ "model.layers.16.mlp.down_proj.weight": "hf_model_0002_2.pt",
154
+ "model.layers.16.mlp.up_proj.weight": "hf_model_0002_2.pt",
155
+ "model.layers.16.input_layernorm.weight": "hf_model_0002_2.pt",
156
+ "model.layers.16.post_attention_layernorm.weight": "hf_model_0002_2.pt",
157
+ "model.layers.17.self_attn.q_proj.weight": "hf_model_0002_2.pt",
158
+ "model.layers.17.self_attn.k_proj.weight": "hf_model_0002_2.pt",
159
+ "model.layers.17.self_attn.v_proj.weight": "hf_model_0002_2.pt",
160
+ "model.layers.17.self_attn.o_proj.weight": "hf_model_0002_2.pt",
161
+ "model.layers.17.mlp.gate_proj.weight": "hf_model_0002_2.pt",
162
+ "model.layers.17.mlp.down_proj.weight": "hf_model_0002_2.pt",
163
+ "model.layers.17.mlp.up_proj.weight": "hf_model_0002_2.pt",
164
+ "model.layers.17.input_layernorm.weight": "hf_model_0002_2.pt",
165
+ "model.layers.17.post_attention_layernorm.weight": "hf_model_0002_2.pt",
166
+ "model.layers.18.self_attn.q_proj.weight": "hf_model_0002_2.pt",
167
+ "model.layers.18.self_attn.k_proj.weight": "hf_model_0002_2.pt",
168
+ "model.layers.18.self_attn.v_proj.weight": "hf_model_0002_2.pt",
169
+ "model.layers.18.self_attn.o_proj.weight": "hf_model_0002_2.pt",
170
+ "model.layers.18.mlp.gate_proj.weight": "hf_model_0002_2.pt",
171
+ "model.layers.18.mlp.down_proj.weight": "hf_model_0002_2.pt",
172
+ "model.layers.18.mlp.up_proj.weight": "hf_model_0002_2.pt",
173
+ "model.layers.18.input_layernorm.weight": "hf_model_0002_2.pt",
174
+ "model.layers.18.post_attention_layernorm.weight": "hf_model_0002_2.pt",
175
+ "model.layers.19.self_attn.q_proj.weight": "hf_model_0002_2.pt",
176
+ "model.layers.19.self_attn.k_proj.weight": "hf_model_0002_2.pt",
177
+ "model.layers.19.self_attn.v_proj.weight": "hf_model_0002_2.pt",
178
+ "model.layers.19.self_attn.o_proj.weight": "hf_model_0002_2.pt",
179
+ "model.layers.19.mlp.gate_proj.weight": "hf_model_0002_2.pt",
180
+ "model.layers.19.mlp.down_proj.weight": "hf_model_0002_2.pt",
181
+ "model.layers.19.mlp.up_proj.weight": "hf_model_0002_2.pt",
182
+ "model.layers.19.input_layernorm.weight": "hf_model_0002_2.pt",
183
+ "model.layers.19.post_attention_layernorm.weight": "hf_model_0002_2.pt",
184
+ "model.layers.20.self_attn.q_proj.weight": "hf_model_0002_2.pt",
185
+ "model.layers.20.self_attn.k_proj.weight": "hf_model_0002_2.pt",
186
+ "model.layers.20.self_attn.v_proj.weight": "hf_model_0002_2.pt",
187
+ "model.layers.20.self_attn.o_proj.weight": "hf_model_0002_2.pt",
188
+ "model.layers.20.mlp.gate_proj.weight": "hf_model_0002_2.pt",
189
+ "model.layers.20.mlp.down_proj.weight": "hf_model_0003_2.pt",
190
+ "model.layers.20.mlp.up_proj.weight": "hf_model_0003_2.pt",
191
+ "model.layers.20.input_layernorm.weight": "hf_model_0003_2.pt",
192
+ "model.layers.20.post_attention_layernorm.weight": "hf_model_0003_2.pt",
193
+ "model.layers.21.self_attn.q_proj.weight": "hf_model_0003_2.pt",
194
+ "model.layers.21.self_attn.k_proj.weight": "hf_model_0003_2.pt",
195
+ "model.layers.21.self_attn.v_proj.weight": "hf_model_0003_2.pt",
196
+ "model.layers.21.self_attn.o_proj.weight": "hf_model_0003_2.pt",
197
+ "model.layers.21.mlp.gate_proj.weight": "hf_model_0003_2.pt",
198
+ "model.layers.21.mlp.down_proj.weight": "hf_model_0003_2.pt",
199
+ "model.layers.21.mlp.up_proj.weight": "hf_model_0003_2.pt",
200
+ "model.layers.21.input_layernorm.weight": "hf_model_0003_2.pt",
201
+ "model.layers.21.post_attention_layernorm.weight": "hf_model_0003_2.pt",
202
+ "model.layers.22.self_attn.q_proj.weight": "hf_model_0003_2.pt",
203
+ "model.layers.22.self_attn.k_proj.weight": "hf_model_0003_2.pt",
204
+ "model.layers.22.self_attn.v_proj.weight": "hf_model_0003_2.pt",
205
+ "model.layers.22.self_attn.o_proj.weight": "hf_model_0003_2.pt",
206
+ "model.layers.22.mlp.gate_proj.weight": "hf_model_0003_2.pt",
207
+ "model.layers.22.mlp.down_proj.weight": "hf_model_0003_2.pt",
208
+ "model.layers.22.mlp.up_proj.weight": "hf_model_0003_2.pt",
209
+ "model.layers.22.input_layernorm.weight": "hf_model_0003_2.pt",
210
+ "model.layers.22.post_attention_layernorm.weight": "hf_model_0003_2.pt",
211
+ "model.layers.23.self_attn.q_proj.weight": "hf_model_0003_2.pt",
212
+ "model.layers.23.self_attn.k_proj.weight": "hf_model_0003_2.pt",
213
+ "model.layers.23.self_attn.v_proj.weight": "hf_model_0003_2.pt",
214
+ "model.layers.23.self_attn.o_proj.weight": "hf_model_0003_2.pt",
215
+ "model.layers.23.mlp.gate_proj.weight": "hf_model_0003_2.pt",
216
+ "model.layers.23.mlp.down_proj.weight": "hf_model_0003_2.pt",
217
+ "model.layers.23.mlp.up_proj.weight": "hf_model_0003_2.pt",
218
+ "model.layers.23.input_layernorm.weight": "hf_model_0003_2.pt",
219
+ "model.layers.23.post_attention_layernorm.weight": "hf_model_0003_2.pt",
220
+ "model.layers.24.self_attn.q_proj.weight": "hf_model_0003_2.pt",
221
+ "model.layers.24.self_attn.k_proj.weight": "hf_model_0003_2.pt",
222
+ "model.layers.24.self_attn.v_proj.weight": "hf_model_0003_2.pt",
223
+ "model.layers.24.self_attn.o_proj.weight": "hf_model_0003_2.pt",
224
+ "model.layers.24.mlp.gate_proj.weight": "hf_model_0003_2.pt",
225
+ "model.layers.24.mlp.down_proj.weight": "hf_model_0003_2.pt",
226
+ "model.layers.24.mlp.up_proj.weight": "hf_model_0003_2.pt",
227
+ "model.layers.24.input_layernorm.weight": "hf_model_0003_2.pt",
228
+ "model.layers.24.post_attention_layernorm.weight": "hf_model_0003_2.pt",
229
+ "model.layers.25.self_attn.q_proj.weight": "hf_model_0003_2.pt",
230
+ "model.layers.25.self_attn.k_proj.weight": "hf_model_0003_2.pt",
231
+ "model.layers.25.self_attn.v_proj.weight": "hf_model_0003_2.pt",
232
+ "model.layers.25.self_attn.o_proj.weight": "hf_model_0003_2.pt",
233
+ "model.layers.25.mlp.gate_proj.weight": "hf_model_0003_2.pt",
234
+ "model.layers.25.mlp.down_proj.weight": "hf_model_0003_2.pt",
235
+ "model.layers.25.mlp.up_proj.weight": "hf_model_0003_2.pt",
236
+ "model.layers.25.input_layernorm.weight": "hf_model_0003_2.pt",
237
+ "model.layers.25.post_attention_layernorm.weight": "hf_model_0003_2.pt",
238
+ "model.layers.26.self_attn.q_proj.weight": "hf_model_0003_2.pt",
239
+ "model.layers.26.self_attn.k_proj.weight": "hf_model_0003_2.pt",
240
+ "model.layers.26.self_attn.v_proj.weight": "hf_model_0003_2.pt",
241
+ "model.layers.26.self_attn.o_proj.weight": "hf_model_0003_2.pt",
242
+ "model.layers.26.mlp.gate_proj.weight": "hf_model_0003_2.pt",
243
+ "model.layers.26.mlp.down_proj.weight": "hf_model_0003_2.pt",
244
+ "model.layers.26.mlp.up_proj.weight": "hf_model_0003_2.pt",
245
+ "model.layers.26.input_layernorm.weight": "hf_model_0003_2.pt",
246
+ "model.layers.26.post_attention_layernorm.weight": "hf_model_0003_2.pt",
247
+ "model.layers.27.self_attn.q_proj.weight": "hf_model_0003_2.pt",
248
+ "model.layers.27.self_attn.k_proj.weight": "hf_model_0003_2.pt",
249
+ "model.layers.27.self_attn.v_proj.weight": "hf_model_0003_2.pt",
250
+ "model.layers.27.self_attn.o_proj.weight": "hf_model_0003_2.pt",
251
+ "model.layers.27.mlp.gate_proj.weight": "hf_model_0003_2.pt",
252
+ "model.layers.27.mlp.down_proj.weight": "hf_model_0003_2.pt",
253
+ "model.layers.27.mlp.up_proj.weight": "hf_model_0003_2.pt",
254
+ "model.layers.27.input_layernorm.weight": "hf_model_0003_2.pt",
255
+ "model.layers.27.post_attention_layernorm.weight": "hf_model_0003_2.pt",
256
+ "model.layers.28.self_attn.q_proj.weight": "hf_model_0003_2.pt",
257
+ "model.layers.28.self_attn.k_proj.weight": "hf_model_0003_2.pt",
258
+ "model.layers.28.self_attn.v_proj.weight": "hf_model_0003_2.pt",
259
+ "model.layers.28.self_attn.o_proj.weight": "hf_model_0003_2.pt",
260
+ "model.layers.28.mlp.gate_proj.weight": "hf_model_0003_2.pt",
261
+ "model.layers.28.mlp.down_proj.weight": "hf_model_0003_2.pt",
262
+ "model.layers.28.mlp.up_proj.weight": "hf_model_0003_2.pt",
263
+ "model.layers.28.input_layernorm.weight": "hf_model_0003_2.pt",
264
+ "model.layers.28.post_attention_layernorm.weight": "hf_model_0003_2.pt",
265
+ "model.layers.29.self_attn.q_proj.weight": "hf_model_0003_2.pt",
266
+ "model.layers.29.self_attn.k_proj.weight": "hf_model_0003_2.pt",
267
+ "model.layers.29.self_attn.v_proj.weight": "hf_model_0003_2.pt",
268
+ "model.layers.29.self_attn.o_proj.weight": "hf_model_0003_2.pt",
269
+ "model.layers.29.mlp.gate_proj.weight": "hf_model_0003_2.pt",
270
+ "model.layers.29.mlp.down_proj.weight": "hf_model_0003_2.pt",
271
+ "model.layers.29.mlp.up_proj.weight": "hf_model_0003_2.pt",
272
+ "model.layers.29.input_layernorm.weight": "hf_model_0003_2.pt",
273
+ "model.layers.29.post_attention_layernorm.weight": "hf_model_0003_2.pt",
274
+ "model.layers.30.self_attn.q_proj.weight": "hf_model_0003_2.pt",
275
+ "model.layers.30.self_attn.k_proj.weight": "hf_model_0003_2.pt",
276
+ "model.layers.30.self_attn.v_proj.weight": "hf_model_0003_2.pt",
277
+ "model.layers.30.self_attn.o_proj.weight": "hf_model_0003_2.pt",
278
+ "model.layers.30.mlp.gate_proj.weight": "hf_model_0003_2.pt",
279
+ "model.layers.30.mlp.down_proj.weight": "hf_model_0003_2.pt",
280
+ "model.layers.30.mlp.up_proj.weight": "hf_model_0003_2.pt",
281
+ "model.layers.30.input_layernorm.weight": "hf_model_0003_2.pt",
282
+ "model.layers.30.post_attention_layernorm.weight": "hf_model_0003_2.pt",
283
+ "model.layers.31.self_attn.q_proj.weight": "hf_model_0003_2.pt",
284
+ "model.layers.31.self_attn.k_proj.weight": "hf_model_0003_2.pt",
285
+ "model.layers.31.self_attn.v_proj.weight": "hf_model_0003_2.pt",
286
+ "model.layers.31.self_attn.o_proj.weight": "hf_model_0003_2.pt",
287
+ "model.layers.31.mlp.gate_proj.weight": "hf_model_0003_2.pt",
288
+ "model.layers.31.mlp.up_proj.weight": "hf_model_0003_2.pt",
289
+ "model.layers.31.mlp.down_proj.weight": "hf_model_0004_2.pt",
290
+ "model.layers.31.input_layernorm.weight": "hf_model_0004_2.pt",
291
+ "model.layers.31.post_attention_layernorm.weight": "hf_model_0004_2.pt",
292
+ "model.norm.weight": "hf_model_0004_2.pt",
293
+ "lm_head.weight": "hf_model_0004_2.pt"
294
+ }
295
+ }