Hugging Face's logo

Abzu
/
mpt-30b-instruct-q8

Text Generation
Transformers
Safetensors
mpt
Composer
MosaicML
llm-foundry
custom_code
Inference Endpoints
text-generation-inference
8-bit precision
Model card Files Community
Danivilanova commited on
Commit
f8be025
•
1 Parent(s): 5668a66

Upload MPTForCausalLM

Browse files
Files changed (7) hide show
  1. config.json +64 -0
  2. generation_config.json +5 -0
  3. pytorch_model-00001-of-00004.bin +3 -0
  4. pytorch_model-00002-of-00004.bin +3 -0
  5. pytorch_model-00003-of-00004.bin +3 -0
  6. pytorch_model-00004-of-00004.bin +3 -0
  7. pytorch_model.bin.index.json +489 -0
config.json ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "mosaicml/mpt-30b-instruct",
3
+ "architectures": [
4
+ "MPTForCausalLM"
5
+ ],
6
+ "attn_config": {
7
+ "alibi": true,
8
+ "alibi_bias_max": 8,
9
+ "attn_impl": "triton",
10
+ "attn_pdrop": 0,
11
+ "attn_type": "multihead_attention",
12
+ "attn_uses_sequence_id": false,
13
+ "clip_qkv": null,
14
+ "prefix_lm": false,
15
+ "qk_ln": false,
16
+ "softmax_scale": null
17
+ },
18
+ "auto_map": {
19
+ "AutoConfig": "mosaicml/mpt-30b-instruct--configuration_mpt.MPTConfig",
20
+ "AutoModelForCausalLM": "mosaicml/mpt-30b-instruct--modeling_mpt.MPTForCausalLM"
21
+ },
22
+ "d_model": 7168,
23
+ "emb_pdrop": 0,
24
+ "embedding_fraction": 1.0,
25
+ "expansion_ratio": 4,
26
+ "init_config": {
27
+ "emb_init_std": null,
28
+ "emb_init_uniform_lim": null,
29
+ "fan_mode": "fan_in",
30
+ "init_div_is_residual": true,
31
+ "init_gain": 0.0,
32
+ "init_nonlinearity": "relu",
33
+ "init_std": null,
34
+ "name": "kaiming_normal_",
35
+ "verbose": 0
36
+ },
37
+ "init_device": "cuda:0",
38
+ "learned_pos_emb": true,
39
+ "logit_scale": null,
40
+ "max_seq_len": 8192,
41
+ "model_type": "mpt",
42
+ "n_heads": 64,
43
+ "n_layers": 48,
44
+ "no_bias": true,
45
+ "norm_type": "low_precision_layernorm",
46
+ "quantization_config": {
47
+ "bnb_4bit_compute_dtype": "float32",
48
+ "bnb_4bit_quant_type": "fp4",
49
+ "bnb_4bit_use_double_quant": false,
50
+ "llm_int8_enable_fp32_cpu_offload": false,
51
+ "llm_int8_has_fp16_weight": false,
52
+ "llm_int8_skip_modules": null,
53
+ "llm_int8_threshold": 6.0,
54
+ "load_in_4bit": false,
55
+ "load_in_8bit": true
56
+ },
57
+ "resid_pdrop": 0,
58
+ "tokenizer_name": "EleutherAI/gpt-neox-20b",
59
+ "torch_dtype": "bfloat16",
60
+ "transformers_version": "4.30.2",
61
+ "use_cache": false,
62
+ "verbose": 0,
63
+ "vocab_size": 50432
64
+ }
generation_config.json ADDED
@@ -0,0 +1,5 @@
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "transformers_version": "4.30.2",
4
+ "use_cache": false
5
+ }
pytorch_model-00001-of-00004.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:126349b854cd6ad5d16359055b045ded9f4e4823d535302b3caf96d3ceea25b3
3
+ size 9975802535
pytorch_model-00002-of-00004.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a0fafb0cc37843b178f824d58e9e6954a221bd975dbacb49d04175c78bd4806d
3
+ size 9869647383
pytorch_model-00003-of-00004.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6ecaf57d06e26c14eb4fcc4242172b6a39bc6a4814d167aef2d246dee7a0ff68
3
+ size 9869647383
pytorch_model-00004-of-00004.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f57bbc01759cc12ee320a82a9d1b426c702e524b7dc0d27125a716efe61cf1e0
3
+ size 616853251
pytorch_model.bin.index.json ADDED
@@ -0,0 +1,489 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 30331779072
4
+ },
5
+ "weight_map": {
6
+ "transformer.blocks.0.attn.Wqkv.SCB": "pytorch_model-00001-of-00004.bin",
7
+ "transformer.blocks.0.attn.Wqkv.weight": "pytorch_model-00001-of-00004.bin",
8
+ "transformer.blocks.0.attn.out_proj.SCB": "pytorch_model-00001-of-00004.bin",
9
+ "transformer.blocks.0.attn.out_proj.weight": "pytorch_model-00001-of-00004.bin",
10
+ "transformer.blocks.0.ffn.down_proj.SCB": "pytorch_model-00001-of-00004.bin",
11
+ "transformer.blocks.0.ffn.down_proj.weight": "pytorch_model-00001-of-00004.bin",
12
+ "transformer.blocks.0.ffn.up_proj.SCB": "pytorch_model-00001-of-00004.bin",
13
+ "transformer.blocks.0.ffn.up_proj.weight": "pytorch_model-00001-of-00004.bin",
14
+ "transformer.blocks.0.norm_1.weight": "pytorch_model-00001-of-00004.bin",
15
+ "transformer.blocks.0.norm_2.weight": "pytorch_model-00001-of-00004.bin",
16
+ "transformer.blocks.1.attn.Wqkv.SCB": "pytorch_model-00001-of-00004.bin",
17
+ "transformer.blocks.1.attn.Wqkv.weight": "pytorch_model-00001-of-00004.bin",
18
+ "transformer.blocks.1.attn.out_proj.SCB": "pytorch_model-00001-of-00004.bin",
19
+ "transformer.blocks.1.attn.out_proj.weight": "pytorch_model-00001-of-00004.bin",
20
+ "transformer.blocks.1.ffn.down_proj.SCB": "pytorch_model-00001-of-00004.bin",
21
+ "transformer.blocks.1.ffn.down_proj.weight": "pytorch_model-00001-of-00004.bin",
22
+ "transformer.blocks.1.ffn.up_proj.SCB": "pytorch_model-00001-of-00004.bin",
23
+ "transformer.blocks.1.ffn.up_proj.weight": "pytorch_model-00001-of-00004.bin",
24
+ "transformer.blocks.1.norm_1.weight": "pytorch_model-00001-of-00004.bin",
25
+ "transformer.blocks.1.norm_2.weight": "pytorch_model-00001-of-00004.bin",
26
+ "transformer.blocks.10.attn.Wqkv.SCB": "pytorch_model-00001-of-00004.bin",
27
+ "transformer.blocks.10.attn.Wqkv.weight": "pytorch_model-00001-of-00004.bin",
28
+ "transformer.blocks.10.attn.out_proj.SCB": "pytorch_model-00001-of-00004.bin",
29
+ "transformer.blocks.10.attn.out_proj.weight": "pytorch_model-00001-of-00004.bin",
30
+ "transformer.blocks.10.ffn.down_proj.SCB": "pytorch_model-00001-of-00004.bin",
31
+ "transformer.blocks.10.ffn.down_proj.weight": "pytorch_model-00001-of-00004.bin",
32
+ "transformer.blocks.10.ffn.up_proj.SCB": "pytorch_model-00001-of-00004.bin",
33
+ "transformer.blocks.10.ffn.up_proj.weight": "pytorch_model-00001-of-00004.bin",
34
+ "transformer.blocks.10.norm_1.weight": "pytorch_model-00001-of-00004.bin",
35
+ "transformer.blocks.10.norm_2.weight": "pytorch_model-00001-of-00004.bin",
36
+ "transformer.blocks.11.attn.Wqkv.SCB": "pytorch_model-00001-of-00004.bin",
37
+ "transformer.blocks.11.attn.Wqkv.weight": "pytorch_model-00001-of-00004.bin",
38
+ "transformer.blocks.11.attn.out_proj.SCB": "pytorch_model-00001-of-00004.bin",
39
+ "transformer.blocks.11.attn.out_proj.weight": "pytorch_model-00001-of-00004.bin",
40
+ "transformer.blocks.11.ffn.down_proj.SCB": "pytorch_model-00001-of-00004.bin",
41
+ "transformer.blocks.11.ffn.down_proj.weight": "pytorch_model-00001-of-00004.bin",
42
+ "transformer.blocks.11.ffn.up_proj.SCB": "pytorch_model-00001-of-00004.bin",
43
+ "transformer.blocks.11.ffn.up_proj.weight": "pytorch_model-00001-of-00004.bin",
44
+ "transformer.blocks.11.norm_1.weight": "pytorch_model-00001-of-00004.bin",
45
+ "transformer.blocks.11.norm_2.weight": "pytorch_model-00001-of-00004.bin",
46
+ "transformer.blocks.12.attn.Wqkv.SCB": "pytorch_model-00001-of-00004.bin",
47
+ "transformer.blocks.12.attn.Wqkv.weight": "pytorch_model-00001-of-00004.bin",
48
+ "transformer.blocks.12.attn.out_proj.SCB": "pytorch_model-00001-of-00004.bin",
49
+ "transformer.blocks.12.attn.out_proj.weight": "pytorch_model-00001-of-00004.bin",
50
+ "transformer.blocks.12.ffn.down_proj.SCB": "pytorch_model-00001-of-00004.bin",
51
+ "transformer.blocks.12.ffn.down_proj.weight": "pytorch_model-00001-of-00004.bin",
52
+ "transformer.blocks.12.ffn.up_proj.SCB": "pytorch_model-00001-of-00004.bin",
53
+ "transformer.blocks.12.ffn.up_proj.weight": "pytorch_model-00001-of-00004.bin",
54
+ "transformer.blocks.12.norm_1.weight": "pytorch_model-00001-of-00004.bin",
55
+ "transformer.blocks.12.norm_2.weight": "pytorch_model-00001-of-00004.bin",
56
+ "transformer.blocks.13.attn.Wqkv.SCB": "pytorch_model-00001-of-00004.bin",
57
+ "transformer.blocks.13.attn.Wqkv.weight": "pytorch_model-00001-of-00004.bin",
58
+ "transformer.blocks.13.attn.out_proj.SCB": "pytorch_model-00001-of-00004.bin",
59
+ "transformer.blocks.13.attn.out_proj.weight": "pytorch_model-00001-of-00004.bin",
60
+ "transformer.blocks.13.ffn.down_proj.SCB": "pytorch_model-00001-of-00004.bin",
61
+ "transformer.blocks.13.ffn.down_proj.weight": "pytorch_model-00001-of-00004.bin",
62
+ "transformer.blocks.13.ffn.up_proj.SCB": "pytorch_model-00001-of-00004.bin",
63
+ "transformer.blocks.13.ffn.up_proj.weight": "pytorch_model-00001-of-00004.bin",
64
+ "transformer.blocks.13.norm_1.weight": "pytorch_model-00001-of-00004.bin",
65
+ "transformer.blocks.13.norm_2.weight": "pytorch_model-00001-of-00004.bin",
66
+ "transformer.blocks.14.attn.Wqkv.SCB": "pytorch_model-00001-of-00004.bin",
67
+ "transformer.blocks.14.attn.Wqkv.weight": "pytorch_model-00001-of-00004.bin",
68
+ "transformer.blocks.14.attn.out_proj.SCB": "pytorch_model-00001-of-00004.bin",
69
+ "transformer.blocks.14.attn.out_proj.weight": "pytorch_model-00001-of-00004.bin",
70
+ "transformer.blocks.14.ffn.down_proj.SCB": "pytorch_model-00001-of-00004.bin",
71
+ "transformer.blocks.14.ffn.down_proj.weight": "pytorch_model-00001-of-00004.bin",
72
+ "transformer.blocks.14.ffn.up_proj.SCB": "pytorch_model-00001-of-00004.bin",
73
+ "transformer.blocks.14.ffn.up_proj.weight": "pytorch_model-00001-of-00004.bin",
74
+ "transformer.blocks.14.norm_1.weight": "pytorch_model-00001-of-00004.bin",
75
+ "transformer.blocks.14.norm_2.weight": "pytorch_model-00001-of-00004.bin",
76
+ "transformer.blocks.15.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
77
+ "transformer.blocks.15.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
78
+ "transformer.blocks.15.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
79
+ "transformer.blocks.15.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
80
+ "transformer.blocks.15.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
81
+ "transformer.blocks.15.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
82
+ "transformer.blocks.15.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
83
+ "transformer.blocks.15.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
84
+ "transformer.blocks.15.norm_1.weight": "pytorch_model-00001-of-00004.bin",
85
+ "transformer.blocks.15.norm_2.weight": "pytorch_model-00002-of-00004.bin",
86
+ "transformer.blocks.16.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
87
+ "transformer.blocks.16.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
88
+ "transformer.blocks.16.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
89
+ "transformer.blocks.16.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
90
+ "transformer.blocks.16.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
91
+ "transformer.blocks.16.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
92
+ "transformer.blocks.16.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
93
+ "transformer.blocks.16.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
94
+ "transformer.blocks.16.norm_1.weight": "pytorch_model-00002-of-00004.bin",
95
+ "transformer.blocks.16.norm_2.weight": "pytorch_model-00002-of-00004.bin",
96
+ "transformer.blocks.17.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
97
+ "transformer.blocks.17.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
98
+ "transformer.blocks.17.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
99
+ "transformer.blocks.17.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
100
+ "transformer.blocks.17.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
101
+ "transformer.blocks.17.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
102
+ "transformer.blocks.17.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
103
+ "transformer.blocks.17.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
104
+ "transformer.blocks.17.norm_1.weight": "pytorch_model-00002-of-00004.bin",
105
+ "transformer.blocks.17.norm_2.weight": "pytorch_model-00002-of-00004.bin",
106
+ "transformer.blocks.18.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
107
+ "transformer.blocks.18.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
108
+ "transformer.blocks.18.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
109
+ "transformer.blocks.18.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
110
+ "transformer.blocks.18.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
111
+ "transformer.blocks.18.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
112
+ "transformer.blocks.18.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
113
+ "transformer.blocks.18.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
114
+ "transformer.blocks.18.norm_1.weight": "pytorch_model-00002-of-00004.bin",
115
+ "transformer.blocks.18.norm_2.weight": "pytorch_model-00002-of-00004.bin",
116
+ "transformer.blocks.19.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
117
+ "transformer.blocks.19.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
118
+ "transformer.blocks.19.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
119
+ "transformer.blocks.19.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
120
+ "transformer.blocks.19.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
121
+ "transformer.blocks.19.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
122
+ "transformer.blocks.19.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
123
+ "transformer.blocks.19.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
124
+ "transformer.blocks.19.norm_1.weight": "pytorch_model-00002-of-00004.bin",
125
+ "transformer.blocks.19.norm_2.weight": "pytorch_model-00002-of-00004.bin",
126
+ "transformer.blocks.2.attn.Wqkv.SCB": "pytorch_model-00001-of-00004.bin",
127
+ "transformer.blocks.2.attn.Wqkv.weight": "pytorch_model-00001-of-00004.bin",
128
+ "transformer.blocks.2.attn.out_proj.SCB": "pytorch_model-00001-of-00004.bin",
129
+ "transformer.blocks.2.attn.out_proj.weight": "pytorch_model-00001-of-00004.bin",
130
+ "transformer.blocks.2.ffn.down_proj.SCB": "pytorch_model-00001-of-00004.bin",
131
+ "transformer.blocks.2.ffn.down_proj.weight": "pytorch_model-00001-of-00004.bin",
132
+ "transformer.blocks.2.ffn.up_proj.SCB": "pytorch_model-00001-of-00004.bin",
133
+ "transformer.blocks.2.ffn.up_proj.weight": "pytorch_model-00001-of-00004.bin",
134
+ "transformer.blocks.2.norm_1.weight": "pytorch_model-00001-of-00004.bin",
135
+ "transformer.blocks.2.norm_2.weight": "pytorch_model-00001-of-00004.bin",
136
+ "transformer.blocks.20.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
137
+ "transformer.blocks.20.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
138
+ "transformer.blocks.20.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
139
+ "transformer.blocks.20.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
140
+ "transformer.blocks.20.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
141
+ "transformer.blocks.20.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
142
+ "transformer.blocks.20.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
143
+ "transformer.blocks.20.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
144
+ "transformer.blocks.20.norm_1.weight": "pytorch_model-00002-of-00004.bin",
145
+ "transformer.blocks.20.norm_2.weight": "pytorch_model-00002-of-00004.bin",
146
+ "transformer.blocks.21.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
147
+ "transformer.blocks.21.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
148
+ "transformer.blocks.21.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
149
+ "transformer.blocks.21.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
150
+ "transformer.blocks.21.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
151
+ "transformer.blocks.21.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
152
+ "transformer.blocks.21.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
153
+ "transformer.blocks.21.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
154
+ "transformer.blocks.21.norm_1.weight": "pytorch_model-00002-of-00004.bin",
155
+ "transformer.blocks.21.norm_2.weight": "pytorch_model-00002-of-00004.bin",
156
+ "transformer.blocks.22.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
157
+ "transformer.blocks.22.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
158
+ "transformer.blocks.22.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
159
+ "transformer.blocks.22.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
160
+ "transformer.blocks.22.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
161
+ "transformer.blocks.22.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
162
+ "transformer.blocks.22.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
163
+ "transformer.blocks.22.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
164
+ "transformer.blocks.22.norm_1.weight": "pytorch_model-00002-of-00004.bin",
165
+ "transformer.blocks.22.norm_2.weight": "pytorch_model-00002-of-00004.bin",
166
+ "transformer.blocks.23.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
167
+ "transformer.blocks.23.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
168
+ "transformer.blocks.23.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
169
+ "transformer.blocks.23.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
170
+ "transformer.blocks.23.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
171
+ "transformer.blocks.23.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
172
+ "transformer.blocks.23.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
173
+ "transformer.blocks.23.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
174
+ "transformer.blocks.23.norm_1.weight": "pytorch_model-00002-of-00004.bin",
175
+ "transformer.blocks.23.norm_2.weight": "pytorch_model-00002-of-00004.bin",
176
+ "transformer.blocks.24.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
177
+ "transformer.blocks.24.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
178
+ "transformer.blocks.24.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
179
+ "transformer.blocks.24.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
180
+ "transformer.blocks.24.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
181
+ "transformer.blocks.24.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
182
+ "transformer.blocks.24.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
183
+ "transformer.blocks.24.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
184
+ "transformer.blocks.24.norm_1.weight": "pytorch_model-00002-of-00004.bin",
185
+ "transformer.blocks.24.norm_2.weight": "pytorch_model-00002-of-00004.bin",
186
+ "transformer.blocks.25.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
187
+ "transformer.blocks.25.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
188
+ "transformer.blocks.25.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
189
+ "transformer.blocks.25.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
190
+ "transformer.blocks.25.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
191
+ "transformer.blocks.25.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
192
+ "transformer.blocks.25.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
193
+ "transformer.blocks.25.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
194
+ "transformer.blocks.25.norm_1.weight": "pytorch_model-00002-of-00004.bin",
195
+ "transformer.blocks.25.norm_2.weight": "pytorch_model-00002-of-00004.bin",
196
+ "transformer.blocks.26.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
197
+ "transformer.blocks.26.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
198
+ "transformer.blocks.26.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
199
+ "transformer.blocks.26.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
200
+ "transformer.blocks.26.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
201
+ "transformer.blocks.26.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
202
+ "transformer.blocks.26.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
203
+ "transformer.blocks.26.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
204
+ "transformer.blocks.26.norm_1.weight": "pytorch_model-00002-of-00004.bin",
205
+ "transformer.blocks.26.norm_2.weight": "pytorch_model-00002-of-00004.bin",
206
+ "transformer.blocks.27.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
207
+ "transformer.blocks.27.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
208
+ "transformer.blocks.27.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
209
+ "transformer.blocks.27.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
210
+ "transformer.blocks.27.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
211
+ "transformer.blocks.27.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
212
+ "transformer.blocks.27.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
213
+ "transformer.blocks.27.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
214
+ "transformer.blocks.27.norm_1.weight": "pytorch_model-00002-of-00004.bin",
215
+ "transformer.blocks.27.norm_2.weight": "pytorch_model-00002-of-00004.bin",
216
+ "transformer.blocks.28.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
217
+ "transformer.blocks.28.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
218
+ "transformer.blocks.28.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
219
+ "transformer.blocks.28.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
220
+ "transformer.blocks.28.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
221
+ "transformer.blocks.28.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
222
+ "transformer.blocks.28.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
223
+ "transformer.blocks.28.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
224
+ "transformer.blocks.28.norm_1.weight": "pytorch_model-00002-of-00004.bin",
225
+ "transformer.blocks.28.norm_2.weight": "pytorch_model-00002-of-00004.bin",
226
+ "transformer.blocks.29.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
227
+ "transformer.blocks.29.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
228
+ "transformer.blocks.29.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
229
+ "transformer.blocks.29.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
230
+ "transformer.blocks.29.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
231
+ "transformer.blocks.29.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
232
+ "transformer.blocks.29.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
233
+ "transformer.blocks.29.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
234
+ "transformer.blocks.29.norm_1.weight": "pytorch_model-00002-of-00004.bin",
235
+ "transformer.blocks.29.norm_2.weight": "pytorch_model-00002-of-00004.bin",
236
+ "transformer.blocks.3.attn.Wqkv.SCB": "pytorch_model-00001-of-00004.bin",
237
+ "transformer.blocks.3.attn.Wqkv.weight": "pytorch_model-00001-of-00004.bin",
238
+ "transformer.blocks.3.attn.out_proj.SCB": "pytorch_model-00001-of-00004.bin",
239
+ "transformer.blocks.3.attn.out_proj.weight": "pytorch_model-00001-of-00004.bin",
240
+ "transformer.blocks.3.ffn.down_proj.SCB": "pytorch_model-00001-of-00004.bin",
241
+ "transformer.blocks.3.ffn.down_proj.weight": "pytorch_model-00001-of-00004.bin",
242
+ "transformer.blocks.3.ffn.up_proj.SCB": "pytorch_model-00001-of-00004.bin",
243
+ "transformer.blocks.3.ffn.up_proj.weight": "pytorch_model-00001-of-00004.bin",
244
+ "transformer.blocks.3.norm_1.weight": "pytorch_model-00001-of-00004.bin",
245
+ "transformer.blocks.3.norm_2.weight": "pytorch_model-00001-of-00004.bin",
246
+ "transformer.blocks.30.attn.Wqkv.SCB": "pytorch_model-00002-of-00004.bin",
247
+ "transformer.blocks.30.attn.Wqkv.weight": "pytorch_model-00002-of-00004.bin",
248
+ "transformer.blocks.30.attn.out_proj.SCB": "pytorch_model-00002-of-00004.bin",
249
+ "transformer.blocks.30.attn.out_proj.weight": "pytorch_model-00002-of-00004.bin",
250
+ "transformer.blocks.30.ffn.down_proj.SCB": "pytorch_model-00002-of-00004.bin",
251
+ "transformer.blocks.30.ffn.down_proj.weight": "pytorch_model-00002-of-00004.bin",
252
+ "transformer.blocks.30.ffn.up_proj.SCB": "pytorch_model-00002-of-00004.bin",
253
+ "transformer.blocks.30.ffn.up_proj.weight": "pytorch_model-00002-of-00004.bin",
254
+ "transformer.blocks.30.norm_1.weight": "pytorch_model-00002-of-00004.bin",
255
+ "transformer.blocks.30.norm_2.weight": "pytorch_model-00002-of-00004.bin",
256
+ "transformer.blocks.31.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
257
+ "transformer.blocks.31.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
258
+ "transformer.blocks.31.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
259
+ "transformer.blocks.31.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
260
+ "transformer.blocks.31.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
261
+ "transformer.blocks.31.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
262
+ "transformer.blocks.31.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
263
+ "transformer.blocks.31.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
264
+ "transformer.blocks.31.norm_1.weight": "pytorch_model-00002-of-00004.bin",
265
+ "transformer.blocks.31.norm_2.weight": "pytorch_model-00003-of-00004.bin",
266
+ "transformer.blocks.32.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
267
+ "transformer.blocks.32.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
268
+ "transformer.blocks.32.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
269
+ "transformer.blocks.32.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
270
+ "transformer.blocks.32.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
271
+ "transformer.blocks.32.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
272
+ "transformer.blocks.32.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
273
+ "transformer.blocks.32.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
274
+ "transformer.blocks.32.norm_1.weight": "pytorch_model-00003-of-00004.bin",
275
+ "transformer.blocks.32.norm_2.weight": "pytorch_model-00003-of-00004.bin",
276
+ "transformer.blocks.33.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
277
+ "transformer.blocks.33.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
278
+ "transformer.blocks.33.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
279
+ "transformer.blocks.33.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
280
+ "transformer.blocks.33.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
281
+ "transformer.blocks.33.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
282
+ "transformer.blocks.33.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
283
+ "transformer.blocks.33.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
284
+ "transformer.blocks.33.norm_1.weight": "pytorch_model-00003-of-00004.bin",
285
+ "transformer.blocks.33.norm_2.weight": "pytorch_model-00003-of-00004.bin",
286
+ "transformer.blocks.34.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
287
+ "transformer.blocks.34.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
288
+ "transformer.blocks.34.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
289
+ "transformer.blocks.34.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
290
+ "transformer.blocks.34.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
291
+ "transformer.blocks.34.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
292
+ "transformer.blocks.34.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
293
+ "transformer.blocks.34.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
294
+ "transformer.blocks.34.norm_1.weight": "pytorch_model-00003-of-00004.bin",
295
+ "transformer.blocks.34.norm_2.weight": "pytorch_model-00003-of-00004.bin",
296
+ "transformer.blocks.35.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
297
+ "transformer.blocks.35.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
298
+ "transformer.blocks.35.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
299
+ "transformer.blocks.35.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
300
+ "transformer.blocks.35.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
301
+ "transformer.blocks.35.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
302
+ "transformer.blocks.35.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
303
+ "transformer.blocks.35.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
304
+ "transformer.blocks.35.norm_1.weight": "pytorch_model-00003-of-00004.bin",
305
+ "transformer.blocks.35.norm_2.weight": "pytorch_model-00003-of-00004.bin",
306
+ "transformer.blocks.36.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
307
+ "transformer.blocks.36.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
308
+ "transformer.blocks.36.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
309
+ "transformer.blocks.36.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
310
+ "transformer.blocks.36.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
311
+ "transformer.blocks.36.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
312
+ "transformer.blocks.36.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
313
+ "transformer.blocks.36.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
314
+ "transformer.blocks.36.norm_1.weight": "pytorch_model-00003-of-00004.bin",
315
+ "transformer.blocks.36.norm_2.weight": "pytorch_model-00003-of-00004.bin",
316
+ "transformer.blocks.37.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
317
+ "transformer.blocks.37.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
318
+ "transformer.blocks.37.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
319
+ "transformer.blocks.37.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
320
+ "transformer.blocks.37.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
321
+ "transformer.blocks.37.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
322
+ "transformer.blocks.37.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
323
+ "transformer.blocks.37.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
324
+ "transformer.blocks.37.norm_1.weight": "pytorch_model-00003-of-00004.bin",
325
+ "transformer.blocks.37.norm_2.weight": "pytorch_model-00003-of-00004.bin",
326
+ "transformer.blocks.38.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
327
+ "transformer.blocks.38.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
328
+ "transformer.blocks.38.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
329
+ "transformer.blocks.38.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
330
+ "transformer.blocks.38.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
331
+ "transformer.blocks.38.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
332
+ "transformer.blocks.38.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
333
+ "transformer.blocks.38.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
334
+ "transformer.blocks.38.norm_1.weight": "pytorch_model-00003-of-00004.bin",
335
+ "transformer.blocks.38.norm_2.weight": "pytorch_model-00003-of-00004.bin",
336
+ "transformer.blocks.39.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
337
+ "transformer.blocks.39.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
338
+ "transformer.blocks.39.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
339
+ "transformer.blocks.39.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
340
+ "transformer.blocks.39.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
341
+ "transformer.blocks.39.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
342
+ "transformer.blocks.39.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
343
+ "transformer.blocks.39.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
344
+ "transformer.blocks.39.norm_1.weight": "pytorch_model-00003-of-00004.bin",
345
+ "transformer.blocks.39.norm_2.weight": "pytorch_model-00003-of-00004.bin",
346
+ "transformer.blocks.4.attn.Wqkv.SCB": "pytorch_model-00001-of-00004.bin",
347
+ "transformer.blocks.4.attn.Wqkv.weight": "pytorch_model-00001-of-00004.bin",
348
+ "transformer.blocks.4.attn.out_proj.SCB": "pytorch_model-00001-of-00004.bin",
349
+ "transformer.blocks.4.attn.out_proj.weight": "pytorch_model-00001-of-00004.bin",
350
+ "transformer.blocks.4.ffn.down_proj.SCB": "pytorch_model-00001-of-00004.bin",
351
+ "transformer.blocks.4.ffn.down_proj.weight": "pytorch_model-00001-of-00004.bin",
352
+ "transformer.blocks.4.ffn.up_proj.SCB": "pytorch_model-00001-of-00004.bin",
353
+ "transformer.blocks.4.ffn.up_proj.weight": "pytorch_model-00001-of-00004.bin",
354
+ "transformer.blocks.4.norm_1.weight": "pytorch_model-00001-of-00004.bin",
355
+ "transformer.blocks.4.norm_2.weight": "pytorch_model-00001-of-00004.bin",
356
+ "transformer.blocks.40.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
357
+ "transformer.blocks.40.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
358
+ "transformer.blocks.40.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
359
+ "transformer.blocks.40.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
360
+ "transformer.blocks.40.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
361
+ "transformer.blocks.40.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
362
+ "transformer.blocks.40.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
363
+ "transformer.blocks.40.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
364
+ "transformer.blocks.40.norm_1.weight": "pytorch_model-00003-of-00004.bin",
365
+ "transformer.blocks.40.norm_2.weight": "pytorch_model-00003-of-00004.bin",
366
+ "transformer.blocks.41.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
367
+ "transformer.blocks.41.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
368
+ "transformer.blocks.41.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
369
+ "transformer.blocks.41.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
370
+ "transformer.blocks.41.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
371
+ "transformer.blocks.41.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
372
+ "transformer.blocks.41.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
373
+ "transformer.blocks.41.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
374
+ "transformer.blocks.41.norm_1.weight": "pytorch_model-00003-of-00004.bin",
375
+ "transformer.blocks.41.norm_2.weight": "pytorch_model-00003-of-00004.bin",
376
+ "transformer.blocks.42.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
377
+ "transformer.blocks.42.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
378
+ "transformer.blocks.42.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
379
+ "transformer.blocks.42.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
380
+ "transformer.blocks.42.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
381
+ "transformer.blocks.42.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
382
+ "transformer.blocks.42.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
383
+ "transformer.blocks.42.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
384
+ "transformer.blocks.42.norm_1.weight": "pytorch_model-00003-of-00004.bin",
385
+ "transformer.blocks.42.norm_2.weight": "pytorch_model-00003-of-00004.bin",
386
+ "transformer.blocks.43.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
387
+ "transformer.blocks.43.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
388
+ "transformer.blocks.43.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
389
+ "transformer.blocks.43.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
390
+ "transformer.blocks.43.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
391
+ "transformer.blocks.43.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
392
+ "transformer.blocks.43.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
393
+ "transformer.blocks.43.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
394
+ "transformer.blocks.43.norm_1.weight": "pytorch_model-00003-of-00004.bin",
395
+ "transformer.blocks.43.norm_2.weight": "pytorch_model-00003-of-00004.bin",
396
+ "transformer.blocks.44.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
397
+ "transformer.blocks.44.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
398
+ "transformer.blocks.44.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
399
+ "transformer.blocks.44.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
400
+ "transformer.blocks.44.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
401
+ "transformer.blocks.44.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
402
+ "transformer.blocks.44.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
403
+ "transformer.blocks.44.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
404
+ "transformer.blocks.44.norm_1.weight": "pytorch_model-00003-of-00004.bin",
405
+ "transformer.blocks.44.norm_2.weight": "pytorch_model-00003-of-00004.bin",
406
+ "transformer.blocks.45.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
407
+ "transformer.blocks.45.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
408
+ "transformer.blocks.45.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
409
+ "transformer.blocks.45.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
410
+ "transformer.blocks.45.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
411
+ "transformer.blocks.45.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
412
+ "transformer.blocks.45.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
413
+ "transformer.blocks.45.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
414
+ "transformer.blocks.45.norm_1.weight": "pytorch_model-00003-of-00004.bin",
415
+ "transformer.blocks.45.norm_2.weight": "pytorch_model-00003-of-00004.bin",
416
+ "transformer.blocks.46.attn.Wqkv.SCB": "pytorch_model-00003-of-00004.bin",
417
+ "transformer.blocks.46.attn.Wqkv.weight": "pytorch_model-00003-of-00004.bin",
418
+ "transformer.blocks.46.attn.out_proj.SCB": "pytorch_model-00003-of-00004.bin",
419
+ "transformer.blocks.46.attn.out_proj.weight": "pytorch_model-00003-of-00004.bin",
420
+ "transformer.blocks.46.ffn.down_proj.SCB": "pytorch_model-00003-of-00004.bin",
421
+ "transformer.blocks.46.ffn.down_proj.weight": "pytorch_model-00003-of-00004.bin",
422
+ "transformer.blocks.46.ffn.up_proj.SCB": "pytorch_model-00003-of-00004.bin",
423
+ "transformer.blocks.46.ffn.up_proj.weight": "pytorch_model-00003-of-00004.bin",
424
+ "transformer.blocks.46.norm_1.weight": "pytorch_model-00003-of-00004.bin",
425
+ "transformer.blocks.46.norm_2.weight": "pytorch_model-00003-of-00004.bin",
426
+ "transformer.blocks.47.attn.Wqkv.SCB": "pytorch_model-00004-of-00004.bin",
427
+ "transformer.blocks.47.attn.Wqkv.weight": "pytorch_model-00004-of-00004.bin",
428
+ "transformer.blocks.47.attn.out_proj.SCB": "pytorch_model-00004-of-00004.bin",
429
+ "transformer.blocks.47.attn.out_proj.weight": "pytorch_model-00004-of-00004.bin",
430
+ "transformer.blocks.47.ffn.down_proj.SCB": "pytorch_model-00004-of-00004.bin",
431
+ "transformer.blocks.47.ffn.down_proj.weight": "pytorch_model-00004-of-00004.bin",
432
+ "transformer.blocks.47.ffn.up_proj.SCB": "pytorch_model-00004-of-00004.bin",
433
+ "transformer.blocks.47.ffn.up_proj.weight": "pytorch_model-00004-of-00004.bin",
434
+ "transformer.blocks.47.norm_1.weight": "pytorch_model-00003-of-00004.bin",
435
+ "transformer.blocks.47.norm_2.weight": "pytorch_model-00004-of-00004.bin",
436
+ "transformer.blocks.5.attn.Wqkv.SCB": "pytorch_model-00001-of-00004.bin",
437
+ "transformer.blocks.5.attn.Wqkv.weight": "pytorch_model-00001-of-00004.bin",
438
+ "transformer.blocks.5.attn.out_proj.SCB": "pytorch_model-00001-of-00004.bin",
439
+ "transformer.blocks.5.attn.out_proj.weight": "pytorch_model-00001-of-00004.bin",
440
+ "transformer.blocks.5.ffn.down_proj.SCB": "pytorch_model-00001-of-00004.bin",
441
+ "transformer.blocks.5.ffn.down_proj.weight": "pytorch_model-00001-of-00004.bin",
442
+ "transformer.blocks.5.ffn.up_proj.SCB": "pytorch_model-00001-of-00004.bin",
443
+ "transformer.blocks.5.ffn.up_proj.weight": "pytorch_model-00001-of-00004.bin",
444
+ "transformer.blocks.5.norm_1.weight": "pytorch_model-00001-of-00004.bin",
445
+ "transformer.blocks.5.norm_2.weight": "pytorch_model-00001-of-00004.bin",
446
+ "transformer.blocks.6.attn.Wqkv.SCB": "pytorch_model-00001-of-00004.bin",
447
+ "transformer.blocks.6.attn.Wqkv.weight": "pytorch_model-00001-of-00004.bin",
448
+ "transformer.blocks.6.attn.out_proj.SCB": "pytorch_model-00001-of-00004.bin",
449
+ "transformer.blocks.6.attn.out_proj.weight": "pytorch_model-00001-of-00004.bin",
450
+ "transformer.blocks.6.ffn.down_proj.SCB": "pytorch_model-00001-of-00004.bin",
451
+ "transformer.blocks.6.ffn.down_proj.weight": "pytorch_model-00001-of-00004.bin",
452
+ "transformer.blocks.6.ffn.up_proj.SCB": "pytorch_model-00001-of-00004.bin",
453
+ "transformer.blocks.6.ffn.up_proj.weight": "pytorch_model-00001-of-00004.bin",
454
+ "transformer.blocks.6.norm_1.weight": "pytorch_model-00001-of-00004.bin",
455
+ "transformer.blocks.6.norm_2.weight": "pytorch_model-00001-of-00004.bin",
456
+ "transformer.blocks.7.attn.Wqkv.SCB": "pytorch_model-00001-of-00004.bin",
457
+ "transformer.blocks.7.attn.Wqkv.weight": "pytorch_model-00001-of-00004.bin",
458
+ "transformer.blocks.7.attn.out_proj.SCB": "pytorch_model-00001-of-00004.bin",
459
+ "transformer.blocks.7.attn.out_proj.weight": "pytorch_model-00001-of-00004.bin",
460
+ "transformer.blocks.7.ffn.down_proj.SCB": "pytorch_model-00001-of-00004.bin",
461
+ "transformer.blocks.7.ffn.down_proj.weight": "pytorch_model-00001-of-00004.bin",
462
+ "transformer.blocks.7.ffn.up_proj.SCB": "pytorch_model-00001-of-00004.bin",
463
+ "transformer.blocks.7.ffn.up_proj.weight": "pytorch_model-00001-of-00004.bin",
464
+ "transformer.blocks.7.norm_1.weight": "pytorch_model-00001-of-00004.bin",
465
+ "transformer.blocks.7.norm_2.weight": "pytorch_model-00001-of-00004.bin",
466
+ "transformer.blocks.8.attn.Wqkv.SCB": "pytorch_model-00001-of-00004.bin",
467
+ "transformer.blocks.8.attn.Wqkv.weight": "pytorch_model-00001-of-00004.bin",
468
+ "transformer.blocks.8.attn.out_proj.SCB": "pytorch_model-00001-of-00004.bin",
469
+ "transformer.blocks.8.attn.out_proj.weight": "pytorch_model-00001-of-00004.bin",
470
+ "transformer.blocks.8.ffn.down_proj.SCB": "pytorch_model-00001-of-00004.bin",
471
+ "transformer.blocks.8.ffn.down_proj.weight": "pytorch_model-00001-of-00004.bin",
472
+ "transformer.blocks.8.ffn.up_proj.SCB": "pytorch_model-00001-of-00004.bin",
473
+ "transformer.blocks.8.ffn.up_proj.weight": "pytorch_model-00001-of-00004.bin",
474
+ "transformer.blocks.8.norm_1.weight": "pytorch_model-00001-of-00004.bin",
475
+ "transformer.blocks.8.norm_2.weight": "pytorch_model-00001-of-00004.bin",
476
+ "transformer.blocks.9.attn.Wqkv.SCB": "pytorch_model-00001-of-00004.bin",
477
+ "transformer.blocks.9.attn.Wqkv.weight": "pytorch_model-00001-of-00004.bin",
478
+ "transformer.blocks.9.attn.out_proj.SCB": "pytorch_model-00001-of-00004.bin",
479
+ "transformer.blocks.9.attn.out_proj.weight": "pytorch_model-00001-of-00004.bin",
480
+ "transformer.blocks.9.ffn.down_proj.SCB": "pytorch_model-00001-of-00004.bin",
481
+ "transformer.blocks.9.ffn.down_proj.weight": "pytorch_model-00001-of-00004.bin",
482
+ "transformer.blocks.9.ffn.up_proj.SCB": "pytorch_model-00001-of-00004.bin",
483
+ "transformer.blocks.9.ffn.up_proj.weight": "pytorch_model-00001-of-00004.bin",
484
+ "transformer.blocks.9.norm_1.weight": "pytorch_model-00001-of-00004.bin",
485
+ "transformer.blocks.9.norm_2.weight": "pytorch_model-00001-of-00004.bin",
486
+ "transformer.norm_f.weight": "pytorch_model-00004-of-00004.bin",
487
+ "transformer.wte.weight": "pytorch_model-00001-of-00004.bin"
488
+ }
489
+ }