Hugging Face's logo

neody
/
riva-translate-4b-instruct-gptq-int8

Safetensors
mistral
vllm
8-bit precision
gptq
Model card Files
xet
Community
googlefan commited on
Commit
a4662d1
·
verified ·
1 Parent(s): 51cd7cf

Upload folder using huggingface_hub

Browse files
Files changed (12) hide show
  1. .gitattributes +1 -0
  2. chat_template.jinja +29 -0
  3. config.json +53 -0
  4. generation_config.json +7 -0
  5. model-00001-of-00002.safetensors +3 -0
  6. model-00002-of-00002.safetensors +3 -0
  7. model.safetensors.index.json +1030 -0
  8. quant_log.csv +239 -0
  9. quantize_config.json +26 -0
  10. special_tokens_map.json +24 -0
  11. tokenizer.json +3 -0
  12. tokenizer_config.json +0 -0
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
chat_template.jinja ADDED
@@ -0,0 +1,29 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- set language_pairs = {"en-zh-cn": {"source": "English", "target": "Simplified Chinese"}, "en-zh": {"source": "English", "target": "Simplified Chinese"}, "en-zh-tw": {"source": "English", "target": "Traditional Chinese"}, "en-ar": {"source": "English", "target": "Arabic"}, "en-de": {"source": "English", "target": "German"}, "en-es-es": {"source": "English", "target": "European Spanish"}, "en-es-us": {"source": "English", "target": "Latin American Spanish"}, "en-fr": {"source": "English", "target": "French"}, "en-ja": {"source": "English", "target": "Japanese"}, "en-ko": {"source": "English", "target": "Korean"}, "en-ru": {"source": "English", "target": "Russian"}, "en-pt-br": {"source": "English", "target": "Brazilian Portuguese"}} -%}
2
+ {%- set system_message = '' -%}
3
+ {%- set source_lang = '' -%}
4
+ {%- set target_lang = '' -%}
5
+ {%- if messages[0]['role'] == 'system' -%}
6
+ {%- set lang_pair = messages[0]['content'] | trim -%}
7
+ {%- set messages = messages[1:] -%}
8
+ {%- if lang_pair in language_pairs -%}
9
+ {%- set source_lang = language_pairs[lang_pair]['source'] -%}
10
+ {%- set target_lang = language_pairs[lang_pair]['target'] -%}
11
+ {%- set system_message = 'You are an expert at translating text from ' + source_lang + ' to ' + target_lang + '.' -%}
12
+ {%- else -%}
13
+ {%- set system_message = 'You are a translation expert.' -%}
14
+ {%- endif -%}
15
+ {%- endif -%}
16
+ {{- '<s>System\n' + system_message + '</s>\n' -}}
17
+ {%- for message in messages -%}
18
+ {%- if (message['role'] in ['user']) != (loop.index0 % 2 == 0) -%}
19
+ {{- raise_exception('Conversation roles must alternate between user and assistant') -}}
20
+ {%- elif message['role'] == 'user' -%}
21
+ {%- set user_content = (target_lang and 'What is the ' + target_lang + ' translation of the sentence: ' + message['content'] | trim or message['content'] | trim) -%}
22
+ {{- '<s>User\n' + user_content + '</s>\n' -}}
23
+ {%- elif message['role'] == 'assistant' -%}
24
+ {{- '<s>Assistant\n' + message['content'] | trim + '</s>\n' -}}
25
+ {%- endif -%}
26
+ {%- endfor -%}
27
+ {%- if add_generation_prompt -%}
28
+ {{ '<s>Assistant\n' }}
29
+ {%- endif -%}
config.json ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "activation": "silu",
3
+ "architectures": [
4
+ "MistralForCausalLM"
5
+ ],
6
+ "attention_dropout": 0.0,
7
+ "bos_token_id": 1,
8
+ "dtype": "bfloat16",
9
+ "eos_token_id": 2,
10
+ "head_dim": 128,
11
+ "hidden_act": "silu",
12
+ "hidden_size": 3072,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 8640,
15
+ "max_position_embeddings": 8192,
16
+ "model_type": "mistral",
17
+ "num_attention_heads": 32,
18
+ "num_hidden_layers": 34,
19
+ "num_key_value_heads": 8,
20
+ "quantization_config": {
21
+ "bits": 8,
22
+ "checkpoint_format": "gptq",
23
+ "desc_act": false,
24
+ "format": "gptq",
25
+ "group_size": 128,
26
+ "lm_head": false,
27
+ "meta": {
28
+ "act_group_aware": true,
29
+ "damp_auto_increment": 0.01,
30
+ "damp_percent": 0.05,
31
+ "gptaq": false,
32
+ "gptaq_alpha": 0.25,
33
+ "mse": 0.0,
34
+ "quantizer": [
35
+ "gptqmodel:5.6.0"
36
+ ],
37
+ "static_groups": false,
38
+ "true_sequential": true,
39
+ "uri": "https://github.com/modelcloud/gptqmodel"
40
+ },
41
+ "pack_dtype": "int32",
42
+ "pack_impl": "cpu",
43
+ "quant_method": "gptq",
44
+ "sym": true
45
+ },
46
+ "rms_norm_eps": 1e-05,
47
+ "rope_theta": 1000000.0,
48
+ "sliding_window": null,
49
+ "tie_word_embeddings": true,
50
+ "transformers_version": "4.57.3",
51
+ "use_cache": true,
52
+ "vocab_size": 131072
53
+ }
generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "do_sample": true,
5
+ "eos_token_id": 2,
6
+ "transformers_version": "4.57.3"
7
+ }
model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f2536bc283ab9cb4469b93bf3be62d351a7447455efddb6510f1aab16d9728a8
3
+ size 4279807430
model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:28467c58bcee9c0b2043a2ceda3b5939296249ab6b9befb114e0fe99473204c0
3
+ size 395375166
model.safetensors.index.json ADDED
@@ -0,0 +1,1030 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 4675182596
4
+ },
5
+ "weight_map": {
6
+ "model.embed_tokens.weight": "model-00001-of-00002.safetensors",
7
+ "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
8
+ "model.layers.0.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
9
+ "model.layers.0.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
10
+ "model.layers.0.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
11
+ "model.layers.0.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
12
+ "model.layers.0.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
13
+ "model.layers.0.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
14
+ "model.layers.0.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
15
+ "model.layers.0.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
16
+ "model.layers.0.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
17
+ "model.layers.0.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
18
+ "model.layers.0.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
19
+ "model.layers.0.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
20
+ "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
21
+ "model.layers.0.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
22
+ "model.layers.0.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
23
+ "model.layers.0.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
24
+ "model.layers.0.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
25
+ "model.layers.0.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
26
+ "model.layers.0.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
27
+ "model.layers.0.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
28
+ "model.layers.0.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
29
+ "model.layers.0.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
30
+ "model.layers.0.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
31
+ "model.layers.0.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
32
+ "model.layers.0.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
33
+ "model.layers.0.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
34
+ "model.layers.0.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
35
+ "model.layers.0.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
36
+ "model.layers.0.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
37
+ "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
38
+ "model.layers.1.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
39
+ "model.layers.1.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
40
+ "model.layers.1.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
41
+ "model.layers.1.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
42
+ "model.layers.1.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
43
+ "model.layers.1.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
44
+ "model.layers.1.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
45
+ "model.layers.1.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
46
+ "model.layers.1.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
47
+ "model.layers.1.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
48
+ "model.layers.1.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
49
+ "model.layers.1.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
50
+ "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
51
+ "model.layers.1.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
52
+ "model.layers.1.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
53
+ "model.layers.1.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
54
+ "model.layers.1.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
55
+ "model.layers.1.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
56
+ "model.layers.1.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
57
+ "model.layers.1.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
58
+ "model.layers.1.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
59
+ "model.layers.1.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
60
+ "model.layers.1.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
61
+ "model.layers.1.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
62
+ "model.layers.1.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
63
+ "model.layers.1.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
64
+ "model.layers.1.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
65
+ "model.layers.1.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
66
+ "model.layers.1.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
67
+ "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
68
+ "model.layers.10.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
69
+ "model.layers.10.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
70
+ "model.layers.10.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
71
+ "model.layers.10.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
72
+ "model.layers.10.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
73
+ "model.layers.10.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
74
+ "model.layers.10.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
75
+ "model.layers.10.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
76
+ "model.layers.10.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
77
+ "model.layers.10.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
78
+ "model.layers.10.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
79
+ "model.layers.10.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
80
+ "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
81
+ "model.layers.10.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
82
+ "model.layers.10.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
83
+ "model.layers.10.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
84
+ "model.layers.10.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
85
+ "model.layers.10.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
86
+ "model.layers.10.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
87
+ "model.layers.10.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
88
+ "model.layers.10.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
89
+ "model.layers.10.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
90
+ "model.layers.10.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
91
+ "model.layers.10.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
92
+ "model.layers.10.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
93
+ "model.layers.10.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
94
+ "model.layers.10.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
95
+ "model.layers.10.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
96
+ "model.layers.10.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
97
+ "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
98
+ "model.layers.11.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
99
+ "model.layers.11.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
100
+ "model.layers.11.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
101
+ "model.layers.11.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
102
+ "model.layers.11.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
103
+ "model.layers.11.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
104
+ "model.layers.11.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
105
+ "model.layers.11.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
106
+ "model.layers.11.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
107
+ "model.layers.11.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
108
+ "model.layers.11.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
109
+ "model.layers.11.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
110
+ "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
111
+ "model.layers.11.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
112
+ "model.layers.11.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
113
+ "model.layers.11.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
114
+ "model.layers.11.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
115
+ "model.layers.11.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
116
+ "model.layers.11.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
117
+ "model.layers.11.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
118
+ "model.layers.11.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
119
+ "model.layers.11.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
120
+ "model.layers.11.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
121
+ "model.layers.11.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
122
+ "model.layers.11.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
123
+ "model.layers.11.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
124
+ "model.layers.11.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
125
+ "model.layers.11.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
126
+ "model.layers.11.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
127
+ "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
128
+ "model.layers.12.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
129
+ "model.layers.12.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
130
+ "model.layers.12.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
131
+ "model.layers.12.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
132
+ "model.layers.12.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
133
+ "model.layers.12.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
134
+ "model.layers.12.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
135
+ "model.layers.12.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
136
+ "model.layers.12.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
137
+ "model.layers.12.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
138
+ "model.layers.12.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
139
+ "model.layers.12.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
140
+ "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
141
+ "model.layers.12.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
142
+ "model.layers.12.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
143
+ "model.layers.12.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
144
+ "model.layers.12.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
145
+ "model.layers.12.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
146
+ "model.layers.12.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
147
+ "model.layers.12.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
148
+ "model.layers.12.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
149
+ "model.layers.12.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
150
+ "model.layers.12.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
151
+ "model.layers.12.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
152
+ "model.layers.12.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
153
+ "model.layers.12.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
154
+ "model.layers.12.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
155
+ "model.layers.12.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
156
+ "model.layers.12.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
157
+ "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
158
+ "model.layers.13.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
159
+ "model.layers.13.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
160
+ "model.layers.13.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
161
+ "model.layers.13.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
162
+ "model.layers.13.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
163
+ "model.layers.13.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
164
+ "model.layers.13.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
165
+ "model.layers.13.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
166
+ "model.layers.13.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
167
+ "model.layers.13.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
168
+ "model.layers.13.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
169
+ "model.layers.13.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
170
+ "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
171
+ "model.layers.13.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
172
+ "model.layers.13.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
173
+ "model.layers.13.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
174
+ "model.layers.13.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
175
+ "model.layers.13.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
176
+ "model.layers.13.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
177
+ "model.layers.13.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
178
+ "model.layers.13.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
179
+ "model.layers.13.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
180
+ "model.layers.13.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
181
+ "model.layers.13.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
182
+ "model.layers.13.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
183
+ "model.layers.13.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
184
+ "model.layers.13.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
185
+ "model.layers.13.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
186
+ "model.layers.13.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
187
+ "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
188
+ "model.layers.14.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
189
+ "model.layers.14.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
190
+ "model.layers.14.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
191
+ "model.layers.14.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
192
+ "model.layers.14.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
193
+ "model.layers.14.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
194
+ "model.layers.14.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
195
+ "model.layers.14.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
196
+ "model.layers.14.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
197
+ "model.layers.14.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
198
+ "model.layers.14.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
199
+ "model.layers.14.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
200
+ "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
201
+ "model.layers.14.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
202
+ "model.layers.14.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
203
+ "model.layers.14.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
204
+ "model.layers.14.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
205
+ "model.layers.14.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
206
+ "model.layers.14.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
207
+ "model.layers.14.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
208
+ "model.layers.14.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
209
+ "model.layers.14.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
210
+ "model.layers.14.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
211
+ "model.layers.14.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
212
+ "model.layers.14.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
213
+ "model.layers.14.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
214
+ "model.layers.14.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
215
+ "model.layers.14.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
216
+ "model.layers.14.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
217
+ "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
218
+ "model.layers.15.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
219
+ "model.layers.15.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
220
+ "model.layers.15.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
221
+ "model.layers.15.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
222
+ "model.layers.15.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
223
+ "model.layers.15.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
224
+ "model.layers.15.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
225
+ "model.layers.15.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
226
+ "model.layers.15.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
227
+ "model.layers.15.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
228
+ "model.layers.15.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
229
+ "model.layers.15.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
230
+ "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
231
+ "model.layers.15.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
232
+ "model.layers.15.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
233
+ "model.layers.15.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
234
+ "model.layers.15.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
235
+ "model.layers.15.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
236
+ "model.layers.15.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
237
+ "model.layers.15.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
238
+ "model.layers.15.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
239
+ "model.layers.15.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
240
+ "model.layers.15.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
241
+ "model.layers.15.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
242
+ "model.layers.15.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
243
+ "model.layers.15.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
244
+ "model.layers.15.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
245
+ "model.layers.15.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
246
+ "model.layers.15.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
247
+ "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors",
248
+ "model.layers.16.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
249
+ "model.layers.16.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
250
+ "model.layers.16.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
251
+ "model.layers.16.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
252
+ "model.layers.16.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
253
+ "model.layers.16.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
254
+ "model.layers.16.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
255
+ "model.layers.16.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
256
+ "model.layers.16.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
257
+ "model.layers.16.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
258
+ "model.layers.16.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
259
+ "model.layers.16.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
260
+ "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
261
+ "model.layers.16.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
262
+ "model.layers.16.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
263
+ "model.layers.16.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
264
+ "model.layers.16.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
265
+ "model.layers.16.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
266
+ "model.layers.16.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
267
+ "model.layers.16.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
268
+ "model.layers.16.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
269
+ "model.layers.16.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
270
+ "model.layers.16.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
271
+ "model.layers.16.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
272
+ "model.layers.16.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
273
+ "model.layers.16.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
274
+ "model.layers.16.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
275
+ "model.layers.16.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
276
+ "model.layers.16.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
277
+ "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors",
278
+ "model.layers.17.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
279
+ "model.layers.17.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
280
+ "model.layers.17.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
281
+ "model.layers.17.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
282
+ "model.layers.17.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
283
+ "model.layers.17.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
284
+ "model.layers.17.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
285
+ "model.layers.17.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
286
+ "model.layers.17.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
287
+ "model.layers.17.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
288
+ "model.layers.17.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
289
+ "model.layers.17.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
290
+ "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
291
+ "model.layers.17.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
292
+ "model.layers.17.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
293
+ "model.layers.17.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
294
+ "model.layers.17.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
295
+ "model.layers.17.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
296
+ "model.layers.17.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
297
+ "model.layers.17.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
298
+ "model.layers.17.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
299
+ "model.layers.17.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
300
+ "model.layers.17.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
301
+ "model.layers.17.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
302
+ "model.layers.17.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
303
+ "model.layers.17.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
304
+ "model.layers.17.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
305
+ "model.layers.17.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
306
+ "model.layers.17.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
307
+ "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors",
308
+ "model.layers.18.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
309
+ "model.layers.18.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
310
+ "model.layers.18.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
311
+ "model.layers.18.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
312
+ "model.layers.18.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
313
+ "model.layers.18.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
314
+ "model.layers.18.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
315
+ "model.layers.18.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
316
+ "model.layers.18.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
317
+ "model.layers.18.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
318
+ "model.layers.18.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
319
+ "model.layers.18.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
320
+ "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
321
+ "model.layers.18.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
322
+ "model.layers.18.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
323
+ "model.layers.18.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
324
+ "model.layers.18.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
325
+ "model.layers.18.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
326
+ "model.layers.18.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
327
+ "model.layers.18.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
328
+ "model.layers.18.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
329
+ "model.layers.18.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
330
+ "model.layers.18.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
331
+ "model.layers.18.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
332
+ "model.layers.18.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
333
+ "model.layers.18.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
334
+ "model.layers.18.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
335
+ "model.layers.18.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
336
+ "model.layers.18.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
337
+ "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors",
338
+ "model.layers.19.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
339
+ "model.layers.19.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
340
+ "model.layers.19.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
341
+ "model.layers.19.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
342
+ "model.layers.19.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
343
+ "model.layers.19.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
344
+ "model.layers.19.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
345
+ "model.layers.19.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
346
+ "model.layers.19.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
347
+ "model.layers.19.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
348
+ "model.layers.19.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
349
+ "model.layers.19.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
350
+ "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
351
+ "model.layers.19.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
352
+ "model.layers.19.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
353
+ "model.layers.19.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
354
+ "model.layers.19.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
355
+ "model.layers.19.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
356
+ "model.layers.19.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
357
+ "model.layers.19.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
358
+ "model.layers.19.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
359
+ "model.layers.19.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
360
+ "model.layers.19.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
361
+ "model.layers.19.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
362
+ "model.layers.19.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
363
+ "model.layers.19.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
364
+ "model.layers.19.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
365
+ "model.layers.19.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
366
+ "model.layers.19.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
367
+ "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
368
+ "model.layers.2.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
369
+ "model.layers.2.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
370
+ "model.layers.2.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
371
+ "model.layers.2.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
372
+ "model.layers.2.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
373
+ "model.layers.2.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
374
+ "model.layers.2.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
375
+ "model.layers.2.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
376
+ "model.layers.2.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
377
+ "model.layers.2.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
378
+ "model.layers.2.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
379
+ "model.layers.2.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
380
+ "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
381
+ "model.layers.2.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
382
+ "model.layers.2.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
383
+ "model.layers.2.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
384
+ "model.layers.2.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
385
+ "model.layers.2.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
386
+ "model.layers.2.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
387
+ "model.layers.2.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
388
+ "model.layers.2.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
389
+ "model.layers.2.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
390
+ "model.layers.2.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
391
+ "model.layers.2.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
392
+ "model.layers.2.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
393
+ "model.layers.2.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
394
+ "model.layers.2.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
395
+ "model.layers.2.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
396
+ "model.layers.2.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
397
+ "model.layers.20.input_layernorm.weight": "model-00001-of-00002.safetensors",
398
+ "model.layers.20.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
399
+ "model.layers.20.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
400
+ "model.layers.20.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
401
+ "model.layers.20.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
402
+ "model.layers.20.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
403
+ "model.layers.20.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
404
+ "model.layers.20.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
405
+ "model.layers.20.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
406
+ "model.layers.20.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
407
+ "model.layers.20.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
408
+ "model.layers.20.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
409
+ "model.layers.20.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
410
+ "model.layers.20.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
411
+ "model.layers.20.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
412
+ "model.layers.20.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
413
+ "model.layers.20.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
414
+ "model.layers.20.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
415
+ "model.layers.20.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
416
+ "model.layers.20.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
417
+ "model.layers.20.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
418
+ "model.layers.20.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
419
+ "model.layers.20.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
420
+ "model.layers.20.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
421
+ "model.layers.20.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
422
+ "model.layers.20.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
423
+ "model.layers.20.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
424
+ "model.layers.20.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
425
+ "model.layers.20.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
426
+ "model.layers.20.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
427
+ "model.layers.21.input_layernorm.weight": "model-00001-of-00002.safetensors",
428
+ "model.layers.21.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
429
+ "model.layers.21.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
430
+ "model.layers.21.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
431
+ "model.layers.21.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
432
+ "model.layers.21.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
433
+ "model.layers.21.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
434
+ "model.layers.21.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
435
+ "model.layers.21.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
436
+ "model.layers.21.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
437
+ "model.layers.21.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
438
+ "model.layers.21.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
439
+ "model.layers.21.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
440
+ "model.layers.21.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
441
+ "model.layers.21.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
442
+ "model.layers.21.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
443
+ "model.layers.21.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
444
+ "model.layers.21.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
445
+ "model.layers.21.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
446
+ "model.layers.21.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
447
+ "model.layers.21.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
448
+ "model.layers.21.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
449
+ "model.layers.21.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
450
+ "model.layers.21.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
451
+ "model.layers.21.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
452
+ "model.layers.21.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
453
+ "model.layers.21.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
454
+ "model.layers.21.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
455
+ "model.layers.21.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
456
+ "model.layers.21.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
457
+ "model.layers.22.input_layernorm.weight": "model-00001-of-00002.safetensors",
458
+ "model.layers.22.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
459
+ "model.layers.22.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
460
+ "model.layers.22.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
461
+ "model.layers.22.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
462
+ "model.layers.22.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
463
+ "model.layers.22.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
464
+ "model.layers.22.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
465
+ "model.layers.22.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
466
+ "model.layers.22.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
467
+ "model.layers.22.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
468
+ "model.layers.22.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
469
+ "model.layers.22.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
470
+ "model.layers.22.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
471
+ "model.layers.22.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
472
+ "model.layers.22.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
473
+ "model.layers.22.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
474
+ "model.layers.22.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
475
+ "model.layers.22.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
476
+ "model.layers.22.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
477
+ "model.layers.22.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
478
+ "model.layers.22.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
479
+ "model.layers.22.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
480
+ "model.layers.22.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
481
+ "model.layers.22.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
482
+ "model.layers.22.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
483
+ "model.layers.22.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
484
+ "model.layers.22.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
485
+ "model.layers.22.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
486
+ "model.layers.22.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
487
+ "model.layers.23.input_layernorm.weight": "model-00001-of-00002.safetensors",
488
+ "model.layers.23.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
489
+ "model.layers.23.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
490
+ "model.layers.23.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
491
+ "model.layers.23.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
492
+ "model.layers.23.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
493
+ "model.layers.23.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
494
+ "model.layers.23.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
495
+ "model.layers.23.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
496
+ "model.layers.23.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
497
+ "model.layers.23.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
498
+ "model.layers.23.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
499
+ "model.layers.23.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
500
+ "model.layers.23.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
501
+ "model.layers.23.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
502
+ "model.layers.23.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
503
+ "model.layers.23.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
504
+ "model.layers.23.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
505
+ "model.layers.23.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
506
+ "model.layers.23.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
507
+ "model.layers.23.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
508
+ "model.layers.23.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
509
+ "model.layers.23.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
510
+ "model.layers.23.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
511
+ "model.layers.23.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
512
+ "model.layers.23.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
513
+ "model.layers.23.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
514
+ "model.layers.23.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
515
+ "model.layers.23.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
516
+ "model.layers.23.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
517
+ "model.layers.24.input_layernorm.weight": "model-00001-of-00002.safetensors",
518
+ "model.layers.24.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
519
+ "model.layers.24.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
520
+ "model.layers.24.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
521
+ "model.layers.24.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
522
+ "model.layers.24.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
523
+ "model.layers.24.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
524
+ "model.layers.24.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
525
+ "model.layers.24.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
526
+ "model.layers.24.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
527
+ "model.layers.24.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
528
+ "model.layers.24.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
529
+ "model.layers.24.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
530
+ "model.layers.24.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
531
+ "model.layers.24.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
532
+ "model.layers.24.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
533
+ "model.layers.24.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
534
+ "model.layers.24.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
535
+ "model.layers.24.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
536
+ "model.layers.24.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
537
+ "model.layers.24.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
538
+ "model.layers.24.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
539
+ "model.layers.24.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
540
+ "model.layers.24.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
541
+ "model.layers.24.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
542
+ "model.layers.24.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
543
+ "model.layers.24.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
544
+ "model.layers.24.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
545
+ "model.layers.24.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
546
+ "model.layers.24.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
547
+ "model.layers.25.input_layernorm.weight": "model-00001-of-00002.safetensors",
548
+ "model.layers.25.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
549
+ "model.layers.25.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
550
+ "model.layers.25.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
551
+ "model.layers.25.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
552
+ "model.layers.25.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
553
+ "model.layers.25.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
554
+ "model.layers.25.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
555
+ "model.layers.25.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
556
+ "model.layers.25.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
557
+ "model.layers.25.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
558
+ "model.layers.25.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
559
+ "model.layers.25.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
560
+ "model.layers.25.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
561
+ "model.layers.25.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
562
+ "model.layers.25.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
563
+ "model.layers.25.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
564
+ "model.layers.25.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
565
+ "model.layers.25.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
566
+ "model.layers.25.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
567
+ "model.layers.25.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
568
+ "model.layers.25.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
569
+ "model.layers.25.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
570
+ "model.layers.25.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
571
+ "model.layers.25.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
572
+ "model.layers.25.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
573
+ "model.layers.25.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
574
+ "model.layers.25.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
575
+ "model.layers.25.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
576
+ "model.layers.25.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
577
+ "model.layers.26.input_layernorm.weight": "model-00001-of-00002.safetensors",
578
+ "model.layers.26.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
579
+ "model.layers.26.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
580
+ "model.layers.26.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
581
+ "model.layers.26.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
582
+ "model.layers.26.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
583
+ "model.layers.26.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
584
+ "model.layers.26.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
585
+ "model.layers.26.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
586
+ "model.layers.26.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
587
+ "model.layers.26.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
588
+ "model.layers.26.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
589
+ "model.layers.26.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
590
+ "model.layers.26.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
591
+ "model.layers.26.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
592
+ "model.layers.26.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
593
+ "model.layers.26.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
594
+ "model.layers.26.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
595
+ "model.layers.26.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
596
+ "model.layers.26.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
597
+ "model.layers.26.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
598
+ "model.layers.26.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
599
+ "model.layers.26.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
600
+ "model.layers.26.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
601
+ "model.layers.26.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
602
+ "model.layers.26.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
603
+ "model.layers.26.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
604
+ "model.layers.26.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
605
+ "model.layers.26.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
606
+ "model.layers.26.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
607
+ "model.layers.27.input_layernorm.weight": "model-00001-of-00002.safetensors",
608
+ "model.layers.27.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
609
+ "model.layers.27.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
610
+ "model.layers.27.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
611
+ "model.layers.27.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
612
+ "model.layers.27.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
613
+ "model.layers.27.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
614
+ "model.layers.27.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
615
+ "model.layers.27.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
616
+ "model.layers.27.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
617
+ "model.layers.27.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
618
+ "model.layers.27.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
619
+ "model.layers.27.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
620
+ "model.layers.27.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
621
+ "model.layers.27.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
622
+ "model.layers.27.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
623
+ "model.layers.27.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
624
+ "model.layers.27.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
625
+ "model.layers.27.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
626
+ "model.layers.27.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
627
+ "model.layers.27.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
628
+ "model.layers.27.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
629
+ "model.layers.27.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
630
+ "model.layers.27.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
631
+ "model.layers.27.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
632
+ "model.layers.27.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
633
+ "model.layers.27.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
634
+ "model.layers.27.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
635
+ "model.layers.27.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
636
+ "model.layers.27.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
637
+ "model.layers.28.input_layernorm.weight": "model-00001-of-00002.safetensors",
638
+ "model.layers.28.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
639
+ "model.layers.28.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
640
+ "model.layers.28.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
641
+ "model.layers.28.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
642
+ "model.layers.28.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
643
+ "model.layers.28.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
644
+ "model.layers.28.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
645
+ "model.layers.28.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
646
+ "model.layers.28.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
647
+ "model.layers.28.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
648
+ "model.layers.28.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
649
+ "model.layers.28.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
650
+ "model.layers.28.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
651
+ "model.layers.28.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
652
+ "model.layers.28.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
653
+ "model.layers.28.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
654
+ "model.layers.28.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
655
+ "model.layers.28.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
656
+ "model.layers.28.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
657
+ "model.layers.28.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
658
+ "model.layers.28.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
659
+ "model.layers.28.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
660
+ "model.layers.28.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
661
+ "model.layers.28.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
662
+ "model.layers.28.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
663
+ "model.layers.28.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
664
+ "model.layers.28.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
665
+ "model.layers.28.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
666
+ "model.layers.28.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
667
+ "model.layers.29.input_layernorm.weight": "model-00001-of-00002.safetensors",
668
+ "model.layers.29.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
669
+ "model.layers.29.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
670
+ "model.layers.29.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
671
+ "model.layers.29.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
672
+ "model.layers.29.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
673
+ "model.layers.29.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
674
+ "model.layers.29.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
675
+ "model.layers.29.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
676
+ "model.layers.29.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
677
+ "model.layers.29.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
678
+ "model.layers.29.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
679
+ "model.layers.29.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
680
+ "model.layers.29.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
681
+ "model.layers.29.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
682
+ "model.layers.29.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
683
+ "model.layers.29.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
684
+ "model.layers.29.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
685
+ "model.layers.29.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
686
+ "model.layers.29.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
687
+ "model.layers.29.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
688
+ "model.layers.29.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
689
+ "model.layers.29.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
690
+ "model.layers.29.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
691
+ "model.layers.29.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
692
+ "model.layers.29.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
693
+ "model.layers.29.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
694
+ "model.layers.29.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
695
+ "model.layers.29.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
696
+ "model.layers.29.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
697
+ "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
698
+ "model.layers.3.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
699
+ "model.layers.3.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
700
+ "model.layers.3.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
701
+ "model.layers.3.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
702
+ "model.layers.3.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
703
+ "model.layers.3.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
704
+ "model.layers.3.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
705
+ "model.layers.3.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
706
+ "model.layers.3.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
707
+ "model.layers.3.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
708
+ "model.layers.3.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
709
+ "model.layers.3.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
710
+ "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
711
+ "model.layers.3.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
712
+ "model.layers.3.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
713
+ "model.layers.3.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
714
+ "model.layers.3.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
715
+ "model.layers.3.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
716
+ "model.layers.3.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
717
+ "model.layers.3.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
718
+ "model.layers.3.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
719
+ "model.layers.3.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
720
+ "model.layers.3.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
721
+ "model.layers.3.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
722
+ "model.layers.3.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
723
+ "model.layers.3.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
724
+ "model.layers.3.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
725
+ "model.layers.3.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
726
+ "model.layers.3.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
727
+ "model.layers.30.input_layernorm.weight": "model-00001-of-00002.safetensors",
728
+ "model.layers.30.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
729
+ "model.layers.30.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
730
+ "model.layers.30.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
731
+ "model.layers.30.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
732
+ "model.layers.30.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
733
+ "model.layers.30.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
734
+ "model.layers.30.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
735
+ "model.layers.30.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
736
+ "model.layers.30.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
737
+ "model.layers.30.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
738
+ "model.layers.30.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
739
+ "model.layers.30.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
740
+ "model.layers.30.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
741
+ "model.layers.30.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
742
+ "model.layers.30.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
743
+ "model.layers.30.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
744
+ "model.layers.30.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
745
+ "model.layers.30.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
746
+ "model.layers.30.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
747
+ "model.layers.30.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
748
+ "model.layers.30.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
749
+ "model.layers.30.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
750
+ "model.layers.30.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
751
+ "model.layers.30.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
752
+ "model.layers.30.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
753
+ "model.layers.30.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
754
+ "model.layers.30.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
755
+ "model.layers.30.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
756
+ "model.layers.30.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
757
+ "model.layers.31.input_layernorm.weight": "model-00001-of-00002.safetensors",
758
+ "model.layers.31.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
759
+ "model.layers.31.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
760
+ "model.layers.31.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
761
+ "model.layers.31.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
762
+ "model.layers.31.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
763
+ "model.layers.31.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
764
+ "model.layers.31.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
765
+ "model.layers.31.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
766
+ "model.layers.31.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
767
+ "model.layers.31.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
768
+ "model.layers.31.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
769
+ "model.layers.31.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
770
+ "model.layers.31.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
771
+ "model.layers.31.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
772
+ "model.layers.31.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
773
+ "model.layers.31.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
774
+ "model.layers.31.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
775
+ "model.layers.31.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
776
+ "model.layers.31.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
777
+ "model.layers.31.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
778
+ "model.layers.31.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
779
+ "model.layers.31.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
780
+ "model.layers.31.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
781
+ "model.layers.31.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
782
+ "model.layers.31.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
783
+ "model.layers.31.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
784
+ "model.layers.31.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
785
+ "model.layers.31.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
786
+ "model.layers.31.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
787
+ "model.layers.32.input_layernorm.weight": "model-00001-of-00002.safetensors",
788
+ "model.layers.32.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
789
+ "model.layers.32.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
790
+ "model.layers.32.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
791
+ "model.layers.32.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
792
+ "model.layers.32.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
793
+ "model.layers.32.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
794
+ "model.layers.32.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
795
+ "model.layers.32.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
796
+ "model.layers.32.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
797
+ "model.layers.32.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
798
+ "model.layers.32.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
799
+ "model.layers.32.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
800
+ "model.layers.32.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
801
+ "model.layers.32.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
802
+ "model.layers.32.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
803
+ "model.layers.32.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
804
+ "model.layers.32.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
805
+ "model.layers.32.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
806
+ "model.layers.32.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
807
+ "model.layers.32.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
808
+ "model.layers.32.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
809
+ "model.layers.32.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
810
+ "model.layers.32.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
811
+ "model.layers.32.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
812
+ "model.layers.32.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
813
+ "model.layers.32.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
814
+ "model.layers.32.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
815
+ "model.layers.32.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
816
+ "model.layers.32.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
817
+ "model.layers.33.input_layernorm.weight": "model-00001-of-00002.safetensors",
818
+ "model.layers.33.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
819
+ "model.layers.33.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
820
+ "model.layers.33.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
821
+ "model.layers.33.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
822
+ "model.layers.33.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
823
+ "model.layers.33.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
824
+ "model.layers.33.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
825
+ "model.layers.33.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
826
+ "model.layers.33.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
827
+ "model.layers.33.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
828
+ "model.layers.33.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
829
+ "model.layers.33.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
830
+ "model.layers.33.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
831
+ "model.layers.33.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
832
+ "model.layers.33.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
833
+ "model.layers.33.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
834
+ "model.layers.33.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
835
+ "model.layers.33.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
836
+ "model.layers.33.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
837
+ "model.layers.33.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
838
+ "model.layers.33.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
839
+ "model.layers.33.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
840
+ "model.layers.33.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
841
+ "model.layers.33.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
842
+ "model.layers.33.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
843
+ "model.layers.33.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
844
+ "model.layers.33.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
845
+ "model.layers.33.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
846
+ "model.layers.33.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
847
+ "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
848
+ "model.layers.4.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
849
+ "model.layers.4.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
850
+ "model.layers.4.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
851
+ "model.layers.4.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
852
+ "model.layers.4.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
853
+ "model.layers.4.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
854
+ "model.layers.4.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
855
+ "model.layers.4.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
856
+ "model.layers.4.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
857
+ "model.layers.4.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
858
+ "model.layers.4.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
859
+ "model.layers.4.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
860
+ "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
861
+ "model.layers.4.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
862
+ "model.layers.4.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
863
+ "model.layers.4.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
864
+ "model.layers.4.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
865
+ "model.layers.4.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
866
+ "model.layers.4.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
867
+ "model.layers.4.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
868
+ "model.layers.4.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
869
+ "model.layers.4.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
870
+ "model.layers.4.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
871
+ "model.layers.4.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
872
+ "model.layers.4.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
873
+ "model.layers.4.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
874
+ "model.layers.4.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
875
+ "model.layers.4.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
876
+ "model.layers.4.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
877
+ "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
878
+ "model.layers.5.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
879
+ "model.layers.5.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
880
+ "model.layers.5.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
881
+ "model.layers.5.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
882
+ "model.layers.5.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
883
+ "model.layers.5.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
884
+ "model.layers.5.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
885
+ "model.layers.5.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
886
+ "model.layers.5.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
887
+ "model.layers.5.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
888
+ "model.layers.5.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
889
+ "model.layers.5.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
890
+ "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
891
+ "model.layers.5.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
892
+ "model.layers.5.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
893
+ "model.layers.5.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
894
+ "model.layers.5.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
895
+ "model.layers.5.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
896
+ "model.layers.5.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
897
+ "model.layers.5.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
898
+ "model.layers.5.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
899
+ "model.layers.5.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
900
+ "model.layers.5.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
901
+ "model.layers.5.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
902
+ "model.layers.5.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
903
+ "model.layers.5.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
904
+ "model.layers.5.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
905
+ "model.layers.5.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
906
+ "model.layers.5.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
907
+ "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
908
+ "model.layers.6.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
909
+ "model.layers.6.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
910
+ "model.layers.6.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
911
+ "model.layers.6.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
912
+ "model.layers.6.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
913
+ "model.layers.6.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
914
+ "model.layers.6.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
915
+ "model.layers.6.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
916
+ "model.layers.6.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
917
+ "model.layers.6.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
918
+ "model.layers.6.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
919
+ "model.layers.6.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
920
+ "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
921
+ "model.layers.6.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
922
+ "model.layers.6.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
923
+ "model.layers.6.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
924
+ "model.layers.6.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
925
+ "model.layers.6.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
926
+ "model.layers.6.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
927
+ "model.layers.6.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
928
+ "model.layers.6.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
929
+ "model.layers.6.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
930
+ "model.layers.6.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
931
+ "model.layers.6.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
932
+ "model.layers.6.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
933
+ "model.layers.6.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
934
+ "model.layers.6.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
935
+ "model.layers.6.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
936
+ "model.layers.6.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
937
+ "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
938
+ "model.layers.7.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
939
+ "model.layers.7.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
940
+ "model.layers.7.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
941
+ "model.layers.7.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
942
+ "model.layers.7.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
943
+ "model.layers.7.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
944
+ "model.layers.7.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
945
+ "model.layers.7.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
946
+ "model.layers.7.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
947
+ "model.layers.7.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
948
+ "model.layers.7.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
949
+ "model.layers.7.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
950
+ "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
951
+ "model.layers.7.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
952
+ "model.layers.7.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
953
+ "model.layers.7.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
954
+ "model.layers.7.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
955
+ "model.layers.7.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
956
+ "model.layers.7.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
957
+ "model.layers.7.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
958
+ "model.layers.7.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
959
+ "model.layers.7.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
960
+ "model.layers.7.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
961
+ "model.layers.7.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
962
+ "model.layers.7.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
963
+ "model.layers.7.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
964
+ "model.layers.7.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
965
+ "model.layers.7.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
966
+ "model.layers.7.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
967
+ "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
968
+ "model.layers.8.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
969
+ "model.layers.8.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
970
+ "model.layers.8.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
971
+ "model.layers.8.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
972
+ "model.layers.8.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
973
+ "model.layers.8.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
974
+ "model.layers.8.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
975
+ "model.layers.8.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
976
+ "model.layers.8.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
977
+ "model.layers.8.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
978
+ "model.layers.8.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
979
+ "model.layers.8.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
980
+ "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
981
+ "model.layers.8.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
982
+ "model.layers.8.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
983
+ "model.layers.8.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
984
+ "model.layers.8.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
985
+ "model.layers.8.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
986
+ "model.layers.8.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
987
+ "model.layers.8.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
988
+ "model.layers.8.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
989
+ "model.layers.8.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
990
+ "model.layers.8.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
991
+ "model.layers.8.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
992
+ "model.layers.8.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
993
+ "model.layers.8.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
994
+ "model.layers.8.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
995
+ "model.layers.8.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
996
+ "model.layers.8.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
997
+ "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
998
+ "model.layers.9.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
999
+ "model.layers.9.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
1000
+ "model.layers.9.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
1001
+ "model.layers.9.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
1002
+ "model.layers.9.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
1003
+ "model.layers.9.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
1004
+ "model.layers.9.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
1005
+ "model.layers.9.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
1006
+ "model.layers.9.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
1007
+ "model.layers.9.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
1008
+ "model.layers.9.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
1009
+ "model.layers.9.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
1010
+ "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
1011
+ "model.layers.9.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
1012
+ "model.layers.9.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
1013
+ "model.layers.9.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
1014
+ "model.layers.9.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
1015
+ "model.layers.9.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
1016
+ "model.layers.9.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
1017
+ "model.layers.9.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
1018
+ "model.layers.9.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
1019
+ "model.layers.9.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
1020
+ "model.layers.9.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
1021
+ "model.layers.9.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
1022
+ "model.layers.9.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
1023
+ "model.layers.9.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
1024
+ "model.layers.9.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
1025
+ "model.layers.9.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
1026
+ "model.layers.9.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
1027
+ "model.norm.weight": "model-00001-of-00002.safetensors",
1028
+ "model.rotary_emb.inv_freq": "model-00002-of-00002.safetensors"
1029
+ }
1030
+ }
quant_log.csv ADDED
@@ -0,0 +1,239 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ layer,module,loss,samples,damp,time
2
+ 0,self_attn.q_proj,0.0000000075,0.05000,3.008
3
+ 0,self_attn.k_proj,0.0000000030,0.05000,3.031
4
+ 0,self_attn.v_proj,0.0000000002,0.05000,3.052
5
+ 0,self_attn.o_proj,0.0000000000,0.05000,1.200
6
+ 0,mlp.up_proj,0.0000000054,0.05000,1.456
7
+ 0,mlp.gate_proj,0.0000000059,0.05000,1.473
8
+ 0,mlp.down_proj,0.0000000113,0.05000,2.276
9
+ 1,self_attn.q_proj,0.0000000300,0.05000,2.750
10
+ 1,self_attn.v_proj,0.0000000024,0.05000,2.774
11
+ 1,self_attn.k_proj,0.0000000121,0.05000,2.783
12
+ 1,self_attn.o_proj,0.0000000000,0.05000,1.217
13
+ 1,mlp.gate_proj,0.0000000097,0.05000,2.606
14
+ 1,mlp.up_proj,0.0000000087,0.05000,2.636
15
+ 1,mlp.down_proj,0.0000000003,0.05000,2.435
16
+ 2,self_attn.v_proj,0.0000000030,0.05000,2.871
17
+ 2,self_attn.k_proj,0.0000000112,0.05000,2.881
18
+ 2,self_attn.q_proj,0.0000000293,0.05000,2.888
19
+ 2,self_attn.o_proj,0.0000000001,0.05000,1.247
20
+ 2,mlp.gate_proj,0.0000000248,0.05000,1.672
21
+ 2,mlp.up_proj,0.0000000208,0.05000,1.680
22
+ 2,mlp.down_proj,0.0000000163,0.05000,2.173
23
+ 3,self_attn.q_proj,0.0000000690,0.05000,2.740
24
+ 3,self_attn.v_proj,0.0000000055,0.05000,2.745
25
+ 3,self_attn.k_proj,0.0000000241,0.05000,2.754
26
+ 3,self_attn.o_proj,0.0000000001,0.05000,1.247
27
+ 3,mlp.gate_proj,0.0000000285,0.05000,1.467
28
+ 3,mlp.up_proj,0.0000000232,0.05000,1.512
29
+ 3,mlp.down_proj,0.0000000009,0.05000,2.859
30
+ 4,self_attn.q_proj,0.0000000442,0.05000,2.484
31
+ 4,self_attn.v_proj,0.0000000051,0.05000,2.488
32
+ 4,self_attn.k_proj,0.0000000158,0.05000,2.498
33
+ 4,self_attn.o_proj,0.0000000002,0.05000,1.251
34
+ 4,mlp.up_proj,0.0000000274,0.05000,1.512
35
+ 4,mlp.gate_proj,0.0000000344,0.05000,1.520
36
+ 4,mlp.down_proj,0.0000000013,0.05000,2.869
37
+ 5,self_attn.q_proj,0.0000000603,0.05000,2.531
38
+ 5,self_attn.k_proj,0.0000000206,0.05000,2.571
39
+ 5,self_attn.v_proj,0.0000000055,0.05000,2.579
40
+ 5,self_attn.o_proj,0.0000000003,0.05000,1.312
41
+ 5,mlp.up_proj,0.0000000321,0.05000,2.469
42
+ 5,mlp.gate_proj,0.0000000390,0.05000,2.480
43
+ 5,mlp.down_proj,0.0000000016,0.05000,2.164
44
+ 6,self_attn.q_proj,0.0000000641,0.05000,3.809
45
+ 6,self_attn.k_proj,0.0000000242,0.05000,3.838
46
+ 6,self_attn.v_proj,0.0000000058,0.05000,3.922
47
+ 6,self_attn.o_proj,0.0000000006,0.05000,1.298
48
+ 6,mlp.up_proj,0.0000000361,0.05000,1.936
49
+ 6,mlp.gate_proj,0.0000000441,0.05000,1.941
50
+ 6,mlp.down_proj,0.0000000020,0.05000,1.926
51
+ 7,self_attn.q_proj,0.0000000627,0.05000,2.615
52
+ 7,self_attn.k_proj,0.0000000221,0.05000,2.627
53
+ 7,self_attn.v_proj,0.0000000070,0.05000,2.644
54
+ 7,self_attn.o_proj,0.0000000008,0.05000,0.956
55
+ 7,mlp.up_proj,0.0000000377,0.05000,1.592
56
+ 7,mlp.gate_proj,0.0000000447,0.05000,1.611
57
+ 7,mlp.down_proj,0.0000000021,0.05000,2.775
58
+ 8,self_attn.q_proj,0.0000000653,0.05000,3.218
59
+ 8,self_attn.k_proj,0.0000000243,0.05000,3.319
60
+ 8,self_attn.v_proj,0.0000000069,0.05000,3.373
61
+ 8,self_attn.o_proj,0.0000000009,0.05000,1.220
62
+ 8,mlp.up_proj,0.0000000404,0.05000,1.542
63
+ 8,mlp.gate_proj,0.0000000455,0.05000,1.555
64
+ 8,mlp.down_proj,0.0000000023,0.05000,2.556
65
+ 9,self_attn.q_proj,0.0000000673,0.05000,2.872
66
+ 9,self_attn.k_proj,0.0000000240,0.05000,2.893
67
+ 9,self_attn.v_proj,0.0000000082,0.05000,2.906
68
+ 9,self_attn.o_proj,0.0000000013,0.05000,1.218
69
+ 9,mlp.gate_proj,0.0000000456,0.05000,1.421
70
+ 9,mlp.up_proj,0.0000000422,0.05000,1.434
71
+ 9,mlp.down_proj,0.0000000027,0.05000,2.207
72
+ 10,self_attn.q_proj,0.0000000675,0.05000,2.962
73
+ 10,self_attn.k_proj,0.0000000243,0.05000,3.045
74
+ 10,self_attn.v_proj,0.0000000080,0.05000,3.081
75
+ 10,self_attn.o_proj,0.0000000014,0.05000,1.270
76
+ 10,mlp.gate_proj,0.0000000466,0.05000,2.136
77
+ 10,mlp.up_proj,0.0000000431,0.05000,2.152
78
+ 10,mlp.down_proj,0.0000000027,0.05000,2.642
79
+ 11,self_attn.v_proj,0.0000000084,0.05000,3.812
80
+ 11,self_attn.k_proj,0.0000000279,0.05000,3.848
81
+ 11,self_attn.q_proj,0.0000000745,0.05000,3.951
82
+ 11,self_attn.o_proj,0.0000000015,0.05000,1.160
83
+ 11,mlp.up_proj,0.0000000442,0.05000,1.585
84
+ 11,mlp.gate_proj,0.0000000537,0.05000,1.601
85
+ 11,mlp.down_proj,0.0000000028,0.05000,2.170
86
+ 12,self_attn.v_proj,0.0000000095,0.05000,2.501
87
+ 12,self_attn.k_proj,0.0000000247,0.05000,2.519
88
+ 12,self_attn.q_proj,0.0000000698,0.05000,2.529
89
+ 12,self_attn.o_proj,0.0000000014,0.05000,1.287
90
+ 12,mlp.up_proj,0.0000000480,0.05000,1.503
91
+ 12,mlp.gate_proj,0.0000000539,0.05000,1.529
92
+ 12,mlp.down_proj,0.0000000031,0.05000,2.669
93
+ 13,self_attn.q_proj,0.0000000884,0.05000,2.548
94
+ 13,self_attn.v_proj,0.0000000099,0.05000,2.572
95
+ 13,self_attn.k_proj,0.0000000310,0.05000,2.574
96
+ 13,self_attn.o_proj,0.0000000016,0.05000,1.187
97
+ 13,mlp.up_proj,0.0000000504,0.05000,2.487
98
+ 13,mlp.gate_proj,0.0000000544,0.05000,2.501
99
+ 13,mlp.down_proj,0.0000000052,0.05000,2.544
100
+ 14,self_attn.q_proj,0.0000000950,0.05000,3.766
101
+ 14,self_attn.k_proj,0.0000000318,0.05000,3.796
102
+ 14,self_attn.v_proj,0.0000000102,0.05000,3.879
103
+ 14,self_attn.o_proj,0.0000000017,0.05000,1.170
104
+ 14,mlp.gate_proj,0.0000000566,0.05000,1.556
105
+ 14,mlp.up_proj,0.0000000554,0.05000,1.584
106
+ 14,mlp.down_proj,0.0000000039,0.05000,2.197
107
+ 15,self_attn.v_proj,0.0000000092,0.05000,2.836
108
+ 15,self_attn.q_proj,0.0000000792,0.05000,2.842
109
+ 15,self_attn.k_proj,0.0000000302,0.05000,2.860
110
+ 15,self_attn.o_proj,0.0000000017,0.05000,1.298
111
+ 15,mlp.gate_proj,0.0000000563,0.05000,1.485
112
+ 15,mlp.up_proj,0.0000000563,0.05000,1.498
113
+ 15,mlp.down_proj,0.0000000041,0.05000,2.854
114
+ 16,self_attn.q_proj,0.0000000807,0.05000,2.401
115
+ 16,self_attn.k_proj,0.0000000299,0.05000,2.433
116
+ 16,self_attn.v_proj,0.0000000123,0.05000,2.447
117
+ 16,self_attn.o_proj,0.0000000022,0.05000,1.260
118
+ 16,mlp.gate_proj,0.0000000590,0.05000,1.909
119
+ 16,mlp.up_proj,0.0000000594,0.05000,1.936
120
+ 16,mlp.down_proj,0.0000000046,0.05000,2.199
121
+ 17,self_attn.q_proj,0.0000000812,0.05000,3.101
122
+ 17,self_attn.v_proj,0.0000000129,0.05000,3.159
123
+ 17,self_attn.k_proj,0.0000000285,0.05000,3.158
124
+ 17,self_attn.o_proj,0.0000000027,0.05000,1.260
125
+ 17,mlp.up_proj,0.0000000625,0.05000,1.718
126
+ 17,mlp.gate_proj,0.0000000618,0.05000,1.748
127
+ 17,mlp.down_proj,0.0000000059,0.05000,2.384
128
+ 18,self_attn.v_proj,0.0000000151,0.05000,2.679
129
+ 18,self_attn.k_proj,0.0000000320,0.05000,2.693
130
+ 18,self_attn.q_proj,0.0000000885,0.05000,2.722
131
+ 18,self_attn.o_proj,0.0000000029,0.05000,1.201
132
+ 18,mlp.up_proj,0.0000000651,0.05000,1.546
133
+ 18,mlp.gate_proj,0.0000000675,0.05000,1.563
134
+ 18,mlp.down_proj,0.0000000067,0.05000,2.212
135
+ 19,self_attn.q_proj,0.0000000979,0.05000,3.052
136
+ 19,self_attn.v_proj,0.0000000190,0.05000,3.059
137
+ 19,self_attn.k_proj,0.0000000297,0.05000,3.083
138
+ 19,self_attn.o_proj,0.0000000024,0.05000,1.308
139
+ 19,mlp.gate_proj,0.0000000689,0.05000,1.671
140
+ 19,mlp.up_proj,0.0000000664,0.05000,1.682
141
+ 19,mlp.down_proj,0.0000000077,0.05000,2.353
142
+ 20,self_attn.q_proj,0.0000000935,0.05000,3.327
143
+ 20,self_attn.k_proj,0.0000000300,0.05000,3.385
144
+ 20,self_attn.v_proj,0.0000000177,0.05000,3.403
145
+ 20,self_attn.o_proj,0.0000000021,0.05000,1.283
146
+ 20,mlp.gate_proj,0.0000000806,0.05000,1.467
147
+ 20,mlp.up_proj,0.0000000756,0.05000,1.479
148
+ 20,mlp.down_proj,0.0000000089,0.05000,2.752
149
+ 21,self_attn.k_proj,0.0000000291,0.05000,3.551
150
+ 21,self_attn.v_proj,0.0000000160,0.05000,3.591
151
+ 21,self_attn.q_proj,0.0000000912,0.05000,3.676
152
+ 21,self_attn.o_proj,0.0000000016,0.05000,1.171
153
+ 21,mlp.up_proj,0.0000000838,0.05000,2.555
154
+ 21,mlp.gate_proj,0.0000000902,0.05000,2.602
155
+ 21,mlp.down_proj,0.0000000101,0.05000,2.382
156
+ 22,self_attn.q_proj,0.0000000906,0.05000,3.905
157
+ 22,self_attn.k_proj,0.0000000262,0.05000,3.912
158
+ 22,self_attn.v_proj,0.0000000213,0.05000,4.015
159
+ 22,self_attn.o_proj,0.0000000024,0.05000,0.851
160
+ 22,mlp.up_proj,0.0000000967,0.05000,1.559
161
+ 22,mlp.gate_proj,0.0000001041,0.05000,1.565
162
+ 22,mlp.down_proj,0.0000000127,0.05000,2.276
163
+ 23,self_attn.v_proj,0.0000000217,0.05000,3.784
164
+ 23,self_attn.k_proj,0.0000000295,0.05000,3.809
165
+ 23,self_attn.q_proj,0.0000000979,0.05000,3.914
166
+ 23,self_attn.o_proj,0.0000000019,0.05000,1.159
167
+ 23,mlp.up_proj,0.0000001080,0.05000,2.357
168
+ 23,mlp.gate_proj,0.0000001121,0.05000,2.384
169
+ 23,mlp.down_proj,0.0000000163,0.05000,2.817
170
+ 24,self_attn.q_proj,0.0000001070,0.05000,3.759
171
+ 24,self_attn.k_proj,0.0000000320,0.05000,3.819
172
+ 24,self_attn.v_proj,0.0000000248,0.05000,3.908
173
+ 24,self_attn.o_proj,0.0000000022,0.05000,1.191
174
+ 24,mlp.up_proj,0.0000001213,0.05000,1.383
175
+ 24,mlp.gate_proj,0.0000001228,0.05000,1.391
176
+ 24,mlp.down_proj,0.0000000182,0.05000,2.273
177
+ 25,self_attn.k_proj,0.0000000321,0.05000,3.559
178
+ 25,self_attn.v_proj,0.0000000275,0.05000,3.625
179
+ 25,self_attn.q_proj,0.0000001104,0.05000,3.706
180
+ 25,self_attn.o_proj,0.0000000023,0.05000,1.030
181
+ 25,mlp.up_proj,0.0000001367,0.05000,2.500
182
+ 25,mlp.gate_proj,0.0000001344,0.05000,2.523
183
+ 25,mlp.down_proj,0.0000000330,0.05000,2.737
184
+ 26,self_attn.q_proj,0.0000001115,0.05000,3.848
185
+ 26,self_attn.k_proj,0.0000000315,0.05000,3.908
186
+ 26,self_attn.v_proj,0.0000000338,0.05000,4.023
187
+ 26,self_attn.o_proj,0.0000000027,0.05000,0.856
188
+ 26,mlp.up_proj,0.0000001578,0.05000,1.510
189
+ 26,mlp.gate_proj,0.0000001507,0.05000,1.538
190
+ 26,mlp.down_proj,0.0000000300,0.05000,2.150
191
+ 27,self_attn.v_proj,0.0000000335,0.05000,3.903
192
+ 27,self_attn.k_proj,0.0000000346,0.05000,3.934
193
+ 27,self_attn.q_proj,0.0000001312,0.05000,4.007
194
+ 27,self_attn.o_proj,0.0000000041,0.05000,1.299
195
+ 27,mlp.up_proj,0.0000001809,0.05000,2.707
196
+ 27,mlp.gate_proj,0.0000001781,0.05000,2.739
197
+ 27,mlp.down_proj,0.0000000425,0.05000,2.745
198
+ 28,self_attn.q_proj,0.0000001430,0.05000,3.753
199
+ 28,self_attn.k_proj,0.0000000326,0.05000,3.801
200
+ 28,self_attn.v_proj,0.0000000781,0.05000,3.903
201
+ 28,self_attn.o_proj,0.0000000079,0.05000,1.300
202
+ 28,mlp.up_proj,0.0000002043,0.05000,2.586
203
+ 28,mlp.gate_proj,0.0000002063,0.05000,2.587
204
+ 28,mlp.down_proj,0.0000000514,0.05000,2.539
205
+ 29,self_attn.v_proj,0.0000000655,0.05000,3.764
206
+ 29,self_attn.k_proj,0.0000000319,0.05000,3.790
207
+ 29,self_attn.q_proj,0.0000001379,0.05000,3.878
208
+ 29,self_attn.o_proj,0.0000000067,0.05000,1.223
209
+ 29,mlp.up_proj,0.0000002308,0.05000,2.468
210
+ 29,mlp.gate_proj,0.0000002363,0.05000,2.504
211
+ 29,mlp.down_proj,0.0000000717,0.05000,2.485
212
+ 30,self_attn.q_proj,0.0000001404,0.05000,3.584
213
+ 30,self_attn.v_proj,0.0000000711,0.05000,3.688
214
+ 30,self_attn.k_proj,0.0000000310,0.05000,3.716
215
+ 30,self_attn.o_proj,0.0000000099,0.05000,1.164
216
+ 30,mlp.gate_proj,0.0000002582,0.05000,1.471
217
+ 30,mlp.up_proj,0.0000002543,0.05000,1.483
218
+ 30,mlp.down_proj,0.0000000925,0.05000,2.235
219
+ 31,self_attn.v_proj,0.0000000747,0.05000,3.657
220
+ 31,self_attn.k_proj,0.0000000291,0.05000,3.669
221
+ 31,self_attn.q_proj,0.0000001379,0.05000,3.747
222
+ 31,self_attn.o_proj,0.0000000234,0.05000,1.285
223
+ 31,mlp.up_proj,0.0000002673,0.05000,2.496
224
+ 31,mlp.gate_proj,0.0000002787,0.05000,2.525
225
+ 31,mlp.down_proj,0.0000001321,0.05000,2.262
226
+ 32,self_attn.q_proj,0.0000001691,0.05000,3.711
227
+ 32,self_attn.k_proj,0.0000000317,0.05000,3.835
228
+ 32,self_attn.v_proj,0.0000001289,0.05000,3.910
229
+ 32,self_attn.o_proj,0.0000000241,0.05000,1.301
230
+ 32,mlp.up_proj,0.0000002878,0.05000,1.836
231
+ 32,mlp.gate_proj,0.0000003177,0.05000,1.855
232
+ 32,mlp.down_proj,0.0000002104,0.05000,2.258
233
+ 33,self_attn.k_proj,0.0000000274,0.05000,3.998
234
+ 33,self_attn.q_proj,0.0000001259,0.05000,4.011
235
+ 33,self_attn.v_proj,0.0000000742,0.05000,4.066
236
+ 33,self_attn.o_proj,0.0000000296,0.05000,1.115
237
+ 33,mlp.up_proj,0.0000002813,0.05000,2.039
238
+ 33,mlp.gate_proj,0.0000003160,0.05000,2.050
239
+ 33,mlp.down_proj,0.0000006095,0.05000,2.117
quantize_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bits": 8,
3
+ "group_size": 128,
4
+ "desc_act": false,
5
+ "sym": true,
6
+ "lm_head": false,
7
+ "quant_method": "gptq",
8
+ "checkpoint_format": "gptq",
9
+ "pack_dtype": "int32",
10
+ "meta": {
11
+ "quantizer": [
12
+ "gptqmodel:5.6.0"
13
+ ],
14
+ "uri": "https://github.com/modelcloud/gptqmodel",
15
+ "damp_percent": 0.05,
16
+ "damp_auto_increment": 0.01,
17
+ "static_groups": false,
18
+ "true_sequential": true,
19
+ "mse": 0.0,
20
+ "gptaq": false,
21
+ "gptaq_alpha": 0.25,
22
+ "act_group_aware": true
23
+ },
24
+ "pack_impl": "cpu",
25
+ "format": "gptq"
26
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,24 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<s>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "</s>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": "<pad>",
17
+ "unk_token": {
18
+ "content": "<unk>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ }
24
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b0240ce510f08e6c2041724e9043e33be9d251d1e4a4d94eb68cd47b954b61d2
3
+ size 17078292
tokenizer_config.json ADDED
The diff for this file is too large to render. See raw diff