Hugging Face's logo

Devbora29
/
Mistral_7b_INT8

Safetensors
mistral
8-bit precision
gptq
Model card Files
xet
 Community
Devbora29 commited on
Commit
c2bfc8a
·
verified ·
1 Parent(s): 25ba4fc

Upload folder using huggingface_hub

Browse files
Files changed (10) hide show
  1. chat_template.jinja +87 -0
  2. config.json +74 -0
  3. generation_config.json +7 -0
  4. model-00001-of-00002.safetensors +3 -0
  5. model-00002-of-00002.safetensors +3 -0
  6. model.safetensors.index.json +970 -0
  7. quant_log.csv +225 -0
  8. quantize_config.json +44 -0
  9. tokenizer.json +0 -0
  10. tokenizer_config.json +16 -0
chat_template.jinja ADDED
@@ -0,0 +1,87 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- if messages[0]["role"] == "system" %}
2
+ {%- set system_message = messages[0]["content"] %}
3
+ {%- set loop_messages = messages[1:] %}
4
+ {%- else %}
5
+ {%- set loop_messages = messages %}
6
+ {%- endif %}
7
+ {%- if not tools is defined %}
8
+ {%- set tools = none %}
9
+ {%- endif %}
10
+ {%- set user_messages = loop_messages | selectattr("role", "equalto", "user") | list %}
11
+
12
+ {#- This block checks for alternating user/assistant messages, skipping tool calling messages #}
13
+ {%- set ns = namespace() %}
14
+ {%- set ns.index = 0 %}
15
+ {%- for message in loop_messages %}
16
+ {%- if not (message.role == "tool" or message.role == "tool_results" or (message.tool_calls is defined and message.tool_calls is not none)) %}
17
+ {%- if (message["role"] == "user") != (ns.index % 2 == 0) %}
18
+ {{- raise_exception("After the optional system message, conversation roles must alternate user/assistant/user/assistant/...") }}
19
+ {%- endif %}
20
+ {%- set ns.index = ns.index + 1 %}
21
+ {%- endif %}
22
+ {%- endfor %}
23
+
24
+ {{- bos_token }}
25
+ {%- for message in loop_messages %}
26
+ {%- if message["role"] == "user" %}
27
+ {%- if tools is not none and (message == user_messages[-1]) %}
28
+ {{- "[AVAILABLE_TOOLS] [" }}
29
+ {%- for tool in tools %}
30
+ {%- set tool = tool.function %}
31
+ {{- '{"type": "function", "function": {' }}
32
+ {%- for key, val in tool.items() if key != "return" %}
33
+ {%- if val is string %}
34
+ {{- '"' + key + '": "' + val + '"' }}
35
+ {%- else %}
36
+ {{- '"' + key + '": ' + val|tojson }}
37
+ {%- endif %}
38
+ {%- if not loop.last %}
39
+ {{- ", " }}
40
+ {%- endif %}
41
+ {%- endfor %}
42
+ {{- "}}" }}
43
+ {%- if not loop.last %}
44
+ {{- ", " }}
45
+ {%- else %}
46
+ {{- "]" }}
47
+ {%- endif %}
48
+ {%- endfor %}
49
+ {{- "[/AVAILABLE_TOOLS]" }}
50
+ {%- endif %}
51
+ {%- if loop.last and system_message is defined %}
52
+ {{- "[INST] " + system_message + "\n\n" + message["content"] + "[/INST]" }}
53
+ {%- else %}
54
+ {{- "[INST] " + message["content"] + "[/INST]" }}
55
+ {%- endif %}
56
+ {%- elif message.tool_calls is defined and message.tool_calls is not none %}
57
+ {{- "[TOOL_CALLS] [" }}
58
+ {%- for tool_call in message.tool_calls %}
59
+ {%- set out = tool_call.function|tojson %}
60
+ {{- out[:-1] }}
61
+ {%- if not tool_call.id is defined or tool_call.id|length != 9 %}
62
+ {{- raise_exception("Tool call IDs should be alphanumeric strings with length 9!") }}
63
+ {%- endif %}
64
+ {{- ', "id": "' + tool_call.id + '"}' }}
65
+ {%- if not loop.last %}
66
+ {{- ", " }}
67
+ {%- else %}
68
+ {{- "]" + eos_token }}
69
+ {%- endif %}
70
+ {%- endfor %}
71
+ {%- elif message["role"] == "assistant" %}
72
+ {{- " " + message["content"]|trim + eos_token}}
73
+ {%- elif message["role"] == "tool_results" or message["role"] == "tool" %}
74
+ {%- if message.content is defined and message.content.content is defined %}
75
+ {%- set content = message.content.content %}
76
+ {%- else %}
77
+ {%- set content = message.content %}
78
+ {%- endif %}
79
+ {{- '[TOOL_RESULTS] {"content": ' + content|string + ", " }}
80
+ {%- if not message.tool_call_id is defined or message.tool_call_id|length != 9 %}
81
+ {{- raise_exception("Tool call IDs should be alphanumeric strings with length 9!") }}
82
+ {%- endif %}
83
+ {{- '"call_id": "' + message.tool_call_id + '"}[/TOOL_RESULTS]' }}
84
+ {%- else %}
85
+ {{- raise_exception("Only user and assistant roles are supported, with the exception of an initial optional system message!") }}
86
+ {%- endif %}
87
+ {%- endfor %}
config.json ADDED
@@ -0,0 +1,74 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "MistralForCausalLM"
4
+ ],
5
+ "attention_dropout": 0.0,
6
+ "bos_token_id": 1,
7
+ "dtype": "bfloat16",
8
+ "eos_token_id": 2,
9
+ "head_dim": 128,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 4096,
12
+ "initializer_range": 0.02,
13
+ "intermediate_size": 14336,
14
+ "max_position_embeddings": 32768,
15
+ "model_type": "mistral",
16
+ "num_attention_heads": 32,
17
+ "num_hidden_layers": 32,
18
+ "num_key_value_heads": 8,
19
+ "pad_token_id": 0,
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
+ "auto_forward_data_parallel": true,
30
+ "damp_auto_increment": 0.01,
31
+ "damp_percent": 0.05,
32
+ "fallback": {
33
+ "smooth": null,
34
+ "strategy": "rtn",
35
+ "threshold": "0.5%"
36
+ },
37
+ "foem": null,
38
+ "gc_mode": "interval",
39
+ "gptaq": null,
40
+ "hessian": {
41
+ "chunk_bytes": null,
42
+ "chunk_size": null,
43
+ "staging_dtype": "float32"
44
+ },
45
+ "mock_quantization": false,
46
+ "mse": 0.0,
47
+ "offload_to_disk": true,
48
+ "offload_to_disk_path": "./gptqmodel_offload/hedxansx-hkyyfauy/",
49
+ "pack_impl": "cpu",
50
+ "quantizer": [
51
+ "gptqmodel:6.0.3"
52
+ ],
53
+ "static_groups": false,
54
+ "true_sequential": true,
55
+ "uri": "https://github.com/modelcloud/gptqmodel",
56
+ "vram_strategy": "exclusive",
57
+ "wait_for_submodule_finalizers": false
58
+ },
59
+ "method": "gptq",
60
+ "pack_dtype": "int32",
61
+ "quant_method": "gptq",
62
+ "sym": true
63
+ },
64
+ "rms_norm_eps": 1e-05,
65
+ "rope_parameters": {
66
+ "rope_theta": 1000000.0,
67
+ "rope_type": "default"
68
+ },
69
+ "sliding_window": null,
70
+ "tie_word_embeddings": false,
71
+ "transformers_version": "5.5.4",
72
+ "use_cache": true,
73
+ "vocab_size": 32768
74
+ }
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": "5.5.4"
7
+ }
model-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:368b75fed9f8535a20c497470dd0944aa7de792500133d30145419d756114cd0
3
+ size 4275596832
model-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:158b2e2be229f6a580e9ecfdfb77d271422edb70ba17d71e9b3b84f10502a566
3
+ size 3409795544
model.safetensors.index.json ADDED
@@ -0,0 +1,970 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 7685392376
4
+ },
5
+ "weight_map": {
6
+ "lm_head.weight": "model-00001-of-00002.safetensors",
7
+ "model.embed_tokens.weight": "model-00001-of-00002.safetensors",
8
+ "model.layers.0.input_layernorm.weight": "model-00001-of-00002.safetensors",
9
+ "model.layers.0.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
10
+ "model.layers.0.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
11
+ "model.layers.0.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
12
+ "model.layers.0.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
13
+ "model.layers.0.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
14
+ "model.layers.0.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
15
+ "model.layers.0.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
16
+ "model.layers.0.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
17
+ "model.layers.0.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
18
+ "model.layers.0.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
19
+ "model.layers.0.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
20
+ "model.layers.0.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
21
+ "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
22
+ "model.layers.0.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
23
+ "model.layers.0.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
24
+ "model.layers.0.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
25
+ "model.layers.0.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
26
+ "model.layers.0.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
27
+ "model.layers.0.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
28
+ "model.layers.0.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
29
+ "model.layers.0.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
30
+ "model.layers.0.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
31
+ "model.layers.0.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
32
+ "model.layers.0.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
33
+ "model.layers.0.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
34
+ "model.layers.0.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
35
+ "model.layers.0.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
36
+ "model.layers.0.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
37
+ "model.layers.0.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
38
+ "model.layers.1.input_layernorm.weight": "model-00001-of-00002.safetensors",
39
+ "model.layers.1.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
40
+ "model.layers.1.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
41
+ "model.layers.1.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
42
+ "model.layers.1.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
43
+ "model.layers.1.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
44
+ "model.layers.1.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
45
+ "model.layers.1.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
46
+ "model.layers.1.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
47
+ "model.layers.1.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
48
+ "model.layers.1.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
49
+ "model.layers.1.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
50
+ "model.layers.1.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
51
+ "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
52
+ "model.layers.1.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
53
+ "model.layers.1.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
54
+ "model.layers.1.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
55
+ "model.layers.1.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
56
+ "model.layers.1.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
57
+ "model.layers.1.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
58
+ "model.layers.1.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
59
+ "model.layers.1.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
60
+ "model.layers.1.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
61
+ "model.layers.1.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
62
+ "model.layers.1.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
63
+ "model.layers.1.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
64
+ "model.layers.1.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
65
+ "model.layers.1.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
66
+ "model.layers.1.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
67
+ "model.layers.1.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
68
+ "model.layers.10.input_layernorm.weight": "model-00001-of-00002.safetensors",
69
+ "model.layers.10.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
70
+ "model.layers.10.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
71
+ "model.layers.10.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
72
+ "model.layers.10.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
73
+ "model.layers.10.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
74
+ "model.layers.10.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
75
+ "model.layers.10.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
76
+ "model.layers.10.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
77
+ "model.layers.10.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
78
+ "model.layers.10.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
79
+ "model.layers.10.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
80
+ "model.layers.10.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
81
+ "model.layers.10.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
82
+ "model.layers.10.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
83
+ "model.layers.10.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
84
+ "model.layers.10.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
85
+ "model.layers.10.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
86
+ "model.layers.10.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
87
+ "model.layers.10.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
88
+ "model.layers.10.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
89
+ "model.layers.10.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
90
+ "model.layers.10.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
91
+ "model.layers.10.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
92
+ "model.layers.10.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
93
+ "model.layers.10.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
94
+ "model.layers.10.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
95
+ "model.layers.10.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
96
+ "model.layers.10.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
97
+ "model.layers.10.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
98
+ "model.layers.11.input_layernorm.weight": "model-00001-of-00002.safetensors",
99
+ "model.layers.11.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
100
+ "model.layers.11.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
101
+ "model.layers.11.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
102
+ "model.layers.11.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
103
+ "model.layers.11.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
104
+ "model.layers.11.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
105
+ "model.layers.11.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
106
+ "model.layers.11.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
107
+ "model.layers.11.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
108
+ "model.layers.11.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
109
+ "model.layers.11.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
110
+ "model.layers.11.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
111
+ "model.layers.11.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
112
+ "model.layers.11.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
113
+ "model.layers.11.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
114
+ "model.layers.11.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
115
+ "model.layers.11.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
116
+ "model.layers.11.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
117
+ "model.layers.11.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
118
+ "model.layers.11.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
119
+ "model.layers.11.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
120
+ "model.layers.11.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
121
+ "model.layers.11.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
122
+ "model.layers.11.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
123
+ "model.layers.11.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
124
+ "model.layers.11.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
125
+ "model.layers.11.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
126
+ "model.layers.11.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
127
+ "model.layers.11.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
128
+ "model.layers.12.input_layernorm.weight": "model-00001-of-00002.safetensors",
129
+ "model.layers.12.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
130
+ "model.layers.12.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
131
+ "model.layers.12.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
132
+ "model.layers.12.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
133
+ "model.layers.12.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
134
+ "model.layers.12.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
135
+ "model.layers.12.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
136
+ "model.layers.12.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
137
+ "model.layers.12.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
138
+ "model.layers.12.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
139
+ "model.layers.12.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
140
+ "model.layers.12.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
141
+ "model.layers.12.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
142
+ "model.layers.12.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
143
+ "model.layers.12.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
144
+ "model.layers.12.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
145
+ "model.layers.12.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
146
+ "model.layers.12.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
147
+ "model.layers.12.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
148
+ "model.layers.12.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
149
+ "model.layers.12.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
150
+ "model.layers.12.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
151
+ "model.layers.12.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
152
+ "model.layers.12.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
153
+ "model.layers.12.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
154
+ "model.layers.12.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
155
+ "model.layers.12.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
156
+ "model.layers.12.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
157
+ "model.layers.12.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
158
+ "model.layers.13.input_layernorm.weight": "model-00001-of-00002.safetensors",
159
+ "model.layers.13.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
160
+ "model.layers.13.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
161
+ "model.layers.13.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
162
+ "model.layers.13.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
163
+ "model.layers.13.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
164
+ "model.layers.13.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
165
+ "model.layers.13.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
166
+ "model.layers.13.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
167
+ "model.layers.13.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
168
+ "model.layers.13.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
169
+ "model.layers.13.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
170
+ "model.layers.13.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
171
+ "model.layers.13.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
172
+ "model.layers.13.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
173
+ "model.layers.13.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
174
+ "model.layers.13.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
175
+ "model.layers.13.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
176
+ "model.layers.13.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
177
+ "model.layers.13.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
178
+ "model.layers.13.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
179
+ "model.layers.13.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
180
+ "model.layers.13.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
181
+ "model.layers.13.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
182
+ "model.layers.13.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
183
+ "model.layers.13.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
184
+ "model.layers.13.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
185
+ "model.layers.13.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
186
+ "model.layers.13.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
187
+ "model.layers.13.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
188
+ "model.layers.14.input_layernorm.weight": "model-00001-of-00002.safetensors",
189
+ "model.layers.14.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
190
+ "model.layers.14.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
191
+ "model.layers.14.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
192
+ "model.layers.14.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
193
+ "model.layers.14.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
194
+ "model.layers.14.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
195
+ "model.layers.14.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
196
+ "model.layers.14.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
197
+ "model.layers.14.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
198
+ "model.layers.14.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
199
+ "model.layers.14.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
200
+ "model.layers.14.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
201
+ "model.layers.14.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
202
+ "model.layers.14.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
203
+ "model.layers.14.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
204
+ "model.layers.14.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
205
+ "model.layers.14.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
206
+ "model.layers.14.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
207
+ "model.layers.14.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
208
+ "model.layers.14.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
209
+ "model.layers.14.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
210
+ "model.layers.14.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
211
+ "model.layers.14.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
212
+ "model.layers.14.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
213
+ "model.layers.14.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
214
+ "model.layers.14.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
215
+ "model.layers.14.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
216
+ "model.layers.14.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
217
+ "model.layers.14.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
218
+ "model.layers.15.input_layernorm.weight": "model-00001-of-00002.safetensors",
219
+ "model.layers.15.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
220
+ "model.layers.15.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
221
+ "model.layers.15.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
222
+ "model.layers.15.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
223
+ "model.layers.15.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
224
+ "model.layers.15.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
225
+ "model.layers.15.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
226
+ "model.layers.15.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
227
+ "model.layers.15.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
228
+ "model.layers.15.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
229
+ "model.layers.15.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
230
+ "model.layers.15.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
231
+ "model.layers.15.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
232
+ "model.layers.15.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
233
+ "model.layers.15.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
234
+ "model.layers.15.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
235
+ "model.layers.15.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
236
+ "model.layers.15.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
237
+ "model.layers.15.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
238
+ "model.layers.15.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
239
+ "model.layers.15.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
240
+ "model.layers.15.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
241
+ "model.layers.15.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
242
+ "model.layers.15.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
243
+ "model.layers.15.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
244
+ "model.layers.15.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
245
+ "model.layers.15.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
246
+ "model.layers.15.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
247
+ "model.layers.15.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
248
+ "model.layers.16.input_layernorm.weight": "model-00001-of-00002.safetensors",
249
+ "model.layers.16.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
250
+ "model.layers.16.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
251
+ "model.layers.16.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
252
+ "model.layers.16.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
253
+ "model.layers.16.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
254
+ "model.layers.16.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
255
+ "model.layers.16.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
256
+ "model.layers.16.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
257
+ "model.layers.16.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
258
+ "model.layers.16.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
259
+ "model.layers.16.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
260
+ "model.layers.16.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
261
+ "model.layers.16.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
262
+ "model.layers.16.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
263
+ "model.layers.16.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
264
+ "model.layers.16.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
265
+ "model.layers.16.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
266
+ "model.layers.16.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
267
+ "model.layers.16.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
268
+ "model.layers.16.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
269
+ "model.layers.16.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
270
+ "model.layers.16.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
271
+ "model.layers.16.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
272
+ "model.layers.16.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
273
+ "model.layers.16.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
274
+ "model.layers.16.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
275
+ "model.layers.16.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
276
+ "model.layers.16.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
277
+ "model.layers.16.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
278
+ "model.layers.17.input_layernorm.weight": "model-00001-of-00002.safetensors",
279
+ "model.layers.17.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
280
+ "model.layers.17.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
281
+ "model.layers.17.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
282
+ "model.layers.17.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
283
+ "model.layers.17.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
284
+ "model.layers.17.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
285
+ "model.layers.17.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
286
+ "model.layers.17.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
287
+ "model.layers.17.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
288
+ "model.layers.17.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
289
+ "model.layers.17.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
290
+ "model.layers.17.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
291
+ "model.layers.17.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
292
+ "model.layers.17.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
293
+ "model.layers.17.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
294
+ "model.layers.17.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
295
+ "model.layers.17.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
296
+ "model.layers.17.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
297
+ "model.layers.17.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
298
+ "model.layers.17.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
299
+ "model.layers.17.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
300
+ "model.layers.17.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
301
+ "model.layers.17.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
302
+ "model.layers.17.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
303
+ "model.layers.17.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
304
+ "model.layers.17.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
305
+ "model.layers.17.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
306
+ "model.layers.17.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
307
+ "model.layers.17.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
308
+ "model.layers.18.input_layernorm.weight": "model-00001-of-00002.safetensors",
309
+ "model.layers.18.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
310
+ "model.layers.18.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
311
+ "model.layers.18.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
312
+ "model.layers.18.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
313
+ "model.layers.18.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
314
+ "model.layers.18.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
315
+ "model.layers.18.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
316
+ "model.layers.18.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
317
+ "model.layers.18.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
318
+ "model.layers.18.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
319
+ "model.layers.18.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
320
+ "model.layers.18.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
321
+ "model.layers.18.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
322
+ "model.layers.18.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
323
+ "model.layers.18.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
324
+ "model.layers.18.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
325
+ "model.layers.18.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
326
+ "model.layers.18.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
327
+ "model.layers.18.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
328
+ "model.layers.18.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
329
+ "model.layers.18.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
330
+ "model.layers.18.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
331
+ "model.layers.18.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
332
+ "model.layers.18.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
333
+ "model.layers.18.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
334
+ "model.layers.18.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
335
+ "model.layers.18.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
336
+ "model.layers.18.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
337
+ "model.layers.18.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
338
+ "model.layers.19.input_layernorm.weight": "model-00001-of-00002.safetensors",
339
+ "model.layers.19.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
340
+ "model.layers.19.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
341
+ "model.layers.19.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
342
+ "model.layers.19.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
343
+ "model.layers.19.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
344
+ "model.layers.19.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
345
+ "model.layers.19.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
346
+ "model.layers.19.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
347
+ "model.layers.19.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
348
+ "model.layers.19.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
349
+ "model.layers.19.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
350
+ "model.layers.19.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
351
+ "model.layers.19.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
352
+ "model.layers.19.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
353
+ "model.layers.19.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
354
+ "model.layers.19.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
355
+ "model.layers.19.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
356
+ "model.layers.19.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
357
+ "model.layers.19.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
358
+ "model.layers.19.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
359
+ "model.layers.19.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
360
+ "model.layers.19.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
361
+ "model.layers.19.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
362
+ "model.layers.19.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
363
+ "model.layers.19.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
364
+ "model.layers.19.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
365
+ "model.layers.19.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
366
+ "model.layers.19.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
367
+ "model.layers.19.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
368
+ "model.layers.2.input_layernorm.weight": "model-00001-of-00002.safetensors",
369
+ "model.layers.2.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
370
+ "model.layers.2.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
371
+ "model.layers.2.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
372
+ "model.layers.2.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
373
+ "model.layers.2.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
374
+ "model.layers.2.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
375
+ "model.layers.2.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
376
+ "model.layers.2.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
377
+ "model.layers.2.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
378
+ "model.layers.2.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
379
+ "model.layers.2.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
380
+ "model.layers.2.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
381
+ "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
382
+ "model.layers.2.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
383
+ "model.layers.2.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
384
+ "model.layers.2.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
385
+ "model.layers.2.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
386
+ "model.layers.2.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
387
+ "model.layers.2.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
388
+ "model.layers.2.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
389
+ "model.layers.2.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
390
+ "model.layers.2.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
391
+ "model.layers.2.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
392
+ "model.layers.2.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
393
+ "model.layers.2.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
394
+ "model.layers.2.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
395
+ "model.layers.2.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
396
+ "model.layers.2.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
397
+ "model.layers.2.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
398
+ "model.layers.20.input_layernorm.weight": "model-00001-of-00002.safetensors",
399
+ "model.layers.20.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
400
+ "model.layers.20.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
401
+ "model.layers.20.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
402
+ "model.layers.20.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
403
+ "model.layers.20.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
404
+ "model.layers.20.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
405
+ "model.layers.20.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
406
+ "model.layers.20.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
407
+ "model.layers.20.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
408
+ "model.layers.20.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
409
+ "model.layers.20.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
410
+ "model.layers.20.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
411
+ "model.layers.20.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
412
+ "model.layers.20.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
413
+ "model.layers.20.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
414
+ "model.layers.20.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
415
+ "model.layers.20.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
416
+ "model.layers.20.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
417
+ "model.layers.20.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
418
+ "model.layers.20.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
419
+ "model.layers.20.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
420
+ "model.layers.20.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
421
+ "model.layers.20.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
422
+ "model.layers.20.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
423
+ "model.layers.20.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
424
+ "model.layers.20.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
425
+ "model.layers.20.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
426
+ "model.layers.20.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
427
+ "model.layers.20.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
428
+ "model.layers.21.input_layernorm.weight": "model-00001-of-00002.safetensors",
429
+ "model.layers.21.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
430
+ "model.layers.21.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
431
+ "model.layers.21.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
432
+ "model.layers.21.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
433
+ "model.layers.21.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
434
+ "model.layers.21.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
435
+ "model.layers.21.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
436
+ "model.layers.21.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
437
+ "model.layers.21.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
438
+ "model.layers.21.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
439
+ "model.layers.21.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
440
+ "model.layers.21.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
441
+ "model.layers.21.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
442
+ "model.layers.21.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
443
+ "model.layers.21.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
444
+ "model.layers.21.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
445
+ "model.layers.21.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
446
+ "model.layers.21.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
447
+ "model.layers.21.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
448
+ "model.layers.21.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
449
+ "model.layers.21.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
450
+ "model.layers.21.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
451
+ "model.layers.21.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
452
+ "model.layers.21.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
453
+ "model.layers.21.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
454
+ "model.layers.21.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
455
+ "model.layers.21.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
456
+ "model.layers.21.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
457
+ "model.layers.21.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
458
+ "model.layers.22.input_layernorm.weight": "model-00001-of-00002.safetensors",
459
+ "model.layers.22.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
460
+ "model.layers.22.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
461
+ "model.layers.22.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
462
+ "model.layers.22.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
463
+ "model.layers.22.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
464
+ "model.layers.22.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
465
+ "model.layers.22.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
466
+ "model.layers.22.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
467
+ "model.layers.22.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
468
+ "model.layers.22.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
469
+ "model.layers.22.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
470
+ "model.layers.22.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
471
+ "model.layers.22.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
472
+ "model.layers.22.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
473
+ "model.layers.22.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
474
+ "model.layers.22.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
475
+ "model.layers.22.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
476
+ "model.layers.22.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
477
+ "model.layers.22.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
478
+ "model.layers.22.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
479
+ "model.layers.22.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
480
+ "model.layers.22.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
481
+ "model.layers.22.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
482
+ "model.layers.22.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
483
+ "model.layers.22.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
484
+ "model.layers.22.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
485
+ "model.layers.22.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
486
+ "model.layers.22.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
487
+ "model.layers.22.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
488
+ "model.layers.23.input_layernorm.weight": "model-00001-of-00002.safetensors",
489
+ "model.layers.23.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
490
+ "model.layers.23.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
491
+ "model.layers.23.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
492
+ "model.layers.23.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
493
+ "model.layers.23.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
494
+ "model.layers.23.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
495
+ "model.layers.23.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
496
+ "model.layers.23.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
497
+ "model.layers.23.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
498
+ "model.layers.23.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
499
+ "model.layers.23.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
500
+ "model.layers.23.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
501
+ "model.layers.23.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
502
+ "model.layers.23.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
503
+ "model.layers.23.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
504
+ "model.layers.23.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
505
+ "model.layers.23.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
506
+ "model.layers.23.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
507
+ "model.layers.23.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
508
+ "model.layers.23.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
509
+ "model.layers.23.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
510
+ "model.layers.23.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
511
+ "model.layers.23.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
512
+ "model.layers.23.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
513
+ "model.layers.23.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
514
+ "model.layers.23.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
515
+ "model.layers.23.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
516
+ "model.layers.23.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
517
+ "model.layers.23.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
518
+ "model.layers.24.input_layernorm.weight": "model-00001-of-00002.safetensors",
519
+ "model.layers.24.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
520
+ "model.layers.24.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
521
+ "model.layers.24.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
522
+ "model.layers.24.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
523
+ "model.layers.24.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
524
+ "model.layers.24.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
525
+ "model.layers.24.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
526
+ "model.layers.24.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
527
+ "model.layers.24.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
528
+ "model.layers.24.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
529
+ "model.layers.24.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
530
+ "model.layers.24.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
531
+ "model.layers.24.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
532
+ "model.layers.24.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
533
+ "model.layers.24.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
534
+ "model.layers.24.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
535
+ "model.layers.24.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
536
+ "model.layers.24.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
537
+ "model.layers.24.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
538
+ "model.layers.24.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
539
+ "model.layers.24.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
540
+ "model.layers.24.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
541
+ "model.layers.24.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
542
+ "model.layers.24.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
543
+ "model.layers.24.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
544
+ "model.layers.24.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
545
+ "model.layers.24.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
546
+ "model.layers.24.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
547
+ "model.layers.24.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
548
+ "model.layers.25.input_layernorm.weight": "model-00001-of-00002.safetensors",
549
+ "model.layers.25.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
550
+ "model.layers.25.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
551
+ "model.layers.25.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
552
+ "model.layers.25.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
553
+ "model.layers.25.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
554
+ "model.layers.25.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
555
+ "model.layers.25.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
556
+ "model.layers.25.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
557
+ "model.layers.25.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
558
+ "model.layers.25.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
559
+ "model.layers.25.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
560
+ "model.layers.25.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
561
+ "model.layers.25.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
562
+ "model.layers.25.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
563
+ "model.layers.25.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
564
+ "model.layers.25.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
565
+ "model.layers.25.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
566
+ "model.layers.25.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
567
+ "model.layers.25.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
568
+ "model.layers.25.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
569
+ "model.layers.25.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
570
+ "model.layers.25.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
571
+ "model.layers.25.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
572
+ "model.layers.25.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
573
+ "model.layers.25.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
574
+ "model.layers.25.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
575
+ "model.layers.25.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
576
+ "model.layers.25.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
577
+ "model.layers.25.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
578
+ "model.layers.26.input_layernorm.weight": "model-00001-of-00002.safetensors",
579
+ "model.layers.26.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
580
+ "model.layers.26.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
581
+ "model.layers.26.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
582
+ "model.layers.26.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
583
+ "model.layers.26.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
584
+ "model.layers.26.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
585
+ "model.layers.26.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
586
+ "model.layers.26.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
587
+ "model.layers.26.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
588
+ "model.layers.26.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
589
+ "model.layers.26.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
590
+ "model.layers.26.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
591
+ "model.layers.26.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
592
+ "model.layers.26.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
593
+ "model.layers.26.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
594
+ "model.layers.26.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
595
+ "model.layers.26.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
596
+ "model.layers.26.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
597
+ "model.layers.26.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
598
+ "model.layers.26.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
599
+ "model.layers.26.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
600
+ "model.layers.26.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
601
+ "model.layers.26.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
602
+ "model.layers.26.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
603
+ "model.layers.26.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
604
+ "model.layers.26.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
605
+ "model.layers.26.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
606
+ "model.layers.26.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
607
+ "model.layers.26.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
608
+ "model.layers.27.input_layernorm.weight": "model-00001-of-00002.safetensors",
609
+ "model.layers.27.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
610
+ "model.layers.27.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
611
+ "model.layers.27.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
612
+ "model.layers.27.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
613
+ "model.layers.27.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
614
+ "model.layers.27.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
615
+ "model.layers.27.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
616
+ "model.layers.27.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
617
+ "model.layers.27.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
618
+ "model.layers.27.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
619
+ "model.layers.27.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
620
+ "model.layers.27.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
621
+ "model.layers.27.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
622
+ "model.layers.27.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
623
+ "model.layers.27.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
624
+ "model.layers.27.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
625
+ "model.layers.27.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
626
+ "model.layers.27.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
627
+ "model.layers.27.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
628
+ "model.layers.27.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
629
+ "model.layers.27.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
630
+ "model.layers.27.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
631
+ "model.layers.27.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
632
+ "model.layers.27.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
633
+ "model.layers.27.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
634
+ "model.layers.27.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
635
+ "model.layers.27.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
636
+ "model.layers.27.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
637
+ "model.layers.27.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
638
+ "model.layers.28.input_layernorm.weight": "model-00001-of-00002.safetensors",
639
+ "model.layers.28.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
640
+ "model.layers.28.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
641
+ "model.layers.28.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
642
+ "model.layers.28.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
643
+ "model.layers.28.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
644
+ "model.layers.28.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
645
+ "model.layers.28.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
646
+ "model.layers.28.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
647
+ "model.layers.28.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
648
+ "model.layers.28.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
649
+ "model.layers.28.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
650
+ "model.layers.28.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
651
+ "model.layers.28.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
652
+ "model.layers.28.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
653
+ "model.layers.28.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
654
+ "model.layers.28.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
655
+ "model.layers.28.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
656
+ "model.layers.28.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
657
+ "model.layers.28.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
658
+ "model.layers.28.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
659
+ "model.layers.28.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
660
+ "model.layers.28.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
661
+ "model.layers.28.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
662
+ "model.layers.28.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
663
+ "model.layers.28.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
664
+ "model.layers.28.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
665
+ "model.layers.28.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
666
+ "model.layers.28.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
667
+ "model.layers.28.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
668
+ "model.layers.29.input_layernorm.weight": "model-00001-of-00002.safetensors",
669
+ "model.layers.29.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
670
+ "model.layers.29.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
671
+ "model.layers.29.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
672
+ "model.layers.29.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
673
+ "model.layers.29.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
674
+ "model.layers.29.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
675
+ "model.layers.29.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
676
+ "model.layers.29.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
677
+ "model.layers.29.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
678
+ "model.layers.29.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
679
+ "model.layers.29.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
680
+ "model.layers.29.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
681
+ "model.layers.29.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
682
+ "model.layers.29.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
683
+ "model.layers.29.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
684
+ "model.layers.29.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
685
+ "model.layers.29.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
686
+ "model.layers.29.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
687
+ "model.layers.29.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
688
+ "model.layers.29.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
689
+ "model.layers.29.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
690
+ "model.layers.29.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
691
+ "model.layers.29.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
692
+ "model.layers.29.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
693
+ "model.layers.29.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
694
+ "model.layers.29.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
695
+ "model.layers.29.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
696
+ "model.layers.29.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
697
+ "model.layers.29.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
698
+ "model.layers.3.input_layernorm.weight": "model-00001-of-00002.safetensors",
699
+ "model.layers.3.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
700
+ "model.layers.3.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
701
+ "model.layers.3.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
702
+ "model.layers.3.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
703
+ "model.layers.3.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
704
+ "model.layers.3.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
705
+ "model.layers.3.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
706
+ "model.layers.3.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
707
+ "model.layers.3.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
708
+ "model.layers.3.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
709
+ "model.layers.3.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
710
+ "model.layers.3.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
711
+ "model.layers.3.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
712
+ "model.layers.3.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
713
+ "model.layers.3.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
714
+ "model.layers.3.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
715
+ "model.layers.3.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
716
+ "model.layers.3.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
717
+ "model.layers.3.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
718
+ "model.layers.3.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
719
+ "model.layers.3.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
720
+ "model.layers.3.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
721
+ "model.layers.3.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
722
+ "model.layers.3.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
723
+ "model.layers.3.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
724
+ "model.layers.3.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
725
+ "model.layers.3.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
726
+ "model.layers.3.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
727
+ "model.layers.3.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
728
+ "model.layers.30.input_layernorm.weight": "model-00001-of-00002.safetensors",
729
+ "model.layers.30.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
730
+ "model.layers.30.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
731
+ "model.layers.30.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
732
+ "model.layers.30.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
733
+ "model.layers.30.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
734
+ "model.layers.30.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
735
+ "model.layers.30.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
736
+ "model.layers.30.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
737
+ "model.layers.30.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
738
+ "model.layers.30.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
739
+ "model.layers.30.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
740
+ "model.layers.30.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
741
+ "model.layers.30.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
742
+ "model.layers.30.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
743
+ "model.layers.30.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
744
+ "model.layers.30.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
745
+ "model.layers.30.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
746
+ "model.layers.30.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
747
+ "model.layers.30.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
748
+ "model.layers.30.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
749
+ "model.layers.30.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
750
+ "model.layers.30.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
751
+ "model.layers.30.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
752
+ "model.layers.30.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
753
+ "model.layers.30.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
754
+ "model.layers.30.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
755
+ "model.layers.30.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
756
+ "model.layers.30.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
757
+ "model.layers.30.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
758
+ "model.layers.31.input_layernorm.weight": "model-00001-of-00002.safetensors",
759
+ "model.layers.31.mlp.down_proj.g_idx": "model-00002-of-00002.safetensors",
760
+ "model.layers.31.mlp.down_proj.qweight": "model-00002-of-00002.safetensors",
761
+ "model.layers.31.mlp.down_proj.qzeros": "model-00002-of-00002.safetensors",
762
+ "model.layers.31.mlp.down_proj.scales": "model-00002-of-00002.safetensors",
763
+ "model.layers.31.mlp.gate_proj.g_idx": "model-00002-of-00002.safetensors",
764
+ "model.layers.31.mlp.gate_proj.qweight": "model-00002-of-00002.safetensors",
765
+ "model.layers.31.mlp.gate_proj.qzeros": "model-00002-of-00002.safetensors",
766
+ "model.layers.31.mlp.gate_proj.scales": "model-00002-of-00002.safetensors",
767
+ "model.layers.31.mlp.up_proj.g_idx": "model-00002-of-00002.safetensors",
768
+ "model.layers.31.mlp.up_proj.qweight": "model-00002-of-00002.safetensors",
769
+ "model.layers.31.mlp.up_proj.qzeros": "model-00002-of-00002.safetensors",
770
+ "model.layers.31.mlp.up_proj.scales": "model-00002-of-00002.safetensors",
771
+ "model.layers.31.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
772
+ "model.layers.31.self_attn.k_proj.g_idx": "model-00002-of-00002.safetensors",
773
+ "model.layers.31.self_attn.k_proj.qweight": "model-00002-of-00002.safetensors",
774
+ "model.layers.31.self_attn.k_proj.qzeros": "model-00002-of-00002.safetensors",
775
+ "model.layers.31.self_attn.k_proj.scales": "model-00002-of-00002.safetensors",
776
+ "model.layers.31.self_attn.o_proj.g_idx": "model-00002-of-00002.safetensors",
777
+ "model.layers.31.self_attn.o_proj.qweight": "model-00002-of-00002.safetensors",
778
+ "model.layers.31.self_attn.o_proj.qzeros": "model-00002-of-00002.safetensors",
779
+ "model.layers.31.self_attn.o_proj.scales": "model-00002-of-00002.safetensors",
780
+ "model.layers.31.self_attn.q_proj.g_idx": "model-00002-of-00002.safetensors",
781
+ "model.layers.31.self_attn.q_proj.qweight": "model-00002-of-00002.safetensors",
782
+ "model.layers.31.self_attn.q_proj.qzeros": "model-00002-of-00002.safetensors",
783
+ "model.layers.31.self_attn.q_proj.scales": "model-00002-of-00002.safetensors",
784
+ "model.layers.31.self_attn.v_proj.g_idx": "model-00002-of-00002.safetensors",
785
+ "model.layers.31.self_attn.v_proj.qweight": "model-00002-of-00002.safetensors",
786
+ "model.layers.31.self_attn.v_proj.qzeros": "model-00002-of-00002.safetensors",
787
+ "model.layers.31.self_attn.v_proj.scales": "model-00002-of-00002.safetensors",
788
+ "model.layers.4.input_layernorm.weight": "model-00001-of-00002.safetensors",
789
+ "model.layers.4.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
790
+ "model.layers.4.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
791
+ "model.layers.4.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
792
+ "model.layers.4.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
793
+ "model.layers.4.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
794
+ "model.layers.4.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
795
+ "model.layers.4.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
796
+ "model.layers.4.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
797
+ "model.layers.4.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
798
+ "model.layers.4.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
799
+ "model.layers.4.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
800
+ "model.layers.4.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
801
+ "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
802
+ "model.layers.4.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
803
+ "model.layers.4.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
804
+ "model.layers.4.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
805
+ "model.layers.4.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
806
+ "model.layers.4.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
807
+ "model.layers.4.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
808
+ "model.layers.4.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
809
+ "model.layers.4.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
810
+ "model.layers.4.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
811
+ "model.layers.4.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
812
+ "model.layers.4.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
813
+ "model.layers.4.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
814
+ "model.layers.4.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
815
+ "model.layers.4.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
816
+ "model.layers.4.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
817
+ "model.layers.4.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
818
+ "model.layers.5.input_layernorm.weight": "model-00001-of-00002.safetensors",
819
+ "model.layers.5.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
820
+ "model.layers.5.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
821
+ "model.layers.5.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
822
+ "model.layers.5.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
823
+ "model.layers.5.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
824
+ "model.layers.5.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
825
+ "model.layers.5.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
826
+ "model.layers.5.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
827
+ "model.layers.5.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
828
+ "model.layers.5.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
829
+ "model.layers.5.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
830
+ "model.layers.5.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
831
+ "model.layers.5.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
832
+ "model.layers.5.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
833
+ "model.layers.5.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
834
+ "model.layers.5.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
835
+ "model.layers.5.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
836
+ "model.layers.5.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
837
+ "model.layers.5.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
838
+ "model.layers.5.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
839
+ "model.layers.5.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
840
+ "model.layers.5.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
841
+ "model.layers.5.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
842
+ "model.layers.5.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
843
+ "model.layers.5.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
844
+ "model.layers.5.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
845
+ "model.layers.5.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
846
+ "model.layers.5.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
847
+ "model.layers.5.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
848
+ "model.layers.6.input_layernorm.weight": "model-00001-of-00002.safetensors",
849
+ "model.layers.6.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
850
+ "model.layers.6.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
851
+ "model.layers.6.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
852
+ "model.layers.6.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
853
+ "model.layers.6.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
854
+ "model.layers.6.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
855
+ "model.layers.6.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
856
+ "model.layers.6.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
857
+ "model.layers.6.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
858
+ "model.layers.6.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
859
+ "model.layers.6.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
860
+ "model.layers.6.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
861
+ "model.layers.6.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
862
+ "model.layers.6.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
863
+ "model.layers.6.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
864
+ "model.layers.6.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
865
+ "model.layers.6.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
866
+ "model.layers.6.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
867
+ "model.layers.6.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
868
+ "model.layers.6.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
869
+ "model.layers.6.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
870
+ "model.layers.6.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
871
+ "model.layers.6.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
872
+ "model.layers.6.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
873
+ "model.layers.6.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
874
+ "model.layers.6.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
875
+ "model.layers.6.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
876
+ "model.layers.6.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
877
+ "model.layers.6.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
878
+ "model.layers.7.input_layernorm.weight": "model-00001-of-00002.safetensors",
879
+ "model.layers.7.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
880
+ "model.layers.7.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
881
+ "model.layers.7.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
882
+ "model.layers.7.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
883
+ "model.layers.7.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
884
+ "model.layers.7.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
885
+ "model.layers.7.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
886
+ "model.layers.7.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
887
+ "model.layers.7.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
888
+ "model.layers.7.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
889
+ "model.layers.7.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
890
+ "model.layers.7.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
891
+ "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
892
+ "model.layers.7.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
893
+ "model.layers.7.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
894
+ "model.layers.7.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
895
+ "model.layers.7.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
896
+ "model.layers.7.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
897
+ "model.layers.7.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
898
+ "model.layers.7.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
899
+ "model.layers.7.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
900
+ "model.layers.7.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
901
+ "model.layers.7.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
902
+ "model.layers.7.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
903
+ "model.layers.7.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
904
+ "model.layers.7.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
905
+ "model.layers.7.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
906
+ "model.layers.7.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
907
+ "model.layers.7.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
908
+ "model.layers.8.input_layernorm.weight": "model-00001-of-00002.safetensors",
909
+ "model.layers.8.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
910
+ "model.layers.8.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
911
+ "model.layers.8.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
912
+ "model.layers.8.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
913
+ "model.layers.8.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
914
+ "model.layers.8.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
915
+ "model.layers.8.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
916
+ "model.layers.8.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
917
+ "model.layers.8.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
918
+ "model.layers.8.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
919
+ "model.layers.8.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
920
+ "model.layers.8.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
921
+ "model.layers.8.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
922
+ "model.layers.8.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
923
+ "model.layers.8.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
924
+ "model.layers.8.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
925
+ "model.layers.8.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
926
+ "model.layers.8.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
927
+ "model.layers.8.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
928
+ "model.layers.8.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
929
+ "model.layers.8.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
930
+ "model.layers.8.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
931
+ "model.layers.8.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
932
+ "model.layers.8.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
933
+ "model.layers.8.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
934
+ "model.layers.8.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
935
+ "model.layers.8.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
936
+ "model.layers.8.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
937
+ "model.layers.8.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
938
+ "model.layers.9.input_layernorm.weight": "model-00001-of-00002.safetensors",
939
+ "model.layers.9.mlp.down_proj.g_idx": "model-00001-of-00002.safetensors",
940
+ "model.layers.9.mlp.down_proj.qweight": "model-00001-of-00002.safetensors",
941
+ "model.layers.9.mlp.down_proj.qzeros": "model-00001-of-00002.safetensors",
942
+ "model.layers.9.mlp.down_proj.scales": "model-00001-of-00002.safetensors",
943
+ "model.layers.9.mlp.gate_proj.g_idx": "model-00001-of-00002.safetensors",
944
+ "model.layers.9.mlp.gate_proj.qweight": "model-00001-of-00002.safetensors",
945
+ "model.layers.9.mlp.gate_proj.qzeros": "model-00001-of-00002.safetensors",
946
+ "model.layers.9.mlp.gate_proj.scales": "model-00001-of-00002.safetensors",
947
+ "model.layers.9.mlp.up_proj.g_idx": "model-00001-of-00002.safetensors",
948
+ "model.layers.9.mlp.up_proj.qweight": "model-00001-of-00002.safetensors",
949
+ "model.layers.9.mlp.up_proj.qzeros": "model-00001-of-00002.safetensors",
950
+ "model.layers.9.mlp.up_proj.scales": "model-00001-of-00002.safetensors",
951
+ "model.layers.9.post_attention_layernorm.weight": "model-00001-of-00002.safetensors",
952
+ "model.layers.9.self_attn.k_proj.g_idx": "model-00001-of-00002.safetensors",
953
+ "model.layers.9.self_attn.k_proj.qweight": "model-00001-of-00002.safetensors",
954
+ "model.layers.9.self_attn.k_proj.qzeros": "model-00001-of-00002.safetensors",
955
+ "model.layers.9.self_attn.k_proj.scales": "model-00001-of-00002.safetensors",
956
+ "model.layers.9.self_attn.o_proj.g_idx": "model-00001-of-00002.safetensors",
957
+ "model.layers.9.self_attn.o_proj.qweight": "model-00001-of-00002.safetensors",
958
+ "model.layers.9.self_attn.o_proj.qzeros": "model-00001-of-00002.safetensors",
959
+ "model.layers.9.self_attn.o_proj.scales": "model-00001-of-00002.safetensors",
960
+ "model.layers.9.self_attn.q_proj.g_idx": "model-00001-of-00002.safetensors",
961
+ "model.layers.9.self_attn.q_proj.qweight": "model-00001-of-00002.safetensors",
962
+ "model.layers.9.self_attn.q_proj.qzeros": "model-00001-of-00002.safetensors",
963
+ "model.layers.9.self_attn.q_proj.scales": "model-00001-of-00002.safetensors",
964
+ "model.layers.9.self_attn.v_proj.g_idx": "model-00001-of-00002.safetensors",
965
+ "model.layers.9.self_attn.v_proj.qweight": "model-00001-of-00002.safetensors",
966
+ "model.layers.9.self_attn.v_proj.qzeros": "model-00001-of-00002.safetensors",
967
+ "model.layers.9.self_attn.v_proj.scales": "model-00001-of-00002.safetensors",
968
+ "model.norm.weight": "model-00001-of-00002.safetensors"
969
+ }
970
+ }
quant_log.csv ADDED
@@ -0,0 +1,225 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ layer,module,loss,samples,damp,time
2
+ 0,self_attn.k_proj,0.0000000099,0.05000,4.817
3
+ 0,self_attn.q_proj,0.0000000326,0.05000,4.873
4
+ 0,self_attn.v_proj,0.0000000008,0.05000,4.883
5
+ 0,self_attn.o_proj,0.0000000000,0.05000,1.506
6
+ 0,mlp.gate_proj,0.0000000322,0.05000,2.852
7
+ 0,mlp.up_proj,0.0000000279,0.05000,2.861
8
+ 0,mlp.down_proj,0.0000000000,0.05000,5.854
9
+ 1,self_attn.v_proj,0.0000000075,0.05000,5.948
10
+ 1,self_attn.q_proj,0.0000000986,0.05000,5.967
11
+ 1,self_attn.k_proj,0.0000000453,0.05000,5.986
12
+ 1,self_attn.o_proj,0.0000000000,0.05000,1.513
13
+ 1,mlp.gate_proj,0.0000001107,0.05000,2.723
14
+ 1,mlp.up_proj,0.0000000970,0.05000,2.728
15
+ 1,mlp.down_proj,0.0000000181,0.05000,5.748
16
+ 2,self_attn.v_proj,0.0000000473,0.05000,5.806
17
+ 2,self_attn.k_proj,0.0000003621,0.05000,5.825
18
+ 2,self_attn.q_proj,0.0000007457,0.05000,5.841
19
+ 2,self_attn.o_proj,0.0000000000,0.05000,1.545
20
+ 2,mlp.gate_proj,0.0000002318,0.05000,2.761
21
+ 2,mlp.up_proj,0.0000002023,0.05000,2.779
22
+ 2,mlp.down_proj,0.0000000001,0.05000,5.801
23
+ 3,self_attn.q_proj,0.0000004040,0.05000,5.748
24
+ 3,self_attn.k_proj,0.0000001964,0.05000,5.757
25
+ 3,self_attn.v_proj,0.0000000324,0.05000,5.836
26
+ 3,self_attn.o_proj,0.0000000000,0.05000,1.507
27
+ 3,mlp.up_proj,0.0000003335,0.05000,2.726
28
+ 3,mlp.gate_proj,0.0000003861,0.05000,2.737
29
+ 3,mlp.down_proj,0.0000000001,0.05000,5.850
30
+ 4,self_attn.k_proj,0.0000002752,0.05000,5.885
31
+ 4,self_attn.q_proj,0.0000006219,0.05000,5.899
32
+ 4,self_attn.v_proj,0.0000000542,0.05000,5.920
33
+ 4,self_attn.o_proj,0.0000000000,0.05000,1.511
34
+ 4,mlp.gate_proj,0.0000005583,0.05000,2.707
35
+ 4,mlp.up_proj,0.0000004521,0.05000,2.712
36
+ 4,mlp.down_proj,0.0000000003,0.05000,5.795
37
+ 5,self_attn.v_proj,0.0000000601,0.05000,5.651
38
+ 5,self_attn.k_proj,0.0000003567,0.05000,5.685
39
+ 5,self_attn.q_proj,0.0000008041,0.05000,5.708
40
+ 5,self_attn.o_proj,0.0000000001,0.05000,1.530
41
+ 5,mlp.gate_proj,0.0000007742,0.05000,2.754
42
+ 5,mlp.up_proj,0.0000005881,0.05000,2.759
43
+ 5,mlp.down_proj,0.0000000004,0.05000,5.792
44
+ 6,self_attn.q_proj,0.0000007546,0.05000,5.715
45
+ 6,self_attn.v_proj,0.0000000610,0.05000,5.753
46
+ 6,self_attn.k_proj,0.0000003538,0.05000,5.763
47
+ 6,self_attn.o_proj,0.0000000001,0.05000,1.491
48
+ 6,mlp.up_proj,0.0000006991,0.05000,2.700
49
+ 6,mlp.gate_proj,0.0000008995,0.05000,2.717
50
+ 6,mlp.down_proj,0.0000000006,0.05000,5.790
51
+ 7,self_attn.k_proj,0.0000004439,0.05000,5.755
52
+ 7,self_attn.q_proj,0.0000009290,0.05000,5.785
53
+ 7,self_attn.v_proj,0.0000000806,0.05000,5.807
54
+ 7,self_attn.o_proj,0.0000000002,0.05000,1.512
55
+ 7,mlp.up_proj,0.0000008049,0.05000,2.738
56
+ 7,mlp.gate_proj,0.0000010599,0.05000,2.751
57
+ 7,mlp.down_proj,0.0000000008,0.05000,5.833
58
+ 8,self_attn.v_proj,0.0000000749,0.05000,5.644
59
+ 8,self_attn.k_proj,0.0000003480,0.05000,5.679
60
+ 8,self_attn.q_proj,0.0000007657,0.05000,5.711
61
+ 8,self_attn.o_proj,0.0000000003,0.05000,1.486
62
+ 8,mlp.gate_proj,0.0000011395,0.05000,2.710
63
+ 8,mlp.up_proj,0.0000008825,0.05000,2.716
64
+ 8,mlp.down_proj,0.0000000009,0.05000,5.785
65
+ 9,self_attn.k_proj,0.0000004894,0.05000,5.779
66
+ 9,self_attn.q_proj,0.0000010457,0.05000,5.846
67
+ 9,self_attn.v_proj,0.0000000885,0.05000,5.857
68
+ 9,self_attn.o_proj,0.0000000003,0.05000,1.521
69
+ 9,mlp.gate_proj,0.0000011959,0.05000,2.735
70
+ 9,mlp.up_proj,0.0000009581,0.05000,2.754
71
+ 9,mlp.down_proj,0.0000000011,0.05000,5.800
72
+ 10,self_attn.v_proj,0.0000000776,0.05000,5.689
73
+ 10,self_attn.k_proj,0.0000004430,0.05000,5.697
74
+ 10,self_attn.q_proj,0.0000009305,0.05000,5.710
75
+ 10,self_attn.o_proj,0.0000000006,0.05000,1.482
76
+ 10,mlp.gate_proj,0.0000012489,0.05000,2.718
77
+ 10,mlp.up_proj,0.0000010243,0.05000,2.723
78
+ 10,mlp.down_proj,0.0000000013,0.05000,5.783
79
+ 11,self_attn.q_proj,0.0000011214,0.05000,5.811
80
+ 11,self_attn.k_proj,0.0000005155,0.05000,5.824
81
+ 11,self_attn.v_proj,0.0000001141,0.05000,5.853
82
+ 11,self_attn.o_proj,0.0000000008,0.05000,1.511
83
+ 11,mlp.gate_proj,0.0000013628,0.05000,2.734
84
+ 11,mlp.up_proj,0.0000011338,0.05000,2.743
85
+ 11,mlp.down_proj,0.0000000015,0.05000,5.785
86
+ 12,self_attn.q_proj,0.0000014866,0.05000,5.701
87
+ 12,self_attn.v_proj,0.0000001297,0.05000,5.702
88
+ 12,self_attn.k_proj,0.0000006694,0.05000,5.701
89
+ 12,self_attn.o_proj,0.0000000009,0.05000,1.521
90
+ 12,mlp.up_proj,0.0000012766,0.05000,2.776
91
+ 12,mlp.gate_proj,0.0000015000,0.05000,2.783
92
+ 12,mlp.down_proj,0.0000000019,0.05000,5.699
93
+ 13,self_attn.k_proj,0.0000005692,0.05000,5.850
94
+ 13,self_attn.v_proj,0.0000001234,0.05000,5.908
95
+ 13,self_attn.q_proj,0.0000011665,0.05000,5.942
96
+ 13,self_attn.o_proj,0.0000000011,0.05000,1.479
97
+ 13,mlp.up_proj,0.0000014674,0.05000,2.721
98
+ 13,mlp.gate_proj,0.0000016768,0.05000,2.732
99
+ 13,mlp.down_proj,0.0000000024,0.05000,5.763
100
+ 14,self_attn.q_proj,0.0000013253,0.05000,5.565
101
+ 14,self_attn.k_proj,0.0000005664,0.05000,5.643
102
+ 14,self_attn.v_proj,0.0000001981,0.05000,5.653
103
+ 14,self_attn.o_proj,0.0000000014,0.05000,1.508
104
+ 14,mlp.gate_proj,0.0000018725,0.05000,2.713
105
+ 14,mlp.up_proj,0.0000016302,0.05000,2.741
106
+ 14,mlp.down_proj,0.0000000030,0.05000,5.708
107
+ 15,self_attn.k_proj,0.0000007359,0.05000,5.857
108
+ 15,self_attn.q_proj,0.0000016581,0.05000,5.863
109
+ 15,self_attn.v_proj,0.0000002239,0.05000,5.899
110
+ 15,self_attn.o_proj,0.0000000015,0.05000,1.513
111
+ 15,mlp.up_proj,0.0000017916,0.05000,2.713
112
+ 15,mlp.gate_proj,0.0000021294,0.05000,2.719
113
+ 15,mlp.down_proj,0.0000000038,0.05000,5.826
114
+ 16,self_attn.v_proj,0.0000002110,0.05000,5.704
115
+ 16,self_attn.k_proj,0.0000006766,0.05000,5.723
116
+ 16,self_attn.q_proj,0.0000014709,0.05000,5.736
117
+ 16,self_attn.o_proj,0.0000000017,0.05000,1.504
118
+ 16,mlp.up_proj,0.0000020874,0.05000,2.717
119
+ 16,mlp.gate_proj,0.0000026221,0.05000,2.732
120
+ 16,mlp.down_proj,0.0000000054,0.05000,5.778
121
+ 17,self_attn.v_proj,0.0000002017,0.05000,5.752
122
+ 17,self_attn.q_proj,0.0000013624,0.05000,5.840
123
+ 17,self_attn.k_proj,0.0000005692,0.05000,5.847
124
+ 17,self_attn.o_proj,0.0000000022,0.05000,1.490
125
+ 17,mlp.gate_proj,0.0000030491,0.05000,2.742
126
+ 17,mlp.up_proj,0.0000024313,0.05000,2.746
127
+ 17,mlp.down_proj,0.0000000074,0.05000,5.748
128
+ 18,self_attn.k_proj,0.0000006751,0.05000,5.822
129
+ 18,self_attn.v_proj,0.0000002480,0.05000,5.845
130
+ 18,self_attn.q_proj,0.0000017297,0.05000,5.888
131
+ 18,self_attn.o_proj,0.0000000020,0.05000,1.496
132
+ 18,mlp.gate_proj,0.0000034365,0.05000,2.732
133
+ 18,mlp.up_proj,0.0000027417,0.05000,2.739
134
+ 18,mlp.down_proj,0.0000000107,0.05000,5.821
135
+ 19,self_attn.v_proj,0.0000002874,0.05000,5.735
136
+ 19,self_attn.q_proj,0.0000015817,0.05000,5.766
137
+ 19,self_attn.k_proj,0.0000006702,0.05000,5.773
138
+ 19,self_attn.o_proj,0.0000000025,0.05000,1.524
139
+ 19,mlp.up_proj,0.0000030297,0.05000,2.756
140
+ 19,mlp.gate_proj,0.0000038671,0.05000,2.773
141
+ 19,mlp.down_proj,0.0000000143,0.05000,5.762
142
+ 20,self_attn.q_proj,0.0000016975,0.05000,5.790
143
+ 20,self_attn.k_proj,0.0000006890,0.05000,5.835
144
+ 20,self_attn.v_proj,0.0000003146,0.05000,5.849
145
+ 20,self_attn.o_proj,0.0000000023,0.05000,1.505
146
+ 20,mlp.up_proj,0.0000033502,0.05000,2.737
147
+ 20,mlp.gate_proj,0.0000043862,0.05000,2.745
148
+ 20,mlp.down_proj,0.0000000149,0.05000,5.762
149
+ 21,self_attn.q_proj,0.0000016742,0.05000,5.741
150
+ 21,self_attn.v_proj,0.0000003297,0.05000,5.744
151
+ 21,self_attn.k_proj,0.0000006758,0.05000,5.762
152
+ 21,self_attn.o_proj,0.0000000024,0.05000,1.527
153
+ 21,mlp.up_proj,0.0000036733,0.05000,2.724
154
+ 21,mlp.gate_proj,0.0000051593,0.05000,2.730
155
+ 21,mlp.down_proj,0.0000000152,0.05000,5.811
156
+ 22,self_attn.k_proj,0.0000006403,0.05000,5.796
157
+ 22,self_attn.v_proj,0.0000003419,0.05000,5.830
158
+ 22,self_attn.q_proj,0.0000016118,0.05000,5.901
159
+ 22,self_attn.o_proj,0.0000000014,0.05000,1.528
160
+ 22,mlp.up_proj,0.0000039479,0.05000,2.729
161
+ 22,mlp.gate_proj,0.0000054861,0.05000,2.733
162
+ 22,mlp.down_proj,0.0000000157,0.05000,5.797
163
+ 23,self_attn.v_proj,0.0000003579,0.05000,5.894
164
+ 23,self_attn.k_proj,0.0000006380,0.05000,5.895
165
+ 23,self_attn.q_proj,0.0000016239,0.05000,5.922
166
+ 23,self_attn.o_proj,0.0000000023,0.05000,1.531
167
+ 23,mlp.gate_proj,0.0000060618,0.05000,2.698
168
+ 23,mlp.up_proj,0.0000043452,0.05000,2.699
169
+ 23,mlp.down_proj,0.0000000177,0.05000,5.782
170
+ 24,self_attn.k_proj,0.0000007371,0.05000,5.596
171
+ 24,self_attn.q_proj,0.0000018533,0.05000,5.663
172
+ 24,self_attn.v_proj,0.0000004165,0.05000,5.676
173
+ 24,self_attn.o_proj,0.0000000021,0.05000,1.492
174
+ 24,mlp.gate_proj,0.0000068435,0.05000,2.771
175
+ 24,mlp.up_proj,0.0000048163,0.05000,2.777
176
+ 24,mlp.down_proj,0.0000000192,0.05000,5.771
177
+ 25,self_attn.k_proj,0.0000007104,0.05000,5.831
178
+ 25,self_attn.v_proj,0.0000004764,0.05000,5.842
179
+ 25,self_attn.q_proj,0.0000018583,0.05000,5.841
180
+ 25,self_attn.o_proj,0.0000000024,0.05000,1.509
181
+ 25,mlp.up_proj,0.0000052669,0.05000,2.721
182
+ 25,mlp.gate_proj,0.0000074224,0.05000,2.730
183
+ 25,mlp.down_proj,0.0000000218,0.05000,5.820
184
+ 26,self_attn.v_proj,0.0000005038,0.05000,5.621
185
+ 26,self_attn.k_proj,0.0000006691,0.05000,5.630
186
+ 26,self_attn.q_proj,0.0000017644,0.05000,5.648
187
+ 26,self_attn.o_proj,0.0000000033,0.05000,1.515
188
+ 26,mlp.up_proj,0.0000057272,0.05000,2.755
189
+ 26,mlp.gate_proj,0.0000077866,0.05000,2.775
190
+ 26,mlp.down_proj,0.0000000250,0.05000,5.808
191
+ 27,self_attn.q_proj,0.0000018230,0.05000,5.916
192
+ 27,self_attn.v_proj,0.0000004564,0.05000,5.989
193
+ 27,self_attn.k_proj,0.0000006762,0.05000,6.016
194
+ 27,self_attn.o_proj,0.0000000037,0.05000,1.485
195
+ 27,mlp.up_proj,0.0000063079,0.05000,2.725
196
+ 27,mlp.gate_proj,0.0000085269,0.05000,2.739
197
+ 27,mlp.down_proj,0.0000000300,0.05000,5.801
198
+ 28,self_attn.v_proj,0.0000006713,0.05000,5.844
199
+ 28,self_attn.k_proj,0.0000006783,0.05000,5.868
200
+ 28,self_attn.q_proj,0.0000017780,0.05000,5.884
201
+ 28,self_attn.o_proj,0.0000000056,0.05000,1.507
202
+ 28,mlp.up_proj,0.0000070183,0.05000,2.716
203
+ 28,mlp.gate_proj,0.0000091368,0.05000,2.723
204
+ 28,mlp.down_proj,0.0000000409,0.05000,5.721
205
+ 29,self_attn.v_proj,0.0000009821,0.05000,5.830
206
+ 29,self_attn.k_proj,0.0000006843,0.05000,5.854
207
+ 29,self_attn.q_proj,0.0000020354,0.05000,5.860
208
+ 29,self_attn.o_proj,0.0000000119,0.05000,1.516
209
+ 29,mlp.gate_proj,0.0000088944,0.05000,2.812
210
+ 29,mlp.up_proj,0.0000071220,0.05000,2.825
211
+ 29,mlp.down_proj,0.0000000519,0.05000,5.835
212
+ 30,self_attn.q_proj,0.0000018947,0.05000,5.718
213
+ 30,self_attn.k_proj,0.0000006325,0.05000,5.845
214
+ 30,self_attn.v_proj,0.0000010680,0.05000,5.874
215
+ 30,self_attn.o_proj,0.0000000106,0.05000,1.507
216
+ 30,mlp.gate_proj,0.0000089889,0.05000,2.782
217
+ 30,mlp.up_proj,0.0000073173,0.05000,2.796
218
+ 30,mlp.down_proj,0.0000000681,0.05000,5.731
219
+ 31,self_attn.v_proj,0.0000010866,0.05000,5.793
220
+ 31,self_attn.k_proj,0.0000006014,0.05000,5.846
221
+ 31,self_attn.q_proj,0.0000017606,0.05000,5.861
222
+ 31,self_attn.o_proj,0.0000000142,0.05000,1.512
223
+ 31,mlp.gate_proj,0.0000075029,0.05000,2.765
224
+ 31,mlp.up_proj,0.0000059300,0.05000,2.774
225
+ 31,mlp.down_proj,0.0000001083,0.05000,5.806
quantize_config.json ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bits": 8,
3
+ "group_size": 128,
4
+ "desc_act": false,
5
+ "lm_head": false,
6
+ "method": "gptq",
7
+ "quant_method": "gptq",
8
+ "format": "gptq",
9
+ "checkpoint_format": "gptq",
10
+ "pack_dtype": "int32",
11
+ "meta": {
12
+ "quantizer": [
13
+ "gptqmodel:6.0.3"
14
+ ],
15
+ "uri": "https://github.com/modelcloud/gptqmodel",
16
+ "damp_percent": 0.05,
17
+ "damp_auto_increment": 0.01,
18
+ "static_groups": false,
19
+ "true_sequential": true,
20
+ "mse": 0.0,
21
+ "gptaq": null,
22
+ "foem": null,
23
+ "act_group_aware": true,
24
+ "fallback": {
25
+ "strategy": "rtn",
26
+ "threshold": "0.5%",
27
+ "smooth": null
28
+ },
29
+ "offload_to_disk": true,
30
+ "offload_to_disk_path": "./gptqmodel_offload/hedxansx-hkyyfauy/",
31
+ "pack_impl": "cpu",
32
+ "gc_mode": "interval",
33
+ "wait_for_submodule_finalizers": false,
34
+ "auto_forward_data_parallel": true,
35
+ "vram_strategy": "exclusive",
36
+ "mock_quantization": false,
37
+ "hessian": {
38
+ "chunk_size": null,
39
+ "chunk_bytes": null,
40
+ "staging_dtype": "float32"
41
+ }
42
+ },
43
+ "sym": true
44
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_prefix_space": true,
3
+ "backend": "tokenizers",
4
+ "bos_token": "<s>",
5
+ "clean_up_tokenization_spaces": false,
6
+ "eos_token": "</s>",
7
+ "is_local": false,
8
+ "legacy": false,
9
+ "model_max_length": 1000000000000000019884624838656,
10
+ "pad_token": null,
11
+ "sp_model_kwargs": {},
12
+ "spaces_between_special_tokens": false,
13
+ "tokenizer_class": "TokenizersBackend",
14
+ "unk_token": "<unk>",
15
+ "use_default_system_prompt": false
16
+ }