llmstack/qwen3_4b_gpro_4bit
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity

初始化项目，由ModelHub XC社区提供模型

Browse Source 
Model: llmstack/qwen3_4b_gpro_4bit
Source: Original Platform
This commit is contained in:
ModelHub XC
2026-08-05 13:51:17 +08:00
commit 52378361cc
9 changed files with 577 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

36
.gitattributes vendored Normal file
View File

@@ -0,0 +1,36 @@
*.7z filter=lfs diff=lfs merge=lfs -text
*.arrow filter=lfs diff=lfs merge=lfs -text
*.bin filter=lfs diff=lfs merge=lfs -text
*.bz2 filter=lfs diff=lfs merge=lfs -text
*.ckpt filter=lfs diff=lfs merge=lfs -text
*.ftz filter=lfs diff=lfs merge=lfs -text
*.gz filter=lfs diff=lfs merge=lfs -text
*.h5 filter=lfs diff=lfs merge=lfs -text
*.joblib filter=lfs diff=lfs merge=lfs -text
*.lfs.* filter=lfs diff=lfs merge=lfs -text
*.mlmodel filter=lfs diff=lfs merge=lfs -text
*.model filter=lfs diff=lfs merge=lfs -text
*.msgpack filter=lfs diff=lfs merge=lfs -text
*.npy filter=lfs diff=lfs merge=lfs -text
*.npz filter=lfs diff=lfs merge=lfs -text
*.onnx filter=lfs diff=lfs merge=lfs -text
*.ot filter=lfs diff=lfs merge=lfs -text
*.parquet filter=lfs diff=lfs merge=lfs -text
*.pb filter=lfs diff=lfs merge=lfs -text
*.pickle filter=lfs diff=lfs merge=lfs -text
*.pkl filter=lfs diff=lfs merge=lfs -text
*.pt filter=lfs diff=lfs merge=lfs -text
*.pth filter=lfs diff=lfs merge=lfs -text
*.rar filter=lfs diff=lfs merge=lfs -text
*.safetensors filter=lfs diff=lfs merge=lfs -text
saved_model/**/* filter=lfs diff=lfs merge=lfs -text
*.tar.* filter=lfs diff=lfs merge=lfs -text
*.tar filter=lfs diff=lfs merge=lfs -text
*.tflite filter=lfs diff=lfs merge=lfs -text
*.tgz filter=lfs diff=lfs merge=lfs -text
*.wasm filter=lfs diff=lfs merge=lfs -text
*.xz filter=lfs diff=lfs merge=lfs -text
*.zip filter=lfs diff=lfs merge=lfs -text
*.zst filter=lfs diff=lfs merge=lfs -text
*tfevents* filter=lfs diff=lfs merge=lfs -text
tokenizer.json filter=lfs diff=lfs merge=lfs -text

86
chat_template.jinja Normal file
View File

@@ -0,0 +1,86 @@
{%- if tools %}
{{- '<|im_start|>system\n' }}
{%- if messages[0].role == 'system' %}
{{- messages[0].content + '\n\n' }}
{%- endif %}
{{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
{%- for tool in tools %}
{{- "\n" }}
{{- tool | tojson }}
{%- endfor %}
{{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
{%- else %}
{%- if messages[0].role == 'system' %}
{{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
{%- endif %}
{%- endif %}
{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
{%- for message in messages[::-1] %}
{%- set index = (messages|length - 1) - loop.index0 %}
{%- if ns.multi_step_tool and message.role == "user" and message.content is string and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
{%- set ns.multi_step_tool = false %}
{%- set ns.last_query_index = index %}
{%- endif %}
{%- endfor %}
{%- for message in messages %}
{%- if message.content is string %}
{%- set content = message.content %}
{%- else %}
{%- set content = '' %}
{%- endif %}
{%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
{{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
{%- elif message.role == "assistant" %}
{%- set reasoning_content = '' %}
{%- if message.reasoning_content is string %}
{%- set reasoning_content = message.reasoning_content %}
{%- else %}
{%- if '</think>' in content %}
{%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
{%- set content = content.split('</think>')[-1].lstrip('\n') %}
{%- endif %}
{%- endif %}
{%- if loop.index0 > ns.last_query_index %}
{%- if reasoning_content %}
{{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
{%- else %}
{{- '<|im_start|>' + message.role + '\n' + content }}
{%- endif %}
{%- else %}
{{- '<|im_start|>' + message.role + '\n' + content }}
{%- endif %}
{%- if message.tool_calls %}
{%- for tool_call in message.tool_calls %}
{%- if (loop.first and content) or (not loop.first) %}
{{- '\n' }}
{%- endif %}
{%- if tool_call.function %}
{%- set tool_call = tool_call.function %}
{%- endif %}
{{- '<tool_call>\n{"name": "' }}
{{- tool_call.name }}
{{- '", "arguments": ' }}
{%- if tool_call.arguments is string %}
{{- tool_call.arguments }}
{%- else %}
{{- tool_call.arguments | tojson }}
{%- endif %}
{{- '}\n</tool_call>' }}
{%- endfor %}
{%- endif %}
{{- '<|im_end|>\n' }}
{%- elif message.role == "tool" %}
{%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
{{- '<|im_start|>user' }}
{%- endif %}
{{- '\n<tool_response>\n' }}
{{- content }}
{{- '\n</tool_response>' }}
{%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
{{- '<|im_end|>\n' }}
{%- endif %}
{%- endif %}
{%- endfor %}
{%- if add_generation_prompt %}
{{- '<|im_start|>assistant\n' }}
{%- endif %}

120
config.json Normal file
View File

@@ -0,0 +1,120 @@
{
"architectures": [
"Qwen3ForCausalLM"
],
"attention_bias": false,
"attention_dropout": 0.0,
"bos_token_id": null,
"dtype": "float16",
"eos_token_id": 151645,
"head_dim": 128,
"hidden_act": "silu",
"hidden_size": 2560,
"initializer_range": 0.02,
"intermediate_size": 9728,
"layer_types": [
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention",
"full_attention"
],
"max_position_embeddings": 262144,
"max_window_layers": 36,
"model_type": "qwen3",
"num_attention_heads": 32,
"num_hidden_layers": 36,
"num_key_value_heads": 8,
"pad_token_id": 151654,
"quantization_config": {
"bits": 4,
"checkpoint_format": "gptq",
"desc_act": false,
"format": "gptq",
"group_size": 128,
"lm_head": false,
"meta": {
"act_group_aware": true,
"auto_forward_data_parallel": true,
"damp_auto_increment": 0.01,
"damp_percent": 0.1,
"dense_vram_strategy": "exclusive",
"dense_vram_strategy_devices": null,
"fallback": {
"smooth": null,
"strategy": "rtn",
"threshold": "0.5%"
},
"foem": null,
"gc_mode": "interval",
"gptaq": null,
"hessian": {
"chunk_bytes": null,
"chunk_size": null,
"staging_dtype": "float32"
},
"mock_quantization": false,
"moe_vram_strategy": "exclusive",
"moe_vram_strategy_devices": null,
"mse": 0.0,
"offload_to_disk": true,
"offload_to_disk_path": "/tmp/gptqmodel_pf205jdd",
"pack_impl": "cpu",
"quantizer": [
"gptqmodel:7.1.0"
],
"static_groups": false,
"true_sequential": true,
"uri": "https://github.com/modelcloud/gptqmodel",
"wait_for_submodule_finalizers": false
},
"method": "gptq",
"pack_dtype": "int32",
"quant_method": "gptq",
"sym": true
},
"rms_norm_eps": 1e-06,
"rope_parameters": {
"rope_theta": 5000000,
"rope_type": "default"
},
"sliding_window": null,
"tie_word_embeddings": true,
"transformers_version": "5.14.1",
"unsloth_fixed": true,
"unsloth_version": "2026.7.2",
"use_cache": true,
"use_sliding_window": false,
"vocab_size": 151936
}

13
generation_config.json Normal file
View File

@@ -0,0 +1,13 @@
{
"do_sample": true,
"eos_token_id": [
151645,
151643
],
"max_length": 262144,
"pad_token_id": 151654,
"temperature": 0.7,
"top_k": 20,
"top_p": 0.8,
"transformers_version": "5.14.1"
}

3
model.safetensors Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:a94685697a6197d8da6d6a3cc4cb592639ca98be6f6c3c8735402a20ab7871cd
size 2669888848

253
quant_log.csv Normal file
View File

@@ -0,0 +1,253 @@
layer,module,loss,samples,damp,time
0,self_attn.k_proj,0.0000000065,0.10000,4.623
0,self_attn.q_proj,0.0000000241,0.10000,5.088
0,self_attn.v_proj,0.0000000058,0.10000,4.741
0,self_attn.o_proj,0.0000000147,0.10000,1.642
0,mlp.gate_proj,0.0000011763,0.10000,1.887
0,mlp.up_proj,0.0000010297,0.10000,1.908
0,mlp.down_proj,0.0000000892,0.10000,4.448
1,self_attn.k_proj,0.0000000102,0.10000,4.257
1,self_attn.v_proj,0.0000000106,0.10000,4.563
1,self_attn.q_proj,0.0000000381,0.10000,4.579
1,self_attn.o_proj,0.0000000215,0.10000,1.522
1,mlp.up_proj,0.0000345402,0.10000,1.994
1,mlp.gate_proj,0.0000774305,0.10000,2.020
1,mlp.down_proj,0.0000001089,0.10000,4.642
2,self_attn.k_proj,0.0000000278,0.10000,4.120
2,self_attn.v_proj,0.0000000273,0.10000,4.477
2,self_attn.q_proj,0.0000000974,0.10000,4.482
2,self_attn.o_proj,0.0000000296,0.10000,1.632
2,mlp.up_proj,0.0000956081,0.10000,2.117
2,mlp.gate_proj,0.0001185081,0.10000,2.137
2,mlp.down_proj,0.0000000926,0.10000,4.488
3,self_attn.k_proj,0.0000000519,0.10000,4.306
3,self_attn.v_proj,0.0000000551,0.10000,4.533
3,self_attn.q_proj,0.0000002047,0.10000,4.572
3,self_attn.o_proj,0.0000000612,0.10000,1.598
3,mlp.up_proj,0.0000566981,0.10000,2.041
3,mlp.gate_proj,0.0001075486,0.10000,2.091
3,mlp.down_proj,0.0000002804,0.10000,4.618
4,self_attn.k_proj,0.0000000993,0.10000,4.159
4,self_attn.q_proj,0.0000003704,0.10000,4.445
4,self_attn.v_proj,0.0000001018,0.10000,4.500
4,self_attn.o_proj,0.0000000856,0.10000,1.605
4,mlp.gate_proj,0.0000857805,0.10000,2.096
4,mlp.up_proj,0.0000351438,0.10000,2.128
4,mlp.down_proj,0.0000006261,0.10000,4.543
5,self_attn.k_proj,0.0000001043,0.10000,4.181
5,self_attn.q_proj,0.0000004168,0.10000,4.518
5,self_attn.v_proj,0.0000001154,0.10000,4.543
5,self_attn.o_proj,0.0000001148,0.10000,1.588
5,mlp.up_proj,0.0000128055,0.10000,1.963
5,mlp.gate_proj,0.0000251472,0.10000,1.991
5,mlp.down_proj,0.0000007686,0.10000,4.505
6,self_attn.k_proj,0.0000002189,0.10000,4.171
6,self_attn.v_proj,0.0000002524,0.10000,4.485
6,self_attn.q_proj,0.0000009372,0.10000,4.513
6,self_attn.o_proj,0.0000002816,0.10000,1.569
6,mlp.gate_proj,0.0000282582,0.10000,2.038
6,mlp.up_proj,0.0000170360,0.10000,2.078
6,mlp.down_proj,0.0000238226,0.10000,4.473
7,self_attn.k_proj,0.0000005651,0.10000,4.312
7,self_attn.v_proj,0.0000006153,0.10000,4.532
7,self_attn.q_proj,0.0000021191,0.10000,4.541
7,self_attn.o_proj,0.0000003873,0.10000,1.559
7,mlp.up_proj,0.0000209582,0.10000,2.027
7,mlp.gate_proj,0.0000333655,0.10000,2.044
7,mlp.down_proj,0.0000011662,0.10000,4.485
8,self_attn.k_proj,0.0000007382,0.10000,4.366
8,self_attn.v_proj,0.0000008672,0.10000,4.553
8,self_attn.q_proj,0.0000029186,0.10000,4.584
8,self_attn.o_proj,0.0000004689,0.10000,1.625
8,mlp.gate_proj,0.0000198027,0.10000,2.106
8,mlp.up_proj,0.0000163943,0.10000,2.119
8,mlp.down_proj,0.0000015403,0.10000,4.599
9,self_attn.k_proj,0.0000009926,0.10000,4.345
9,self_attn.v_proj,0.0000010791,0.10000,4.597
9,self_attn.q_proj,0.0000036375,0.10000,4.625
9,self_attn.o_proj,0.0000006776,0.10000,1.550
9,mlp.up_proj,0.0000205884,0.10000,2.038
9,mlp.gate_proj,0.0000315698,0.10000,2.079
9,mlp.down_proj,0.0000016813,0.10000,4.645
10,self_attn.k_proj,0.0000013627,0.10000,4.412
10,self_attn.q_proj,0.0000053466,0.10000,4.665
10,self_attn.v_proj,0.0000016548,0.10000,4.706
10,self_attn.o_proj,0.0000007416,0.10000,1.613
10,mlp.gate_proj,0.0000238936,0.10000,2.020
10,mlp.up_proj,0.0000168923,0.10000,2.067
10,mlp.down_proj,0.0000015627,0.10000,4.520
11,self_attn.k_proj,0.0000007019,0.10000,4.288
11,self_attn.v_proj,0.0000007862,0.10000,4.550
11,self_attn.q_proj,0.0000026579,0.10000,4.598
11,self_attn.o_proj,0.0000006333,0.10000,1.561
11,mlp.up_proj,0.0000156422,0.10000,1.964
11,mlp.gate_proj,0.0000196474,0.10000,1.983
11,mlp.down_proj,0.0000015437,0.10000,4.595
12,self_attn.k_proj,0.0000008867,0.10000,4.297
12,self_attn.q_proj,0.0000034566,0.10000,4.574
12,self_attn.v_proj,0.0000010242,0.10000,4.616
12,self_attn.o_proj,0.0000010202,0.10000,1.577
12,mlp.gate_proj,0.0000181967,0.10000,2.092
12,mlp.up_proj,0.0000159937,0.10000,2.128
12,mlp.down_proj,0.0000016049,0.10000,4.597
13,self_attn.k_proj,0.0000006896,0.10000,4.313
13,self_attn.q_proj,0.0000026921,0.10000,4.535
13,self_attn.v_proj,0.0000007320,0.10000,4.566
13,self_attn.o_proj,0.0000008842,0.10000,1.626
13,mlp.up_proj,0.0000177848,0.10000,1.996
13,mlp.gate_proj,0.0000184155,0.10000,2.023
13,mlp.down_proj,0.0000019769,0.10000,4.748
14,self_attn.k_proj,0.0000012724,0.10000,4.292
14,self_attn.v_proj,0.0000014461,0.10000,4.545
14,self_attn.q_proj,0.0000050922,0.10000,4.607
14,self_attn.o_proj,0.0000014442,0.10000,1.618
14,mlp.gate_proj,0.0000188562,0.10000,2.122
14,mlp.up_proj,0.0000183811,0.10000,2.164
14,mlp.down_proj,0.0000018007,0.10000,4.510
15,self_attn.k_proj,0.0000012884,0.10000,4.345
15,self_attn.q_proj,0.0000051236,0.10000,4.570
15,self_attn.v_proj,0.0000013641,0.10000,4.614
15,self_attn.o_proj,0.0000014086,0.10000,1.631
15,mlp.up_proj,0.0000184349,0.10000,2.067
15,mlp.gate_proj,0.0000181649,0.10000,2.091
15,mlp.down_proj,0.0000018890,0.10000,4.577
16,self_attn.k_proj,0.0000021029,0.10000,4.328
16,self_attn.v_proj,0.0000025386,0.10000,4.556
16,self_attn.q_proj,0.0000086758,0.10000,4.568
16,self_attn.o_proj,0.0000012714,0.10000,1.562
16,mlp.gate_proj,0.0000200963,0.10000,1.935
16,mlp.up_proj,0.0000192835,0.10000,1.973
16,mlp.down_proj,0.0000187656,0.10000,4.514
17,self_attn.k_proj,0.0000017093,0.10000,4.341
17,self_attn.q_proj,0.0000077938,0.10000,4.594
17,self_attn.v_proj,0.0000020613,0.10000,4.622
17,self_attn.o_proj,0.0000016575,0.10000,1.643
17,mlp.up_proj,0.0000181763,0.10000,2.057
17,mlp.gate_proj,0.0000186384,0.10000,2.093
17,mlp.down_proj,0.0000018738,0.10000,4.600
18,self_attn.k_proj,0.0000020693,0.10000,4.363
18,self_attn.q_proj,0.0000086855,0.10000,4.631
18,self_attn.v_proj,0.0000025088,0.10000,4.649
18,self_attn.o_proj,0.0000017786,0.10000,1.606
18,mlp.gate_proj,0.0000201907,0.10000,2.099
18,mlp.up_proj,0.0000200689,0.10000,2.166
18,mlp.down_proj,0.0000023515,0.10000,4.576
19,self_attn.k_proj,0.0000038297,0.10000,4.432
19,self_attn.v_proj,0.0000044097,0.10000,4.653
19,self_attn.q_proj,0.0000169872,0.10000,4.682
19,self_attn.o_proj,0.0000028907,0.10000,1.582
19,mlp.up_proj,0.0000229501,0.10000,2.107
19,mlp.gate_proj,0.0000226941,0.10000,2.159
19,mlp.down_proj,0.0000030679,0.10000,4.579
20,self_attn.k_proj,0.0000036831,0.10000,4.317
20,self_attn.v_proj,0.0000043706,0.10000,4.591
20,self_attn.q_proj,0.0000174276,0.10000,4.627
20,self_attn.o_proj,0.0000029771,0.10000,1.599
20,mlp.gate_proj,0.0000253506,0.10000,2.059
20,mlp.up_proj,0.0000262646,0.10000,2.086
20,mlp.down_proj,0.0000036408,0.10000,4.510
21,self_attn.k_proj,0.0000050001,0.10000,4.363
21,self_attn.q_proj,0.0000228820,0.10000,4.653
21,self_attn.v_proj,0.0000059135,0.10000,4.667
21,self_attn.o_proj,0.0000029943,0.10000,1.538
21,mlp.gate_proj,0.0000304431,0.10000,2.065
21,mlp.up_proj,0.0000316683,0.10000,2.093
21,mlp.down_proj,0.0000045086,0.10000,4.625
22,self_attn.k_proj,0.0000092675,0.10000,4.294
22,self_attn.q_proj,0.0000424574,0.10000,4.589
22,self_attn.v_proj,0.0000114349,0.10000,4.613
22,self_attn.o_proj,0.0000045903,0.10000,1.557
22,mlp.gate_proj,0.0000354832,0.10000,2.096
22,mlp.up_proj,0.0000353631,0.10000,2.145
22,mlp.down_proj,0.0000065555,0.10000,4.581
23,self_attn.k_proj,0.0000091701,0.10000,4.238
23,self_attn.v_proj,0.0000109907,0.10000,4.710
23,self_attn.q_proj,0.0000420592,0.10000,4.755
23,self_attn.o_proj,0.0000054694,0.10000,1.579
23,mlp.up_proj,0.0000414041,0.10000,2.125
23,mlp.gate_proj,0.0000440309,0.10000,2.163
23,mlp.down_proj,0.0000084183,0.10000,4.566
24,self_attn.k_proj,0.0000152363,0.10000,4.191
24,self_attn.q_proj,0.0000699393,0.10000,4.506
24,self_attn.v_proj,0.0000197641,0.10000,4.548
24,self_attn.o_proj,0.0000070709,0.10000,1.614
24,mlp.gate_proj,0.0000503237,0.10000,2.141
24,mlp.up_proj,0.0000460797,0.10000,2.191
24,mlp.down_proj,0.0000104877,0.10000,4.497
25,self_attn.k_proj,0.0000114676,0.10000,4.133
25,self_attn.q_proj,0.0000481714,0.10000,4.443
25,self_attn.v_proj,0.0000137399,0.10000,4.481
25,self_attn.o_proj,0.0000035874,0.10000,1.621
25,mlp.up_proj,0.0000504789,0.10000,2.020
25,mlp.gate_proj,0.0000562083,0.10000,2.065
25,mlp.down_proj,0.0000124710,0.10000,4.571
26,self_attn.k_proj,0.0000163119,0.10000,4.307
26,self_attn.v_proj,0.0000199003,0.10000,4.560
26,self_attn.q_proj,0.0000721851,0.10000,4.593
26,self_attn.o_proj,0.0000030045,0.10000,1.607
26,mlp.gate_proj,0.0000650178,0.10000,1.944
26,mlp.up_proj,0.0000600384,0.10000,2.016
26,mlp.down_proj,0.0000118864,0.10000,4.514
27,self_attn.k_proj,0.0000191612,0.10000,4.194
27,self_attn.v_proj,0.0000258756,0.10000,4.484
27,self_attn.q_proj,0.0000881676,0.10000,4.499
27,self_attn.o_proj,0.0000036633,0.10000,1.582
27,mlp.up_proj,0.0000666810,0.10000,2.011
27,mlp.gate_proj,0.0000699586,0.10000,2.086
27,mlp.down_proj,0.0000171271,0.10000,4.543
28,self_attn.k_proj,0.0000253908,0.10000,3.950
28,self_attn.v_proj,0.0000305473,0.10000,4.400
28,self_attn.q_proj,0.0001051568,0.10000,4.419
28,self_attn.o_proj,0.0000075650,0.10000,1.628
28,mlp.gate_proj,0.0000770289,0.10000,2.025
28,mlp.up_proj,0.0000763184,0.10000,2.118
28,mlp.down_proj,0.0000230418,0.10000,4.608
29,self_attn.k_proj,0.0000528532,0.10000,4.391
29,self_attn.v_proj,0.0000706905,0.10000,4.621
29,self_attn.q_proj,0.0002377818,0.10000,4.649
29,self_attn.o_proj,0.0000114592,0.10000,1.577
29,mlp.up_proj,0.0000926004,0.10000,2.121
29,mlp.gate_proj,0.0000892738,0.10000,2.158
29,mlp.down_proj,0.0000262264,0.10000,4.575
30,self_attn.k_proj,0.0000615581,0.10000,4.263
30,self_attn.q_proj,0.0002475710,0.10000,4.521
30,self_attn.v_proj,0.0000790300,0.10000,4.546
30,self_attn.o_proj,0.0000080313,0.10000,1.577
30,mlp.gate_proj,0.0000938009,0.10000,1.948
30,mlp.up_proj,0.0001005060,0.10000,1.981
30,mlp.down_proj,0.0000340703,0.10000,4.439
31,self_attn.k_proj,0.0000925253,0.10000,4.279
31,self_attn.q_proj,0.0003612229,0.10000,4.514
31,self_attn.v_proj,0.0001289460,0.10000,4.533
31,self_attn.o_proj,0.0000124489,0.10000,1.552
31,mlp.up_proj,0.0001066470,0.10000,2.058
31,mlp.gate_proj,0.0000948500,0.10000,2.078
31,mlp.down_proj,0.0000454290,0.10000,4.558
32,self_attn.k_proj,0.0001231427,0.10000,4.320
32,self_attn.q_proj,0.0005124021,0.10000,4.625
32,self_attn.v_proj,0.0001872253,0.10000,4.658
32,self_attn.o_proj,0.0000203278,0.10000,1.620
32,mlp.gate_proj,0.0001001873,0.10000,1.977
32,mlp.up_proj,0.0001154902,0.10000,2.016
32,mlp.down_proj,0.0000601530,0.10000,4.641
33,self_attn.k_proj,0.0002222856,0.10000,4.128
33,self_attn.q_proj,0.0011057818,0.10000,4.438
33,self_attn.v_proj,0.0004192334,0.10000,4.453
33,self_attn.o_proj,0.0000454634,0.10000,1.567
33,mlp.up_proj,0.0001329813,0.10000,2.029
33,mlp.gate_proj,0.0001132789,0.10000,2.081
33,mlp.down_proj,0.0001011796,0.10000,4.507
34,self_attn.k_proj,0.0001888980,0.10000,4.234
34,self_attn.v_proj,0.0003253079,0.10000,4.554
34,self_attn.q_proj,0.0008556856,0.10000,4.575
34,self_attn.o_proj,0.0000674757,0.10000,1.632
34,mlp.gate_proj,0.0001431082,0.10000,2.044
34,mlp.up_proj,0.0001571695,0.10000,2.083
34,mlp.down_proj,0.0001871779,0.10000,4.492
35,self_attn.k_proj,0.0000936730,0.10000,4.291
35,self_attn.v_proj,0.0001312502,0.10000,4.533
35,self_attn.q_proj,0.0003771549,0.10000,4.553
35,self_attn.o_proj,0.0000736388,0.10000,1.576
35,mlp.up_proj,0.0002436513,0.10000,1.918
35,mlp.gate_proj,0.0002332021,0.10000,1.968
35,mlp.down_proj,0.0006246001,0.10000,4.539
1 layer,module,loss,samples,damp,time
2 0,self_attn.k_proj,0.0000000065,0.10000,4.623
3 0,self_attn.q_proj,0.0000000241,0.10000,5.088
4 0,self_attn.v_proj,0.0000000058,0.10000,4.741
5 0,self_attn.o_proj,0.0000000147,0.10000,1.642
6 0,mlp.gate_proj,0.0000011763,0.10000,1.887
7 0,mlp.up_proj,0.0000010297,0.10000,1.908
8 0,mlp.down_proj,0.0000000892,0.10000,4.448
9 1,self_attn.k_proj,0.0000000102,0.10000,4.257
10 1,self_attn.v_proj,0.0000000106,0.10000,4.563
11 1,self_attn.q_proj,0.0000000381,0.10000,4.579
12 1,self_attn.o_proj,0.0000000215,0.10000,1.522
13 1,mlp.up_proj,0.0000345402,0.10000,1.994
14 1,mlp.gate_proj,0.0000774305,0.10000,2.020
15 1,mlp.down_proj,0.0000001089,0.10000,4.642
16 2,self_attn.k_proj,0.0000000278,0.10000,4.120
17 2,self_attn.v_proj,0.0000000273,0.10000,4.477
18 2,self_attn.q_proj,0.0000000974,0.10000,4.482
19 2,self_attn.o_proj,0.0000000296,0.10000,1.632
20 2,mlp.up_proj,0.0000956081,0.10000,2.117
21 2,mlp.gate_proj,0.0001185081,0.10000,2.137
22 2,mlp.down_proj,0.0000000926,0.10000,4.488
23 3,self_attn.k_proj,0.0000000519,0.10000,4.306
24 3,self_attn.v_proj,0.0000000551,0.10000,4.533
25 3,self_attn.q_proj,0.0000002047,0.10000,4.572
26 3,self_attn.o_proj,0.0000000612,0.10000,1.598
27 3,mlp.up_proj,0.0000566981,0.10000,2.041
28 3,mlp.gate_proj,0.0001075486,0.10000,2.091
29 3,mlp.down_proj,0.0000002804,0.10000,4.618
30 4,self_attn.k_proj,0.0000000993,0.10000,4.159
31 4,self_attn.q_proj,0.0000003704,0.10000,4.445
32 4,self_attn.v_proj,0.0000001018,0.10000,4.500
33 4,self_attn.o_proj,0.0000000856,0.10000,1.605
34 4,mlp.gate_proj,0.0000857805,0.10000,2.096
35 4,mlp.up_proj,0.0000351438,0.10000,2.128
36 4,mlp.down_proj,0.0000006261,0.10000,4.543
37 5,self_attn.k_proj,0.0000001043,0.10000,4.181
38 5,self_attn.q_proj,0.0000004168,0.10000,4.518
39 5,self_attn.v_proj,0.0000001154,0.10000,4.543
40 5,self_attn.o_proj,0.0000001148,0.10000,1.588
41 5,mlp.up_proj,0.0000128055,0.10000,1.963
42 5,mlp.gate_proj,0.0000251472,0.10000,1.991
43 5,mlp.down_proj,0.0000007686,0.10000,4.505
44 6,self_attn.k_proj,0.0000002189,0.10000,4.171
45 6,self_attn.v_proj,0.0000002524,0.10000,4.485
46 6,self_attn.q_proj,0.0000009372,0.10000,4.513
47 6,self_attn.o_proj,0.0000002816,0.10000,1.569
48 6,mlp.gate_proj,0.0000282582,0.10000,2.038
49 6,mlp.up_proj,0.0000170360,0.10000,2.078
50 6,mlp.down_proj,0.0000238226,0.10000,4.473
51 7,self_attn.k_proj,0.0000005651,0.10000,4.312
52 7,self_attn.v_proj,0.0000006153,0.10000,4.532
53 7,self_attn.q_proj,0.0000021191,0.10000,4.541
54 7,self_attn.o_proj,0.0000003873,0.10000,1.559
55 7,mlp.up_proj,0.0000209582,0.10000,2.027
56 7,mlp.gate_proj,0.0000333655,0.10000,2.044
57 7,mlp.down_proj,0.0000011662,0.10000,4.485
58 8,self_attn.k_proj,0.0000007382,0.10000,4.366
59 8,self_attn.v_proj,0.0000008672,0.10000,4.553
60 8,self_attn.q_proj,0.0000029186,0.10000,4.584
61 8,self_attn.o_proj,0.0000004689,0.10000,1.625
62 8,mlp.gate_proj,0.0000198027,0.10000,2.106
63 8,mlp.up_proj,0.0000163943,0.10000,2.119
64 8,mlp.down_proj,0.0000015403,0.10000,4.599
65 9,self_attn.k_proj,0.0000009926,0.10000,4.345
66 9,self_attn.v_proj,0.0000010791,0.10000,4.597
67 9,self_attn.q_proj,0.0000036375,0.10000,4.625
68 9,self_attn.o_proj,0.0000006776,0.10000,1.550
69 9,mlp.up_proj,0.0000205884,0.10000,2.038
70 9,mlp.gate_proj,0.0000315698,0.10000,2.079
71 9,mlp.down_proj,0.0000016813,0.10000,4.645
72 10,self_attn.k_proj,0.0000013627,0.10000,4.412
73 10,self_attn.q_proj,0.0000053466,0.10000,4.665
74 10,self_attn.v_proj,0.0000016548,0.10000,4.706
75 10,self_attn.o_proj,0.0000007416,0.10000,1.613
76 10,mlp.gate_proj,0.0000238936,0.10000,2.020
77 10,mlp.up_proj,0.0000168923,0.10000,2.067
78 10,mlp.down_proj,0.0000015627,0.10000,4.520
79 11,self_attn.k_proj,0.0000007019,0.10000,4.288
80 11,self_attn.v_proj,0.0000007862,0.10000,4.550
81 11,self_attn.q_proj,0.0000026579,0.10000,4.598
82 11,self_attn.o_proj,0.0000006333,0.10000,1.561
83 11,mlp.up_proj,0.0000156422,0.10000,1.964
84 11,mlp.gate_proj,0.0000196474,0.10000,1.983
85 11,mlp.down_proj,0.0000015437,0.10000,4.595
86 12,self_attn.k_proj,0.0000008867,0.10000,4.297
87 12,self_attn.q_proj,0.0000034566,0.10000,4.574
88 12,self_attn.v_proj,0.0000010242,0.10000,4.616
89 12,self_attn.o_proj,0.0000010202,0.10000,1.577
90 12,mlp.gate_proj,0.0000181967,0.10000,2.092
91 12,mlp.up_proj,0.0000159937,0.10000,2.128
92 12,mlp.down_proj,0.0000016049,0.10000,4.597
93 13,self_attn.k_proj,0.0000006896,0.10000,4.313
94 13,self_attn.q_proj,0.0000026921,0.10000,4.535
95 13,self_attn.v_proj,0.0000007320,0.10000,4.566
96 13,self_attn.o_proj,0.0000008842,0.10000,1.626
97 13,mlp.up_proj,0.0000177848,0.10000,1.996
98 13,mlp.gate_proj,0.0000184155,0.10000,2.023
99 13,mlp.down_proj,0.0000019769,0.10000,4.748
100 14,self_attn.k_proj,0.0000012724,0.10000,4.292
101 14,self_attn.v_proj,0.0000014461,0.10000,4.545
102 14,self_attn.q_proj,0.0000050922,0.10000,4.607
103 14,self_attn.o_proj,0.0000014442,0.10000,1.618
104 14,mlp.gate_proj,0.0000188562,0.10000,2.122
105 14,mlp.up_proj,0.0000183811,0.10000,2.164
106 14,mlp.down_proj,0.0000018007,0.10000,4.510
107 15,self_attn.k_proj,0.0000012884,0.10000,4.345
108 15,self_attn.q_proj,0.0000051236,0.10000,4.570
109 15,self_attn.v_proj,0.0000013641,0.10000,4.614
110 15,self_attn.o_proj,0.0000014086,0.10000,1.631
111 15,mlp.up_proj,0.0000184349,0.10000,2.067
112 15,mlp.gate_proj,0.0000181649,0.10000,2.091
113 15,mlp.down_proj,0.0000018890,0.10000,4.577
114 16,self_attn.k_proj,0.0000021029,0.10000,4.328
115 16,self_attn.v_proj,0.0000025386,0.10000,4.556
116 16,self_attn.q_proj,0.0000086758,0.10000,4.568
117 16,self_attn.o_proj,0.0000012714,0.10000,1.562
118 16,mlp.gate_proj,0.0000200963,0.10000,1.935
119 16,mlp.up_proj,0.0000192835,0.10000,1.973
120 16,mlp.down_proj,0.0000187656,0.10000,4.514
121 17,self_attn.k_proj,0.0000017093,0.10000,4.341
122 17,self_attn.q_proj,0.0000077938,0.10000,4.594
123 17,self_attn.v_proj,0.0000020613,0.10000,4.622
124 17,self_attn.o_proj,0.0000016575,0.10000,1.643
125 17,mlp.up_proj,0.0000181763,0.10000,2.057
126 17,mlp.gate_proj,0.0000186384,0.10000,2.093
127 17,mlp.down_proj,0.0000018738,0.10000,4.600
128 18,self_attn.k_proj,0.0000020693,0.10000,4.363
129 18,self_attn.q_proj,0.0000086855,0.10000,4.631
130 18,self_attn.v_proj,0.0000025088,0.10000,4.649
131 18,self_attn.o_proj,0.0000017786,0.10000,1.606
132 18,mlp.gate_proj,0.0000201907,0.10000,2.099
133 18,mlp.up_proj,0.0000200689,0.10000,2.166
134 18,mlp.down_proj,0.0000023515,0.10000,4.576
135 19,self_attn.k_proj,0.0000038297,0.10000,4.432
136 19,self_attn.v_proj,0.0000044097,0.10000,4.653
137 19,self_attn.q_proj,0.0000169872,0.10000,4.682
138 19,self_attn.o_proj,0.0000028907,0.10000,1.582
139 19,mlp.up_proj,0.0000229501,0.10000,2.107
140 19,mlp.gate_proj,0.0000226941,0.10000,2.159
141 19,mlp.down_proj,0.0000030679,0.10000,4.579
142 20,self_attn.k_proj,0.0000036831,0.10000,4.317
143 20,self_attn.v_proj,0.0000043706,0.10000,4.591
144 20,self_attn.q_proj,0.0000174276,0.10000,4.627
145 20,self_attn.o_proj,0.0000029771,0.10000,1.599
146 20,mlp.gate_proj,0.0000253506,0.10000,2.059
147 20,mlp.up_proj,0.0000262646,0.10000,2.086
148 20,mlp.down_proj,0.0000036408,0.10000,4.510
149 21,self_attn.k_proj,0.0000050001,0.10000,4.363
150 21,self_attn.q_proj,0.0000228820,0.10000,4.653
151 21,self_attn.v_proj,0.0000059135,0.10000,4.667
152 21,self_attn.o_proj,0.0000029943,0.10000,1.538
153 21,mlp.gate_proj,0.0000304431,0.10000,2.065
154 21,mlp.up_proj,0.0000316683,0.10000,2.093
155 21,mlp.down_proj,0.0000045086,0.10000,4.625
156 22,self_attn.k_proj,0.0000092675,0.10000,4.294
157 22,self_attn.q_proj,0.0000424574,0.10000,4.589
158 22,self_attn.v_proj,0.0000114349,0.10000,4.613
159 22,self_attn.o_proj,0.0000045903,0.10000,1.557
160 22,mlp.gate_proj,0.0000354832,0.10000,2.096
161 22,mlp.up_proj,0.0000353631,0.10000,2.145
162 22,mlp.down_proj,0.0000065555,0.10000,4.581
163 23,self_attn.k_proj,0.0000091701,0.10000,4.238
164 23,self_attn.v_proj,0.0000109907,0.10000,4.710
165 23,self_attn.q_proj,0.0000420592,0.10000,4.755
166 23,self_attn.o_proj,0.0000054694,0.10000,1.579
167 23,mlp.up_proj,0.0000414041,0.10000,2.125
168 23,mlp.gate_proj,0.0000440309,0.10000,2.163
169 23,mlp.down_proj,0.0000084183,0.10000,4.566
170 24,self_attn.k_proj,0.0000152363,0.10000,4.191
171 24,self_attn.q_proj,0.0000699393,0.10000,4.506
172 24,self_attn.v_proj,0.0000197641,0.10000,4.548
173 24,self_attn.o_proj,0.0000070709,0.10000,1.614
174 24,mlp.gate_proj,0.0000503237,0.10000,2.141
175 24,mlp.up_proj,0.0000460797,0.10000,2.191
176 24,mlp.down_proj,0.0000104877,0.10000,4.497
177 25,self_attn.k_proj,0.0000114676,0.10000,4.133
178 25,self_attn.q_proj,0.0000481714,0.10000,4.443
179 25,self_attn.v_proj,0.0000137399,0.10000,4.481
180 25,self_attn.o_proj,0.0000035874,0.10000,1.621
181 25,mlp.up_proj,0.0000504789,0.10000,2.020
182 25,mlp.gate_proj,0.0000562083,0.10000,2.065
183 25,mlp.down_proj,0.0000124710,0.10000,4.571
184 26,self_attn.k_proj,0.0000163119,0.10000,4.307
185 26,self_attn.v_proj,0.0000199003,0.10000,4.560
186 26,self_attn.q_proj,0.0000721851,0.10000,4.593
187 26,self_attn.o_proj,0.0000030045,0.10000,1.607
188 26,mlp.gate_proj,0.0000650178,0.10000,1.944
189 26,mlp.up_proj,0.0000600384,0.10000,2.016
190 26,mlp.down_proj,0.0000118864,0.10000,4.514
191 27,self_attn.k_proj,0.0000191612,0.10000,4.194
192 27,self_attn.v_proj,0.0000258756,0.10000,4.484
193 27,self_attn.q_proj,0.0000881676,0.10000,4.499
194 27,self_attn.o_proj,0.0000036633,0.10000,1.582
195 27,mlp.up_proj,0.0000666810,0.10000,2.011
196 27,mlp.gate_proj,0.0000699586,0.10000,2.086
197 27,mlp.down_proj,0.0000171271,0.10000,4.543
198 28,self_attn.k_proj,0.0000253908,0.10000,3.950
199 28,self_attn.v_proj,0.0000305473,0.10000,4.400
200 28,self_attn.q_proj,0.0001051568,0.10000,4.419
201 28,self_attn.o_proj,0.0000075650,0.10000,1.628
202 28,mlp.gate_proj,0.0000770289,0.10000,2.025
203 28,mlp.up_proj,0.0000763184,0.10000,2.118
204 28,mlp.down_proj,0.0000230418,0.10000,4.608
205 29,self_attn.k_proj,0.0000528532,0.10000,4.391
206 29,self_attn.v_proj,0.0000706905,0.10000,4.621
207 29,self_attn.q_proj,0.0002377818,0.10000,4.649
208 29,self_attn.o_proj,0.0000114592,0.10000,1.577
209 29,mlp.up_proj,0.0000926004,0.10000,2.121
210 29,mlp.gate_proj,0.0000892738,0.10000,2.158
211 29,mlp.down_proj,0.0000262264,0.10000,4.575
212 30,self_attn.k_proj,0.0000615581,0.10000,4.263
213 30,self_attn.q_proj,0.0002475710,0.10000,4.521
214 30,self_attn.v_proj,0.0000790300,0.10000,4.546
215 30,self_attn.o_proj,0.0000080313,0.10000,1.577
216 30,mlp.gate_proj,0.0000938009,0.10000,1.948
217 30,mlp.up_proj,0.0001005060,0.10000,1.981
218 30,mlp.down_proj,0.0000340703,0.10000,4.439
219 31,self_attn.k_proj,0.0000925253,0.10000,4.279
220 31,self_attn.q_proj,0.0003612229,0.10000,4.514
221 31,self_attn.v_proj,0.0001289460,0.10000,4.533
222 31,self_attn.o_proj,0.0000124489,0.10000,1.552
223 31,mlp.up_proj,0.0001066470,0.10000,2.058
224 31,mlp.gate_proj,0.0000948500,0.10000,2.078
225 31,mlp.down_proj,0.0000454290,0.10000,4.558
226 32,self_attn.k_proj,0.0001231427,0.10000,4.320
227 32,self_attn.q_proj,0.0005124021,0.10000,4.625
228 32,self_attn.v_proj,0.0001872253,0.10000,4.658
229 32,self_attn.o_proj,0.0000203278,0.10000,1.620
230 32,mlp.gate_proj,0.0001001873,0.10000,1.977
231 32,mlp.up_proj,0.0001154902,0.10000,2.016
232 32,mlp.down_proj,0.0000601530,0.10000,4.641
233 33,self_attn.k_proj,0.0002222856,0.10000,4.128
234 33,self_attn.q_proj,0.0011057818,0.10000,4.438
235 33,self_attn.v_proj,0.0004192334,0.10000,4.453
236 33,self_attn.o_proj,0.0000454634,0.10000,1.567
237 33,mlp.up_proj,0.0001329813,0.10000,2.029
238 33,mlp.gate_proj,0.0001132789,0.10000,2.081
239 33,mlp.down_proj,0.0001011796,0.10000,4.507
240 34,self_attn.k_proj,0.0001888980,0.10000,4.234
241 34,self_attn.v_proj,0.0003253079,0.10000,4.554
242 34,self_attn.q_proj,0.0008556856,0.10000,4.575
243 34,self_attn.o_proj,0.0000674757,0.10000,1.632
244 34,mlp.gate_proj,0.0001431082,0.10000,2.044
245 34,mlp.up_proj,0.0001571695,0.10000,2.083
246 34,mlp.down_proj,0.0001871779,0.10000,4.492
247 35,self_attn.k_proj,0.0000936730,0.10000,4.291
248 35,self_attn.v_proj,0.0001312502,0.10000,4.533
249 35,self_attn.q_proj,0.0003771549,0.10000,4.553
250 35,self_attn.o_proj,0.0000736388,0.10000,1.576
251 35,mlp.up_proj,0.0002436513,0.10000,1.918
252 35,mlp.gate_proj,0.0002332021,0.10000,1.968
253 35,mlp.down_proj,0.0006246001,0.10000,4.539

47
quantize_config.json Normal file
View File

@@ -0,0 +1,47 @@
{
"bits": 4,
"group_size": 128,
"desc_act": false,
"lm_head": false,
"method": "gptq",
"quant_method": "gptq",
"format": "gptq",
"checkpoint_format": "gptq",
"pack_dtype": "int32",
"meta": {
"quantizer": [
"gptqmodel:7.1.0"
],
"uri": "https://github.com/modelcloud/gptqmodel",
"damp_percent": 0.1,
"damp_auto_increment": 0.01,
"static_groups": false,
"true_sequential": true,
"mse": 0.0,
"gptaq": null,
"foem": null,
"act_group_aware": true,
"fallback": {
"strategy": "rtn",
"threshold": "0.5%",
"smooth": null
},
"offload_to_disk": true,
"offload_to_disk_path": "/tmp/gptqmodel_pf205jdd",
"pack_impl": "cpu",
"gc_mode": "interval",
"wait_for_submodule_finalizers": false,
"auto_forward_data_parallel": true,
"dense_vram_strategy": "exclusive",
"dense_vram_strategy_devices": null,
"moe_vram_strategy": "exclusive",
"moe_vram_strategy_devices": null,
"mock_quantization": false,
"hessian": {
"chunk_size": null,
"chunk_bytes": null,
"staging_dtype": "float32"
}
},
"sym": true
}

3
tokenizer.json Normal file
View File

@@ -0,0 +1,3 @@
version https://git-lfs.github.com/spec/v1
oid sha256:be75606093db2094d7cd20f3c2f385c212750648bd6ea4fb2bf507a6a4c55506
size 11422650

16
tokenizer_config.json Normal file
View File

@@ -0,0 +1,16 @@
{
"add_prefix_space": false,
"backend": "tokenizers",
"bos_token": null,
"clean_up_tokenization_spaces": false,
"eos_token": "<|im_end|>",
"errors": "replace",
"is_local": false,
"local_files_only": false,
"model_max_length": 262144,
"pad_token": "<|vision_pad|>",
"padding_side": "left",
"split_special_tokens": false,
"tokenizer_class": "Qwen2Tokenizer",
"unk_token": null
}