Thomaschtl commited on
Commit
4e9354e
·
verified ·
1 Parent(s): 48ccf9f

Upload quantization_info.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. quantization_info.json +25 -0
quantization_info.json ADDED
@@ -0,0 +1,25 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "quantization_method": "awq_4bit",
3
+ "linear_layers": "0/115 (4-bit AWQ)",
4
+ "embedding_layers": "0/0 (not quantized by AWQ)",
5
+ "total_quantized": "0/115",
6
+ "original_model": "Qwen/Qwen3-0.6B-Base",
7
+ "calibration_data_path": "smoothquant/smoothquant/calibration_prompts.json",
8
+ "calibration_samples": 100,
9
+ "quantization_timestamp": "2025-06-10 20:40:48",
10
+ "pytorch_version": "2.6.0+cu118",
11
+ "estimated_compression_ratio": "4.0x",
12
+ "estimated_size_mb": 284.21875,
13
+ "original_size_mb": 1136.875,
14
+ "formats_included": [
15
+ "awq_safetensors"
16
+ ],
17
+ "lighteval_compatible": true,
18
+ "awq_config": {
19
+ "zero_point": true,
20
+ "q_group_size": 128,
21
+ "w_bit": 4,
22
+ "version": "GEMM"
23
+ },
24
+ "notes": "4-bit AWQ quantization with activation-aware weight optimization"
25
+ }