Model Compression on Qwen3-0.6B
7.03Compression RatioEDGERAZOR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| EDGERAZORGroup Sizes=256, Quantized Layers (Decoder)=Yes, Quantized Layers (Emb)=Yes, Quantized Layers (Lm_head)=Yes, Bit-Widths=1.582026.04 | 7.03 | 99.99 | |
| EDGERAZORGroup Sizes=256, Quantized Layers (Decoder)=Yes, Quantized Layers (Emb)=Yes, Quantized Layers (Lm_head)=Yes, Bit-Widths=1.882026.04 | 6.4 | 99.99 | |
| EDGERAZORGroup Sizes=256, Quantized Layers (Decoder)=Yes, Quantized Layers (Emb)=Yes, Quantized Layers (Lm_head)=Yes, Bit-Widths=2.792026.04 | 5.05 | 99.99 | |
| EDGERAZORGroup Sizes=256, Quantized Layers (Decoder)=Yes, Quantized Layers (Emb)=Yes, Quantized Layers (Lm_head)=Yes, Bit-Widths=42026.04 | 3.94 | 99.99 | |
| Other MethodsGroup Sizes=channel, Quantized Layers (Decoder)=Yes, Quantized Layers (Emb)=No, Quantized Layers (Lm_head)=No, Bit-Widths=1.582026.04 | 2.99 | 73.89 | |
| Other MethodsGroup Sizes=128, Quantized Layers (Decoder)=Yes, Quantized Layers (Emb)=No, Quantized Layers (Lm_head)=No, Bit-Widths=1.582026.04 | 2.94 | 73.89 | |
| Other MethodsGroup Sizes=channel, Quantized Layers (Decoder)=Yes, Quantized Layers (Emb)=No, Quantized Layers (Lm_head)=No, Bit-Widths=22026.04 | 2.83 | 73.89 | |
| Other MethodsGroup Sizes=128, Quantized Layers (Decoder)=Yes, Quantized Layers (Emb)=No, Quantized Layers (Lm_head)=No, Bit-Widths=22026.04 | 2.78 | 73.89 | |
| Other MethodsGroup Sizes=channel, Quantized Layers (Decoder)=Yes, Quantized Layers (Emb)=No, Quantized Layers (Lm_head)=No, Bit-Widths=32026.04 | 2.5 | 73.89 | |
| Other MethodsGroup Sizes=128, Quantized Layers (Decoder)=Yes, Quantized Layers (Emb)=No, Quantized Layers (Lm_head)=No, Bit-Widths=32026.04 | 2.47 | 73.89 | |
| Other MethodsGroup Sizes=channel, Quantized Layers (Decoder)=Yes, Quantized Layers (Emb)=No, Quantized Layers (Lm_head)=No, Bit-Widths=42026.04 | 2.24 | 73.89 | |
| Other MethodsGroup Sizes=128, Quantized Layers (Decoder)=Yes, Quantized Layers (Emb)=No, Quantized Layers (Lm_head)=No, Bit-Widths=42026.04 | 2.21 | 73.89 | |
| Qwen3-0.6BBit-Widths=162026.04 | 1 | — |