Language Generation on Wiki
3.53Perplexity-
Evaluation Results
| Method | Links | |
|---|---|---|
| -#Bits=W16A16, Model=Llama-65B2026.03 | 3.53 | |
| SliderQuant#Bits=W4A16, Model=Llama-65B2026.03 | 3.65 | |
| CBQ#Bits=W4A16, Model=Llama-65B2026.03 | 3.68 | |
| OmniQuant#Bits=W4A16, Model=Llama-65B2026.03 | 3.71 | |
| AWQ#Bits=W4A16, Model=Llama-65B2026.03 | 3.76 | |
| GPTQ#Bits=W4A16, Model=Llama-65B2026.03 | 3.83 | |
| RTN#Bits=W4A16, Model=Llama-65B2026.03 | 3.87 | |
| -#Bits=W16A16, Model=Llama-13B2026.03 | 5.09 | |
| SliderQuant#Bits=W4A16, Model=Llama-13B2026.03 | 5.19 | |
| OmniQuant#Bits=W4A16, Model=Llama-13B2026.03 | 5.21 | |
| CBQ#Bits=W4A16, Model=Llama-13B2026.03 | 5.22 | |
| AWQ#Bits=W4A16, Model=Llama-13B2026.03 | 5.34 | |
| GPTQ#Bits=W4A16, Model=Llama-13B2026.03 | 5.4 | |
| RTN#Bits=W4A16, Model=Llama-13B2026.03 | 5.55 | |
| -#Bits=W16A16, Model=Llama-7B2026.03 | 5.68 | |
| SliderQuant#Bits=W4A16, Model=Llama-7B2026.03 | 5.81 | |
| OmniQuant#Bits=W4A16, Model=Llama-7B2026.03 | 5.86 | |
| CBQ#Bits=W4A16, Model=Llama-7B2026.03 | 5.86 | |
| SliderQuant#Bits=W2A16, Model=Llama-65B2026.03 | 5.95 | |
| AWQ#Bits=W4A16, Model=Llama-7B2026.03 | 6.08 | |
| GPTQ#Bits=W4A16, Model=Llama-7B2026.03 | 6.13 | |
| SliderQuant#Bits=W4A4, Model=Llama-65B2026.03 | 6.2 | |
| RTN#Bits=W4A16, Model=Llama-7B2026.03 | 6.43 | |
| CBQ#Bits=W2A16, Model=Llama-65B2026.03 | 6.56 | |
| SliderQuant#Bits=W4A4, Model=Llama-13B2026.03 | 7.22 | |
| CBQ#Bits=W4A4, Model=Llama-65B2026.03 | 7.23 | |
| SliderQuant#Bits=W2A16, Model=Llama-13B2026.03 | 7.51 | |
| OmniQuant#Bits=W2A16, Model=Llama-65B2026.03 | 7.58 | |
| CBQ#Bits=W2A16, Model=Llama-13B2026.03 | 7.96 | |
| SliderQuant#Bits=W4A4, Model=Llama-7B2026.03 | 8.01 | |
| SliderQuant#Bits=W2A16, Model=Llama-7B2026.03 | 9 | |
| OmniQuant#Bits=W4A4, Model=Llama-65B2026.03 | 9.17 | |
| CBQ#Bits=W2A16, Model=Llama-7B2026.03 | 9.65 | |
| CBQ#Bits=W4A4, Model=Llama-13B2026.03 | 9.69 | |
| CBQ#Bits=W4A4, Model=Llama-7B2026.03 | 10.39 | |
| OmniQuant#Bits=W4A4, Model=Llama-13B2026.03 | 10.87 | |
| OmniQuant#Bits=W4A4, Model=Llama-7B2026.03 | 11.26 | |
| OmniQuant#Bits=W2A16, Model=Llama-13B2026.03 | 13.21 | |
| OmniQuant#Bits=W2A16, Model=Llama-7B2026.03 | 15.47 | |
| SmoothQuant#Bits=W4A4, Model=Llama-7B2026.03 | 25.25 | |
| SmoothQuant#Bits=W4A4, Model=Llama-13B2026.03 | 40.05 | |
| GPTQ#Bits=W2A16, Model=Llama-65B2026.03 | 55.91 | |
| AWQ#Bits=W2A16, Model=Llama-65B2026.03 | 75.43 | |
| RTN#Bits=W4A4, Model=Llama-7B2026.03 | 270 | |
| SmoothQuant#Bits=W4A4, Model=Llama-65B2026.03 | 280 | |
| GPTQ#Bits=W2A16, Model=Llama-7B2026.03 | 2,100 | |
| RTN#Bits=W4A4, Model=Llama-13B2026.03 | 2,400 | |
| GPTQ#Bits=W2A16, Model=Llama-13B2026.03 | 5,500 | |
| RTN#Bits=W2A16, Model=Llama-65B2026.03 | 22,000 | |
| RTN#Bits=W4A4, Model=Llama-65B2026.03 | 37,000 | |
| RTN#Bits=W2A16, Model=Llama-13B2026.03 | 68,000 | |
| RTN#Bits=W2A16, Model=Llama-7B2026.03 | 110,000 | |
| AWQ#Bits=W2A16, Model=Llama-7B2026.03 | 260,000 | |
| AWQ#Bits=W2A16, Model=Llama-13B2026.03 | 280,000 |