Language Modeling on C4 v1 (test)
6.63PerplexityLlama 2-7B
Evaluation Results
| Method | Links | |
|---|---|---|
| Llama 2-7BRatio=-, Avg bits=322025.11 | 6.63 | |
| QTIPRatio=∼ 8×, Avg bits=4.01, Weight Updation=Yes2025.11 | 6.69 | |
| AQLMRatio=∼ 8×, Avg bits=4.04, Weight Updation=Yes2025.11 | 6.75 | |
| QuIP#Ratio=∼ 8×, Avg bits=4.02, Weight Updation=Yes2025.11 | 6.86 | |
| PocketLLMRatio=∼ 8×, Avg bits=3.98, Weight Updation=Yes2025.11 | 6.86 | |
| SpQRRatio=∼ 8×, Avg bits=3.98, Weight Updation=Yes2025.11 | 6.87 | |
| VPTQRatio=∼ 8×, Avg bits=4.01, Weight Updation=Yes2025.11 | 7.13 | |
| GPTQRatio=∼ 8×, Avg bits=4.00, Weight Updation=Yes2025.11 | 7.2 | |
| PocketLLMRatio=∼ 8×, Avg bits=3.98, Weight Updation=No2025.11 | 7.49 | |
| DuQuantRatio=∼ 8×, Avg bits=4.00, Weight Updation=Yes2025.11 | 7.79 | |
| DuQuant*Ratio=∼ 8×, Avg bits=4.01, Weight Updation=No2025.11 | 7.9 |