Language Generation on WikiText-2 (test)
3.319PerplexityOPTQ
Evaluation Results
| Method | Links | |
|---|---|---|
| OPTQWBits=162023.07 | 3.319 | |
| QuIPWBits=162023.07 | 3.319 | |
| QuIPWBits=42023.07 | 3.531 | |
| OPTQWBits=42023.07 | 3.596 | |
| QuIPWBits=32023.07 | 3.853 | |
| OPTQWBits=32023.07 | 4.907 | |
| FP16Bits=-, Backbone=LLAMA-2-7B2026.04 | 5.47 | |
| QuIPWBits=22023.07 | 6.326 | |
| BWABits=W(1+1)A4, Backbone=LLAMA-2-7B2026.04 | 8.89 | |
| LBLLMBits=W(1+1)A4, Backbone=LLAMA-2-7B2026.04 | 9.18 | |
| CBQBits=W4A4, Backbone=LLAMA-2-7B2026.04 | 11.32 | |
| ARB-LLMBits=W(1+1)A4, Backbone=LLAMA-2-7B2026.04 | 19.14 | |
| BiLLMBits=W(1+1)A4, Backbone=LLAMA-2-7B2026.04 | 32.48 | |
| QuaRotBits=W2A4, Backbone=LLAMA-2-7B2026.04 | 49.98 | |
| OPTQWBits=22023.07 | 123.908 | |
| BWA w/o CBits=W(1+1)A4, Backbone=LLAMA-2-7B2026.04 | 243.73 |