Text Generation on WikiText-2
4.88PerplexityFP16
Evaluation Results
| Method | Links | |
|---|---|---|
| FP16Avg. Bit=16, Backbone=LLaMA2-13B2024.10 | 4.88 | |
| VPTQAvg. Bit=4, Backbone=LLaMA2-13B2024.10 | 4.96 | |
| AWQAvg. Bit=4, Backbone=LLaMA2-13B2024.10 | 4.98 | |
| LLM-MQAvg. Bit=4, Backbone=LLaMA2-13B2024.10 | 4.98 | |
| CMPQAvg. Bit=4, Backbone=LLaMA2-13B2024.10 | 4.98 | |
| QuIP#Avg. Bit=4, Backbone=LLaMA2-13B2024.10 | 5 | |
| GPTQAvg. Bit=4, Backbone=LLaMA2-13B2024.10 | 5.08 | |
| RTNAvg. Bit=4, Backbone=LLaMA2-13B2024.10 | 5.2 | |
| QuIPAvg. Bit=4, Backbone=LLaMA2-13B2024.10 | 5.29 | |
| AWQAvg. Bit=3, Backbone=LLaMA2-13B2024.10 | 5.32 | |
| QuIP#Avg. Bit=3, Backbone=LLaMA2-13B2024.10 | 5.34 | |
| CMPQAvg. Bit=3, Backbone=LLaMA2-13B2024.10 | 5.34 | |
| GPTVQAvg. Bit=3, Backbone=LLaMA2-13B2024.10 | 5.41 | |
| VPTQAvg. Bit=3, Backbone=LLaMA2-13B2024.10 | 5.42 | |
| FP16Avg. Bit=16, Backbone=LLaMA2-7B2024.10 | 5.47 | |
| CMPQAvg. Bit=4, Backbone=LLaMA2-7B2024.10 | 5.61 | |
| VPTQAvg. Bit=4, Backbone=LLaMA2-7B2024.10 | 5.64 | |
| QuIPAvg. Bit=3, Backbone=LLaMA2-13B2024.10 | 5.65 | |
| QuIP#Avg. Bit=4, Backbone=LLaMA2-7B2024.10 | 5.66 | |
| GPTVQAvg. Bit=4, Backbone=LLaMA2-7B2024.10 | 5.68 | |
| GPTVQAvg. Bit=4, Backbone=LLaMA2-13B2024.10 | 5.68 | |
| LLM-MQAvg. Bit=4, Backbone=LLaMA2-7B2024.10 | 5.68 | |
| GPTQAvg. Bit=4, Backbone=LLaMA2-7B2024.10 | 5.72 | |
| AWQAvg. Bit=4, Backbone=LLaMA2-7B2024.10 | 5.72 | |
| QuIPAvg. Bit=4, Backbone=LLaMA2-7B2024.10 | 5.72 | |
| LLM-MQAvg. Bit=3, Backbone=LLaMA2-13B2024.10 | 5.89 | |
| RTNAvg. Bit=4, Backbone=LLaMA2-7B2024.10 | 6.12 | |
| CMPQAvg. Bit=3, Backbone=LLaMA2-7B2024.10 | 6.14 | |
| GPTQAvg. Bit=3, Backbone=LLaMA2-13B2024.10 | 6.17 | |
| VPTQAvg. Bit=3, Backbone=LLaMA2-7B2024.10 | 6.17 | |
| QuIP#Avg. Bit=3, Backbone=LLaMA2-7B2024.10 | 6.19 | |
| GPTVQAvg. Bit=3, Backbone=LLaMA2-7B2024.10 | 6.19 | |
| AWQAvg. Bit=3, Backbone=LLaMA2-7B2024.10 | 6.24 | |
| GPTQAvg. Bit=3, Backbone=LLaMA2-7B2024.10 | 6.25 | |
| QuIPAvg. Bit=3, Backbone=LLaMA2-7B2024.10 | 6.8 | |
| LLM-MQAvg. Bit=3, Backbone=LLaMA2-7B2024.10 | 7.16 | |
| QuIP#Avg. Bit=2, Backbone=LLaMA2-13B2024.10 | 7.6 | |
| CMPQAvg. Bit=2, Backbone=LLaMA2-13B2024.10 | 9.14 | |
| QuIPAvg. Bit=2, Backbone=LLaMA2-13B2024.10 | 10.09 | |
| RTNAvg. Bit=3, Backbone=LLaMA2-13B2024.10 | 10.68 | |
| FP16Avg. Bit=16, Backbone=OPT-6.7B2024.10 | 10.86 | |
| AWQAvg. Bit=4, Backbone=OPT-6.7B2024.10 | 10.93 | |
| CMPQAvg. Bit=4, Backbone=OPT-6.7B2024.10 | 10.95 | |
| DensePruning Ratio=0%, Backbone=Llama2-13B2025.10 | 10.98 | |
| QuIPAvg. Bit=4, Backbone=OPT-6.7B2024.10 | 10.98 | |
| LLM-MQAvg. Bit=4, Backbone=OPT-6.7B2024.10 | 11.04 | |
| AWQAvg. Bit=3, Backbone=OPT-6.7B2024.10 | 11.41 | |
| CMPQAvg. Bit=3, Backbone=OPT-6.7B2024.10 | 11.45 | |
| GPTQAvg. Bit=4, Backbone=OPT-6.7B2024.10 | 11.49 | |
| QuIPAvg. Bit=3, Backbone=OPT-6.7B2024.10 | 11.51 | |
| RTNAvg. Bit=4, Backbone=OPT-6.7B2024.10 | 12.15 | |
| DensePruning Ratio=0%, Backbone=Llama2-7B2025.10 | 12.19 | |
| QuIP#Avg. Bit=2, Backbone=LLaMA2-7B2024.10 | 12.3 | |
| FP16Avg. Bit=16, Backbone=OPT-2.7B2024.10 | 12.47 | |
| CMPQAvg. Bit=4, Backbone=OPT-2.7B2024.10 | 12.63 | |
| QuIPAvg. Bit=4, Backbone=OPT-2.7B2024.10 | 12.69 | |
| AWQAvg. Bit=4, Backbone=OPT-2.7B2024.10 | 12.73 | |
| GPTQAvg. Bit=4, Backbone=OPT-2.7B2024.10 | 12.93 | |
| LLM-MQAvg. Bit=4, Backbone=OPT-2.7B2024.10 | 13.06 | |
| CMPQAvg. Bit=3, Backbone=OPT-2.7B2024.10 | 13.38 | |
| GPTQAvg. Bit=2, Backbone=LLaMA2-13B2024.10 | 13.67 | |
| CMPQAvg. Bit=2, Backbone=LLaMA2-7B2024.10 | 14.37 | |
| Wanda-spPruning Ratio=20%, Backbone=Llama2-13B2025.10 | 14.64 | |
| OWLPruning Ratio=20%, Backbone=Llama2-13B2025.10 | 14.76 | |
| GPTQAvg. Bit=3, Backbone=OPT-6.7B2024.10 | 14.87 | |
| GISPPruning Ratio=20%, Backbone=Llama2-13B2025.10 | 15.1 | |
| LLM-MQAvg. Bit=2, Backbone=LLaMA2-13B2024.10 | 15.69 | |
| LLM-MQAvg. Bit=3, Backbone=OPT-6.7B2024.10 | 16.01 | |
| AWQAvg. Bit=3, Backbone=OPT-2.7B2024.10 | 16.32 | |
| RTNAvg. Bit=4, Backbone=OPT-2.7B2024.10 | 16.69 | |
| FLAPPruning Ratio=20%, Backbone=Llama2-13B2025.10 | 16.95 | |
| GISPPruning Ratio=20%, Backbone=Llama2-7B2025.10 | 17.01 | |
| GPTQAvg. Bit=3, Backbone=OPT-2.7B2024.10 | 17.09 | |
| QuIPAvg. Bit=3, Backbone=OPT-2.7B2024.10 | 17.44 | |
| LLM-MQAvg. Bit=3, Backbone=OPT-2.7B2024.10 | 18.09 | |
| CMPQAvg. Bit=2, Backbone=OPT-6.7B2024.10 | 18.63 | |
| OWLPruning Ratio=30%, Backbone=Llama2-13B2025.10 | 19.02 | |
| GISPPruning Ratio=30%, Backbone=Llama2-13B2025.10 | 19.53 | |
| Wanda-spPruning Ratio=30%, Backbone=Llama2-13B2025.10 | 19.73 | |
| ShortGPTPruning Ratio=20%, Backbone=Llama2-13B2025.10 | 19.95 | |
| LLM-PrunerPruning Ratio=20%, Backbone=Llama2-13B2025.10 | 19.99 | |
| FLAPPruning Ratio=30%, Backbone=Llama2-13B2025.10 | 21.32 | |
| OWLPruning Ratio=20%, Backbone=Llama2-7B2025.10 | 21.8 | |
| QuIPAvg. Bit=2, Backbone=OPT-6.7B2024.10 | 22.33 | |
| Wanda-spPruning Ratio=20%, Backbone=Llama2-7B2025.10 | 22.71 | |
| LLM-PrunerPruning Ratio=20%, Backbone=Llama2-7B2025.10 | 24.25 | |
| GISPPruning Ratio=30%, Backbone=Llama2-7B2025.10 | 24.27 | |
| GISPPruning Ratio=40%, Backbone=Llama2-13B2025.10 | 26.56 | |
| QuIPAvg. Bit=2, Backbone=LLaMA2-7B2024.10 | 27.12 | |
| LLM-PrunerPruning Ratio=30%, Backbone=Llama2-13B2025.10 | 28.47 | |
| FLAPPruning Ratio=20%, Backbone=Llama2-7B2025.10 | 29.19 | |
| OWLPruning Ratio=40%, Backbone=Llama2-13B2025.10 | 31.13 | |
| CMPQAvg. Bit=2, Backbone=OPT-2.7B2024.10 | 32.46 | |
| Wanda-spPruning Ratio=40%, Backbone=Llama2-13B2025.10 | 32.91 | |
| GISPPruning Ratio=40%, Backbone=Llama2-7B2025.10 | 34.54 | |
| OWLPruning Ratio=30%, Backbone=Llama2-7B2025.10 | 34.64 | |
| Wanda-spPruning Ratio=30%, Backbone=Llama2-7B2025.10 | 35.43 | |
| GPTQAvg. Bit=2, Backbone=LLaMA2-7B2024.10 | 36.77 | |
| FLAPPruning Ratio=40%, Backbone=Llama2-13B2025.10 | 37.76 | |
| LLM-PrunerPruning Ratio=30%, Backbone=Llama2-7B2025.10 | 41.24 |