| 第150行: |
第150行: |
| | |} | | |} |
| | | | |
| | + | == RWKV 的评测数据 == |
| | + | 以下是 RWKV 和其他模型的 Uncheatable Eval 评分对比: |
| | + | |
| | + | === 14B 参数模型 === |
| | + | {| class="wikitable" |
| | + | !Name |
| | + | !Params (B) |
| | + | !Average (lower=better) |
| | + | !ao3 english |
| | + | !bbc news |
| | + | !wikipedia english |
| | + | !arxiv computer science |
| | + | !arxiv physics |
| | + | !github cpp |
| | + | !github python |
| | + | |- |
| | + | |Mistral-Nemo-Base-2407 |
| | + | |12.248 |
| | + | |7.107 |
| | + | |10.07 |
| | + | |8.081 |
| | + | |7.954 |
| | + | |7.419 |
| | + | |7.656 |
| | + | |4.203 |
| | + | |4.368 |
| | + | |- |
| | + | |RWKV-6-14B-v2.1 |
| | + | |14.069 |
| | + | |7.609 |
| | + | |10.188 |
| | + | |8.518 |
| | + | |8.343 |
| | + | |7.916 |
| | + | |8.04 |
| | + | |4.93 |
| | + | |5.33 |
| | + | |- |
| | + | |Llama-2-13b-hf |
| | + | |13.016 |
| | + | |7.676 |
| | + | |10.524 |
| | + | |8.279 |
| | + | |8.187 |
| | + | |8.075 |
| | + | |8.311 |
| | + | |4.929 |
| | + | |5.426 |
| | + | |- |
| | + | |Qwen1.5-14B |
| | + | |14.167 |
| | + | |7.697 |
| | + | |10.88 |
| | + | |8.884 |
| | + | |9.102 |
| | + | |7.752 |
| | + | |7.862 |
| | + | |4.665 |
| | + | |4.736 |
| | + | |- |
| | + | |pythia-12b-v0 |
| | + | |11.846 |
| | + | |8.356 |
| | + | |11.285 |
| | + | |9.19 |
| | + | |9.527 |
| | + | |8.535 |
| | + | |8.398 |
| | + | |5.43 |
| | + | |6.125 |
| | + | |} |
| | + | |
| | + | === 7B 参数模型 === |
| | + | {| class="wikitable" |
| | + | !Name |
| | + | !Params (B) |
| | + | !Average (lower=better) |
| | + | !ao3 english |
| | + | !bbc news |
| | + | !wikipedia english |
| | + | !arxiv computer science |
| | + | !arxiv physics |
| | + | !github cpp |
| | + | !github python |
| | + | |- |
| | + | |Meta-Llama-3.1-8B |
| | + | |8.030 |
| | + | |7.227 |
| | + | |10.529 |
| | + | |8.201 |
| | + | |7.932 |
| | + | |7.549 |
| | + | |7.686 |
| | + | |4.204 |
| | + | |4.487 |
| | + | |- |
| | + | |Qwen2.5-7B |
| | + | |7.616 |
| | + | |7.453 |
| | + | |10.931 |
| | + | |8.742 |
| | + | |8.815 |
| | + | |7.683 |
| | + | |7.991 |
| | + | |3.920 |
| | + | |4.091 |
| | + | |- |
| | + | |Qwen2-7B |
| | + | |7.616 |
| | + | |7.528 |
| | + | |10.814 |
| | + | |8.582 |
| | + | |8.709 |
| | + | |7.822 |
| | + | |8.249 |
| | + | |4.204 |
| | + | |4.317 |
| | + | |- |
| | + | |Mistral-7B-v0.1 |
| | + | |7.242 |
| | + | |7.580 |
| | + | |10.546 |
| | + | |8.282 |
| | + | |8.435 |
| | + | |7.852 |
| | + | |8.092 |
| | + | |4.800 |
| | + | |5.051 |
| | + | |- |
| | + | |RWKV-6-World-7B-v2.1 |
| | + | |7.636 |
| | + | |7.817 |
| | + | |10.410 |
| | + | |8.740 |
| | + | |8.577 |
| | + | |8.107 |
| | + | |8.248 |
| | + | |5.122 |
| | + | |5.516 |
| | + | |- |
| | + | |Yi-1.5-6B |
| | + | |6.061 |
| | + | |7.830 |
| | + | |10.926 |
| | + | |8.789 |
| | + | |8.948 |
| | + | |8.098 |
| | + | |8.410 |
| | + | |4.746 |
| | + | |4.890 |
| | + | |- |
| | + | |OLMo-1.7-7B-hf |
| | + | |6.888 |
| | + | |7.881 |
| | + | |11.012 |
| | + | |8.647 |
| | + | |8.987 |
| | + | |8.004 |
| | + | |8.200 |
| | + | |4.966 |
| | + | |5.354 |
| | + | |- |
| | + | |RWKV-5-World-7B-v2 |
| | + | |7.518 |
| | + | |7.905 |
| | + | |10.495 |
| | + | |8.878 |
| | + | |8.694 |
| | + | |8.184 |
| | + | |8.311 |
| | + | |5.187 |
| | + | |5.587 |
| | + | |- |
| | + | |Qwen1.5-7B |
| | + | |7.721 |
| | + | |7.916 |
| | + | |11.100 |
| | + | |9.126 |
| | + | |9.357 |
| | + | |7.953 |
| | + | |8.107 |
| | + | |4.845 |
| | + | |4.924 |
| | + | |- |
| | + | |mpt-7b |
| | + | |6.649 |
| | + | |7.950 |
| | + | |11.185 |
| | + | |8.682 |
| | + | |8.771 |
| | + | |8.157 |
| | + | |8.438 |
| | + | |4.954 |
| | + | |5.466 |
| | + | |- |
| | + | |Llama-2-7b-hf |
| | + | |6.738 |
| | + | |7.967 |
| | + | |10.842 |
| | + | |8.511 |
| | + | |8.521 |
| | + | |8.330 |
| | + | |8.630 |
| | + | |5.235 |
| | + | |5.703 |
| | + | |- |
| | + | |Zamba-7B-v1 |
| | + | |7.232 |
| | + | |8.089 |
| | + | |10.846 |
| | + | |8.520 |
| | + | |8.639 |
| | + | |8.060 |
| | + | |8.290 |
| | + | |5.938 |
| | + | |6.328 |
| | + | |- |
| | + | |open_llama_7b_v2 |
| | + | |6.738 |
| | + | |8.103 |
| | + | |11.086 |
| | + | |8.839 |
| | + | |9.053 |
| | + | |8.404 |
| | + | |8.757 |
| | + | |4.887 |
| | + | |5.698 |
| | + | |- |
| | + | |falcon-7b |
| | + | |6.922 |
| | + | |8.302 |
| | + | |10.756 |
| | + | |8.691 |
| | + | |9.147 |
| | + | |8.549 |
| | + | |9.060 |
| | + | |5.763 |
| | + | |6.149 |
| | + | |- |
| | + | |pythia-6.9b-v0 |
| | + | |6.857 |
| | + | |8.545 |
| | + | |11.495 |
| | + | |9.375 |
| | + | |9.757 |
| | + | |8.683 |
| | + | |8.572 |
| | + | |5.611 |
| | + | |6.321 |
| | + | |- |
| | + | |mamba-7b-rw |
| | + | |6.947 |
| | + | |9.782 |
| | + | |10.808 |
| | + | |8.548 |
| | + | |8.990 |
| | + | |8.613 |
| | + | |9.142 |
| | + | |11.107 |
| | + | |11.268 |
| | + | |} |
| | + | |
| | + | === 3B 参数模型 === |
| | + | {| class="wikitable" |
| | + | !Name |
| | + | !Params (B) |
| | + | !Average (lower=better) |
| | + | !ao3 english |
| | + | !bbc news |
| | + | !wikipedia english |
| | + | !arxiv computer science |
| | + | !arxiv physics |
| | + | !github cpp |
| | + | !github python |
| | + | |- |
| | + | |Llama-3.2-3B |
| | + | |3.213 |
| | + | |7.573 |
| | + | |10.897 |
| | + | |8.702 |
| | + | |8.283 |
| | + | |7.784 |
| | + | |8.101 |
| | + | |4.590 |
| | + | |4.654 |
| | + | |- |
| | + | |Qwen2.5-3B |
| | + | |3.086 |
| | + | |7.663 |
| | + | |11.226 |
| | + | |9.154 |
| | + | |8.959 |
| | + | |7.787 |
| | + | |8.246 |
| | + | |4.119 |
| | + | |4.151 |
| | + | |- |
| | + | |RWKV-x070-World-2.9B-v3-20250211-ctx4096 |
| | + | |2.948 |
| | + | |7.737 |
| | + | |10.481 |
| | + | |8.919 |
| | + | |8.470 |
| | + | |7.904 |
| | + | |8.337 |
| | + | |4.882 |
| | + | |5.164 |
| | + | |- |
| | + | |stablelm-3b-4e1t |
| | + | |2.795 |
| | + | |7.856 |
| | + | |10.887 |
| | + | |8.818 |
| | + | |8.509 |
| | + | |8.153 |
| | + | |8.500 |
| | + | |4.847 |
| | + | |5.281 |
| | + | |- |
| | + | |recurrentgemma-2b |
| | + | |2.683 |
| | + | |7.987 |
| | + | |11.302 |
| | + | |8.938 |
| | + | |8.883 |
| | + | |8.237 |
| | + | |8.520 |
| | + | |4.802 |
| | + | |5.224 |
| | + | |- |
| | + | |RWKV-x060-World-3B-v2.1-20240417-ctx4096 |
| | + | |3.100 |
| | + | |8.078 |
| | + | |10.674 |
| | + | |9.174 |
| | + | |8.816 |
| | + | |8.268 |
| | + | |8.576 |
| | + | |5.385 |
| | + | |5.655 |
| | + | |- |
| | + | |gemma-2-2b |
| | + | |2.614 |
| | + | |8.122 |
| | + | |11.352 |
| | + | |8.898 |
| | + | |9.031 |
| | + | |8.385 |
| | + | |8.814 |
| | + | |5.012 |
| | + | |5.364 |
| | + | |- |
| | + | |mamba2attn-2.7b |
| | + | |2.698 |
| | + | |8.183 |
| | + | |11.126 |
| | + | |9.279 |
| | + | |9.263 |
| | + | |8.325 |
| | + | |8.287 |
| | + | |5.223 |
| | + | |5.780 |
| | + | |- |
| | + | |RWKV-5-World-3B-v2-20231113-ctx4096 |
| | + | |3.063 |
| | + | |8.227 |
| | + | |10.831 |
| | + | |9.356 |
| | + | |9.004 |
| | + | |8.415 |
| | + | |8.698 |
| | + | |5.509 |
| | + | |5.776 |
| | + | |- |
| | + | |mamba2-2.7b |
| | + | |2.703 |
| | + | |8.292 |
| | + | |11.213 |
| | + | |9.370 |
| | + | |9.384 |
| | + | |8.427 |
| | + | |8.371 |
| | + | |5.343 |
| | + | |5.933 |
| | + | |- |
| | + | |Zamba2-2.7B |
| | + | |2.662 |
| | + | |8.317 |
| | + | |10.970 |
| | + | |8.948 |
| | + | |8.743 |
| | + | |8.169 |
| | + | |8.699 |
| | + | |6.393 |
| | + | |6.295 |
| | + | |- |
| | + | |mamba-2.8b-hf |
| | + | |2.768 |
| | + | |8.414 |
| | + | |11.311 |
| | + | |9.486 |
| | + | |9.530 |
| | + | |8.566 |
| | + | |8.517 |
| | + | |5.461 |
| | + | |6.027 |
| | + | |- |
| | + | |RWKV-4-World-3B-v1-20230619-ctx4096 |
| | + | |3.063 |
| | + | |8.525 |
| | + | |10.898 |
| | + | |9.565 |
| | + | |9.301 |
| | + | |8.902 |
| | + | |9.274 |
| | + | |5.671 |
| | + | |6.065 |
| | + | |- |
| | + | |pythia-2.8b-v0 |
| | + | |2.775 |
| | + | |8.667 |
| | + | |11.664 |
| | + | |9.741 |
| | + | |9.815 |
| | + | |8.724 |
| | + | |8.729 |
| | + | |5.709 |
| | + | |6.288 |
| | + | |- |
| | + | |granite-3.2-2b-instruct |
| | + | |2.534 |
| | + | |8.800 |
| | + | |11.961 |
| | + | |10.119 |
| | + | |10.309 |
| | + | |8.997 |
| | + | |9.254 |
| | + | |5.502 |
| | + | |5.460 |
| | + | |- |
| | + | |granite-3.1-2b-instruct |
| | + | |2.534 |
| | + | |8.837 |
| | + | |12.004 |
| | + | |10.169 |
| | + | |10.361 |
| | + | |9.043 |
| | + | |9.278 |
| | + | |5.519 |
| | + | |5.484 |
| | + | |} |
| | + | |
| | + | === 1.5B 参数模型 === |
| | + | {| class="wikitable" |
| | + | !Name |
| | + | !Params (B) |
| | + | !Average (lower=better) |
| | + | !ao3 english |
| | + | !bbc news |
| | + | !wikipedia english |
| | + | !arxiv computer science |
| | + | !arxiv physics |
| | + | !github cpp |
| | + | !github python |
| | + | |- |
| | + | |Qwen3-1.7B-Base |
| | + | |1.721 |
| | + | |7.915 |
| | + | |11.644 |
| | + | |9.761 |
| | + | |9.415 |
| | + | |7.816 |
| | + | |8.377 |
| | + | |4.177 |
| | + | |4.214 |
| | + | |- |
| | + | |rwkv7-g1-1.5b-20250429-ctx4096 |
| | + | |1.528 |
| | + | |7.975 |
| | + | |10.622 |
| | + | |9.269 |
| | + | |8.936 |
| | + | |8.069 |
| | + | |8.649 |
| | + | |4.973 |
| | + | |5.308 |
| | + | |- |
| | + | |Qwen2.5-1.5B |
| | + | |1.544 |
| | + | |8.058 |
| | + | |11.756 |
| | + | |9.576 |
| | + | |9.489 |
| | + | |8.118 |
| | + | |8.653 |
| | + | |4.396 |
| | + | |4.421 |
| | + | |- |
| | + | |RWKV-x070-World-1.5B-v3-20250127-ctx4096 |
| | + | |1.528 |
| | + | |8.160 |
| | + | |10.932 |
| | + | |9.337 |
| | + | |8.972 |
| | + | |8.250 |
| | + | |8.770 |
| | + | |5.288 |
| | + | |5.571 |
| | + | |- |
| | + | |Llama-3.2-1B |
| | + | |1.236 |
| | + | |8.225 |
| | + | |11.693 |
| | + | |9.343 |
| | + | |9.066 |
| | + | |8.365 |
| | + | |8.762 |
| | + | |5.160 |
| | + | |5.184 |
| | + | |- |
| | + | |SmolLM2-1.7B |
| | + | |1.711 |
| | + | |8.227 |
| | + | |11.204 |
| | + | |9.396 |
| | + | |9.459 |
| | + | |8.382 |
| | + | |9.041 |
| | + | |4.937 |
| | + | |5.172 |
| | + | |- |
| | + | |Index-1.9B |
| | + | |2.173 |
| | + | |8.301 |
| | + | |11.490 |
| | + | |9.510 |
| | + | |9.227 |
| | + | |8.344 |
| | + | |8.589 |
| | + | |5.292 |
| | + | |5.652 |
| | + | |- |
| | + | |stablelm-2-1_6b |
| | + | |1.645 |
| | + | |8.338 |
| | + | |11.423 |
| | + | |9.243 |
| | + | |9.057 |
| | + | |8.579 |
| | + | |9.076 |
| | + | |5.450 |
| | + | |5.535 |
| | + | |- |
| | + | |RWKV-x060-World-1B6-v2.1-20240328-ctx4096 |
| | + | |1.600 |
| | + | |8.491 |
| | + | |11.094 |
| | + | |9.569 |
| | + | |9.302 |
| | + | |8.619 |
| | + | |8.997 |
| | + | |5.798 |
| | + | |6.061 |
| | + | |- |
| | + | |RWKV-5-World-1B5-v2-20231025-ctx4096 |
| | + | |1.578 |
| | + | |8.640 |
| | + | |11.247 |
| | + | |9.746 |
| | + | |9.499 |
| | + | |8.765 |
| | + | |9.106 |
| | + | |5.918 |
| | + | |6.196 |
| | + | |- |
| | + | |mamba2-1.3b |
| | + | |1.344 |
| | + | |8.676 |
| | + | |11.625 |
| | + | |9.743 |
| | + | |9.863 |
| | + | |8.741 |
| | + | |8.742 |
| | + | |5.706 |
| | + | |6.315 |
| | + | |- |
| | + | |MobileLLM-1.5B |
| | + | |1.562 |
| | + | |8.733 |
| | + | |11.588 |
| | + | |9.145 |
| | + | |9.215 |
| | + | |8.816 |
| | + | |9.292 |
| | + | |6.293 |
| | + | |6.785 |
| | + | |- |
| | + | |mamba-1.4b-hf |
| | + | |1.372 |
| | + | |8.783 |
| | + | |11.703 |
| | + | |9.828 |
| | + | |9.973 |
| | + | |8.876 |
| | + | |8.863 |
| | + | |5.808 |
| | + | |6.429 |
| | + | |- |
| | + | |Zamba2-1.2B |
| | + | |1.215 |
| | + | |8.830 |
| | + | |11.392 |
| | + | |9.382 |
| | + | |9.263 |
| | + | |8.571 |
| | + | |9.209 |
| | + | |7.082 |
| | + | |6.914 |
| | + | |- |
| | + | |SmolLM-1.7B |
| | + | |1.711 |
| | + | |8.877 |
| | + | |12.684 |
| | + | |9.847 |
| | + | |9.890 |
| | + | |8.384 |
| | + | |9.024 |
| | + | |6.552 |
| | + | |5.757 |
| | + | |- |
| | + | |gemma-3-1b-pt |
| | + | |1.000 |
| | + | |8.903 |
| | + | |12.188 |
| | + | |9.364 |
| | + | |9.597 |
| | + | |8.874 |
| | + | |9.559 |
| | + | |6.591 |
| | + | |6.150 |
| | + | |- |
| | + | |MobileLLM-1B |
| | + | |1.005 |
| | + | |8.972 |
| | + | |11.863 |
| | + | |9.349 |
| | + | |9.432 |
| | + | |9.028 |
| | + | |9.566 |
| | + | |6.531 |
| | + | |7.033 |
| | + | |- |
| | + | |RWKV-4-World-1.5B-v1-fixed-20230612-ctx4096 |
| | + | |1.578 |
| | + | |8.999 |
| | + | |11.334 |
| | + | |10.000 |
| | + | |9.821 |
| | + | |9.344 |
| | + | |9.797 |
| | + | |6.158 |
| | + | |6.539 |
| | + | |- |
| | + | |Hymba-1.5B-Base |
| | + | |1.523 |
| | + | |9.125 |
| | + | |11.797 |
| | + | |9.446 |
| | + | |9.502 |
| | + | |8.497 |
| | + | |9.375 |
| | + | |7.714 |
| | + | |7.542 |
| | + | |- |
| | + | |pythia-1.4b-v0 |
| | + | |1.415 |
| | + | |9.154 |
| | + | |12.192 |
| | + | |10.200 |
| | + | |10.429 |
| | + | |9.122 |
| | + | |9.198 |
| | + | |6.149 |
| | + | |6.788 |
| | + | |- |
| | + | |Falcon3-1B-Base |
| | + | |1.669 |
| | + | |9.445 |
| | + | |13.039 |
| | + | |10.448 |
| | + | |10.750 |
| | + | |8.604 |
| | + | |9.200 |
| | + | |7.155 |
| | + | |6.919 |
| | + | |} |
| | ---- | | ---- |
| | == RWKV 与其他模型对比 == | | == RWKV 与其他模型对比 == |