Spaces:
Runtime error
Runtime error
| ,baichuan-inc/Baichuan-7B,hfl/chinese-alpaca-2-7b,Qwen/Qwen-7B-Chat,Qwen/Qwen1.5-7B-Chat,HuggingFaceH4/zephyr-7b-beta,01-ai/Yi-6B-Chat,BAAI/AquilaChat2-7B-16K | |
| "Overall (lower the better)",17,15,11,10,14,24,Generate nonsense | |
| Ready to use,"2 | |
| Some special tokens returned but easy to clean up","1 | |
| No special tokens in responses","2 | |
| Some special tokens returned but easy to clean up","1 | |
| Robust even without system prompt","1 | |
| No special tokens in responses","3 | |
| Returning many unrelated texts, indicating post-processing requirements",NA | |
| Instruction following - general,"2 | |
| Having problem distinguish the poem types","2 | |
| Having problem distinguish the poem types","1 | |
| Perfectly follows","2 | |
| almost follows except one case with Chinese-only instruction","2 | |
| Perfectly follows if ignoring language requirements on poems","5 | |
| Occasionally not answering questions at all",NA | |
| Instruction Following - language,"3 | |
| Always answers in Chinese","3 | |
| Always answers in Chinese","1 | |
| Perfectly distinguishing output language requirements","2 | |
| almost follows except one case with Chinese-only instruction","2 | |
| Having problem with citing Chinese poems","1 | |
| Perfectly distinguishing output language requirements",NA | |
| Helpfulness and Creativeness,"1 | |
| Answer questions with helpful contexts","1 | |
| Answer questions with helpful contexts","2 | |
| Very concise, sometimes too concise","1 | |
| Answer questions with helpful contexts","1 | |
| Answer questions with helpful contexts","3 | |
| Too verbose",NA | |
| Fact,"2 | |
| Wrong answers for citing poem","2 | |
| Wrong answers for citing poem","1 | |
| No obvious mistakes","1 | |
| No obvious mistakes","2 | |
| Wrong answers for citing poem","3 | |
| Wrong answers for citing poem and country terriory",NA | |
| Reasoning,"2 | |
| Self-consistent in reasoning but factually wrong","2 | |
| Self-consistent in reasoning but factually wrong","2 | |
| Self-consistent in reasoning but factually wrong","1 | |
| Self-consistent and factually correct","2 | |
| Self-consistent in reasoning but factually wrong","2 | |
| Self-consistent in reasoning but factually wrong",NA | |
| Coding,"3 | |
| Valid code but wrong formatting or explanation","3 | |
| Valid code but wrong formatting or explanation","1 | |
| Perfect codes with explanation","1 | |
| Perfect codes with explanation","1 | |
| Perfect codes with explanation","4 | |
| Nonsense",NA | |
| Inference Speed,2,1,1,1,3,3,3 | |