THUDM / AgentBench

A Comprehensive Benchmark to Evaluate LLMs as Agents (ICLR'24)
https://llmbench.ai
Apache License 2.0
2.01k stars 136 forks source link

[Bug/Assistance] 测试kg-std任务时,输出文件中全部状态都是task limit reached #115

Open 13416157913 opened 5 months ago

13416157913 commented 5 months ago

{"index": 145, "error": null, "info": null, "output": {"index": 145, "status": "task limit reached", "result": {"predict": [], "actions": []}, "history":

zhc7 commented 5 months ago

Hi, @13416157913 这可能是因为模型不够强或者模型配置错误,建议检查一下输出