search

EmergenceAI / Agent-E

Agent driven automation starting with the web. Discord: https://discord.gg/wgNfmFuqJF
MIT License
702 stars 88 forks source link

能运行但是没反应[BUG] #76

Open luaonze opened 3 months ago

luaonze commented 3 months ago

It works but it doesn't respond and my interface is different from others

截屏2024-07-06 00 23 41

After I run llama3 with ollama locally, I open the page like the screenshot, which is completely different from others. I don't know what to do. I have found many request methods, but they are all wrong.截屏2024-07-06 00 25 36 There is nothing wrong with my running command. I don't know where the problem is. This problem has troubled me for 3 days. Please help me. I give you my most sincere thanks.

teaxio commented 3 months ago

@luaonze we have noticed that when using ollama that all the details related to function calling are gone. Meaning ollama drops them before they reach the intended LLM. There is a section in our readme that shows how to setup ollama correctly. Have you tried it? Another point is that the LLM you use must have function calling capability as that's underlying this. However, even with that there is limited success I suspect due to the verbosity of the planner. We are in need for community members to help us make this work with local LLMs. We have not had time to free up cycles for it :( Please let me know about the above question

teaxio commented 3 months ago

Once you do set this up successfully, I suspect that you will run into this #75 As indicated in the other bug, I welcome ideas on how to defeat this.

luaonze commented 3 months ago

I tried ollama, and my interface is the same as others, especially the robot icon is completely different. I asked questions locally, but there seems to be no response.

And I saw you sent me the solution, but I asked him to operate the model but he didn't reply to me. He didn't respond at all. I uninstalled and reinstalled many times but it still didn't work.

teaxio commented 3 months ago

So you have a problem with Ollama setup then @luaonze ? If that's the case, there is not much I can do

luaonze commented 3 months ago

I think it might be, but I deployed it on several computers and got the same result.

luaonze commented 3 months ago

I think my model is not running. The llama3 model in ollama in docker that I use may not be running.

liuyanmei22 commented 2 months ago

@luaonze 你好，请问这个问题后面解决了吗？

bolt163 commented 2 months ago

这个Agent不支持开源类未实现函数调用的llm... This project builds on AutoGen agent framework which relies on the LLM being used must support function calling. https://github.com/EmergenceAI/Agent-E/issues/28