Giters

darrenburns / elia

A snappy, keyboard-centric terminal user interface for interacting with large language models. Chat with ChatGPT, Claude, Llama 3, Phi 3, Mistral, Gemma and more.

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Same answer after a while

LFd3v opened this issue · comments

commented

Hi, there!

It seems that after a while I start to get the same response over and over again, even if I ask use another prompt. It happened with 3 different chats while I was testing using the some model.

Settings:

  • Linux (ArchLinux), zsh
  • elia installed via pipx
  • Using ollama
  • Model phi3
  • In the modelfile, it has "num_keep=4" which means it should keep the last 4 messages in context.

I will try with other models later. Please let me know if you need more info.

Edit: I have just done some tests with another model that does not have "num_keep" in its modelfile (stablelm2:zephyr) and it seems that it has the same problem: after a while it keeps repeating info about previous inferences.

Thanks again. Regards