zed icon indicating copy to clipboard operation
zed copied to clipboard
verified publisher icon zed-industries

Locally run codellama hangs at the generating step

Open unparadise opened this issue 8 months ago • 1 comments

Summary

Locally run codellama hangs at the generating step while deepseek r1 runs successfully.

Description

Steps to reproduce:

  1. Setup Ollama and run codellama
  2. Configure Zed agent to use Ollama and start a new thread
  3. Add the currently opened file as context and prompt codellama

Expected Behavior: returns result Actual Behavior: it hangs at the generating... step

Zed Version and System Specs

Zed: v0.182.11 (Zed) OS: macOS 13.7.5 Memory: 16 GiB Architecture: x86_64

unparadise avatar Apr 19 '25 23:04 unparadise

Same for me using the qwen3 models, if by "hangs" you mean it takes a while and then outputs everything at once. Feels like it takes about the same amount of time as when asking the query directly with ollama.

maun avatar May 14 '25 12:05 maun

I was not able to reproduce. I pulled codellama with: ollama pull codellama Then connected the Zed assistant and it worked as expected.

You're running this on Intel macOS hardware, with only 16GB of ram on a macOS release two versions old. Does this work via the normally ollama cli?

Closing as unactionable.

notpeter avatar May 20 '25 14:05 notpeter