Can't get connected to my Ollama
Closed this issue · 7 comments
Describe the bug
Can't detect Ollama models.
To Reproduce
Steps to reproduce the behavior:
- brew services start ollama
- open VSCode
- Click twinny
- Select active model
Expected behavior
A drag up menu appears
Screenshots
Desktop (please complete the following information):
- OS: macOS Sonoma
- Browser Chrome
- Version 122 for Chrome, 1.87 for VSCode
Additional context
No.
Please could you provide more information to your problem? Many thanks.
Sure.
- I installed Ollama and twinny, it worked then, because I saw a pullable menu for the model.
- However, after I updated Ollama this week, twinny somehow couldn't detect my Ollama that was running backgroud.
- I am sure Ollama works well, as I can see
❯ curl http://localhost:11434/api/tags
{"models":[{"name":"gemma:7b","model":"gemma:7b","modified_at":"2024-03-03T11:21:28.356683666+08:00","size":5202317200,"digest":"430ed3535049f562814f612a653457ba9fd390cccc77a94ab88964b8d8fd18a8","details":{"parent_model":"","format":"gguf","family":"gemma","families":["gemma"],"parameter_size":"9B","quantization_level":"Q4_0"}}]}%
- The environment variable is set by
launchctl setenv OLLAMA_ORIGINS "http://localhost:*"
So, I guess the requests from twinny have been rejected. Is there any way I can trace twinny's activity by a detailed log to check what happened?
Plus, my VSCode settings for twinny is
{
"twinny.enableCompletionCache": true,
"twinny.fimTemplateFormat": "deepseek",
"twinny.useFileContext": true,
"twinny.useMultiLineCompletions": true,
"twinny.useTls": true,
}
You are using useTls
but running on http://localhost:11434/
? What happens when you set the useTls
to false?
I set it to false, and restarted VSCode, then it works now! Thank you so much @AntonKrug ! You can close it issue as complete!
Thanks @AntonKrug for helping.
Not using TLS and I seem to be having similar issue, not sure how to debug it. I just get spinning wheel forever and no memory allocation. I have no idea what is going on everything is set right even the following works
curl http://localhost:11434/api/generate -d '{
"model": "deepseek-coder:1.3b",
"prompt": "Why is the sky blue?"
}'
Please try to update Ollama and twinny to latest version @nonetrix