Skip to content

moved to llama3.2 and moondream for local compute - #54

Open
meltyli wants to merge 1 commit into
iyaja:mainfrom
meltyli:localllamafs
Open

moved to llama3.2 and moondream for local compute#54
meltyli wants to merge 1 commit into
iyaja:mainfrom
meltyli:localllamafs

Conversation

@meltyli

@meltyli meltyli commented Oct 1, 2024

Copy link
Copy Markdown

removed groq and now completely relies on ollama for chat

Comment thread src/loader.py
Comment thread src/loader.py
Comment thread src/loader.py
Comment thread src/loader.py
Comment on lines +281 to +290
client = ollama.Client()
chat_completion = client.chat(
messages = [
{
"role": "system",
"content": PROMPT},
{
"role": "user",
"content": json.dumps(doc),
},

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The Ollama SDK might not be the greatest tool to use. I'd strongly recommend using LiteLLM as the replacement and then setting the model to ollama/<model>

https://github.com/BerriAI/litellm
https://docs.litellm.ai/docs/providers/ollama

That said, LiteLLM has not updated the model registry to include llama 3.2.

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I will definitely look into this as I was unaware of it before.

I won't have time to work on this anytime soon, however.

Comment thread src/loader.py
Comment thread src/loader.py
Comment thread src/loader.py
Comment thread requirements.txt
@meltyli

meltyli commented Oct 3, 2024

Copy link
Copy Markdown
Author

Besides using LiteLLM instead of Ollama's Python library, it was working on Kubuntu 24.04 (CPU)

I tried running on windows and there were dependency issues (pyyaml iirc). I will try a sandboxed environment for windows later in the week.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants