Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Add GPT-4 #1457

Merged
merged 3 commits into from
Apr 20, 2023
Merged

Add GPT-4 #1457

merged 3 commits into from
Apr 20, 2023

Conversation

yifanmai
Copy link
Collaborator

@yifanmai yifanmai commented Apr 4, 2023

Addresses #1456

@yifanmai yifanmai requested review from teetone and percyliang April 4, 2023 20:21
release_date: 2023-03-14
- name: openai/gpt-4-32k-0314
display_name: gpt-4-32k-0314
description: GPT-4 is a large multimodal model (currently only accepting text inputs and emitting text outputs) that is optimized for chat but works well for traditional completions tasks. Snapshot of gpt-4 from March 14th 2023.
Copy link
Contributor

Choose a reason for hiding this comment

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

Description should mention that the context length is 32K tokens.

Copy link
Collaborator Author

Choose a reason for hiding this comment

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

Added to description.

Copy link
Contributor

@percyliang percyliang left a comment

Choose a reason for hiding this comment

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

Great! Yes, agreed that we should make the window size less clunky...

@pramitchoudhary
Copy link

Hey Guys, this is a great initiative. Thanks for starting it.
I was curious to try it out on the GPT4 models, so I was playing with the changes on this branch.

  1. From my limited understanding, with these changes, one would need to specify an open API key using by adding it to prod_env/credentials.conf
  2. echo 'entries: [{description: "mmlu:subject=philosophy,model=openai/gpt-4-32k-031", priority: 1}]' > run_specs.conf
  3. On attempting helm-run --conf-paths run_specs.conf --local --suite v1 --max-eval-instances 10 throws error,
File "./helm/helm-venv/lib/python3.8/site-packages/helm/proxy/models.py", line 684, in get_model
    raise ValueError(f"No model with name: {model_name}")
ValueError: No model with name: openai/gpt-4-32k-031

Am i missing an obvious step?

@yifanmai
Copy link
Collaborator Author

yifanmai commented Apr 7, 2023

You are probably misspelling the model name: it is openai/gpt-4-32k-0314, not openai/gpt-4-32k-031.

The other steps are correct; I will add them to the documentation soon.

@pramitchoudhary
Copy link

pramitchoudhary commented Apr 8, 2023

openai/gpt-4-32k-031

Sorry, above was a typo. Same error with openai/gpt-4-32k-0314 --> ValueError: No model with name: openai/gpt-4-32k-0314

Also, on, trying with other previous versions of GPT, getting ChatGPT error: local variable 'driver' referenced before assignment, probably from the ChatGPT client used? Current version: revChatGPT==0.1.1

I will add them to the documentation soon.

No problem, 🙏

@yifanmai yifanmai merged commit 44dc5fa into main Apr 20, 2023
@yifanmai yifanmai deleted the yifanmai/1456-gpt-4 branch April 20, 2023 04:02
@yifanmai
Copy link
Collaborator Author

@pramitchoudhary, we will be releasing v0.2.4 of the package next week; you should try installing the updated package (or running from main today) and see if that resolves your issues. You should open a new GitHub issue if this still doesn't work for you.

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.

3 participants