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

Update Llama.cpp & Llama 3 Support #55

Merged
merged 9 commits into from
May 13, 2024
Merged

Conversation

PSchmiedmayer
Copy link
Member

Update Llama.cpp & Llama 3 Support

⚙️ Release Notes

  • Updates llama.cpp
  • Supports Llama 3

📝 Code of Conduct & Contributing Guidelines

By submitting creating this pull request, you agree to follow our Code of Conduct and Contributing Guidelines:

@PSchmiedmayer PSchmiedmayer added the enhancement New feature or request label May 2, 2024
@PSchmiedmayer PSchmiedmayer self-assigned this May 2, 2024
@PSchmiedmayer
Copy link
Member Author

@vishnuravi @philippzagar Feel free to add to this WIP here after lifting llama.cpp. @vishnuravi can you share some of the issues you had with Llama3 so we can document this here in case this gets stale or someone else wants to pick it up?

@vishnuravi
Copy link
Member

vishnuravi commented May 10, 2024

@vishnuravi @philippzagar Feel free to add to this WIP here after lifting llama.cpp. @vishnuravi can you share some of the issues you had with Llama3 so we can document this here in case this gets stale or someone else wants to pick it up?

Hi @PSchmiedmayer, I'm running into the same issue documented here: ollama/ollama#3759 and here: https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct/discussions/4. The model does not stop at the stop token and continues generating indefinitely.

@vishnuravi vishnuravi self-assigned this May 10, 2024
@vishnuravi vishnuravi self-requested a review May 10, 2024 19:01
@vishnuravi
Copy link
Member

@vishnuravi @philippzagar Feel free to add to this WIP here after lifting llama.cpp. @vishnuravi can you share some of the issues you had with Llama3 so we can document this here in case this gets stale or someone else wants to pick it up?

Hi @PSchmiedmayer, I'm running into the same issue documented here: ollama/ollama#3759 and here: https://huggingface.co/meta-llama/Meta-Llama-3-8B-Instruct/discussions/4. The model does not stop at the stop token and continues generating indefinitely.

I was able to resolve this issue by using a different GGUF with the end token correctly configured.

Copy link

codecov bot commented May 11, 2024

Codecov Report

Attention: Patch coverage is 0% with 91 lines in your changes are missing coverage. Please review.

Project coverage is 31.18%. Comparing base (cbaf204) to head (98753fc).

Additional details and impacted files

Impacted file tree graph

@@            Coverage Diff             @@
##             main      #55      +/-   ##
==========================================
- Coverage   32.03%   31.18%   -0.85%     
==========================================
  Files          67       67              
  Lines        2932     3012      +80     
==========================================
  Hits          939      939              
- Misses       1993     2073      +80     
Files Coverage Δ
...ces/SpeziLLMLocal/LLMLocalSession+Generation.swift 0.00% <0.00%> (ø)
...s/SpeziLLMLocal/LLMLocalSession+Tokenization.swift 0.00% <0.00%> (ø)
...ocal/Configuration/LLMLocalContextParameters.swift 0.00% <0.00%> (ø)
...Download/LLMLocalDownloadManager+DefaultUrls.swift 0.00% <0.00%> (ø)
...peziLLMLocal/LLMLocalSchema+PromptFormatting.swift 0.00% <0.00%> (ø)

Continue to review full report in Codecov by Sentry.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update cbaf204...98753fc. Read the comment docs.

@PSchmiedmayer PSchmiedmayer merged commit 94f14f6 into main May 13, 2024
17 of 18 checks passed
@PSchmiedmayer PSchmiedmayer deleted the feature/updatellamacpp branch May 13, 2024 06:48
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
enhancement New feature or request
Projects
None yet
Development

Successfully merging this pull request may close these issues.

None yet

2 participants