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 lm-eval-harness to 0.4.3 #91

Merged
merged 5 commits into from
Jul 18, 2024
Merged

Conversation

AguirreNicolas
Copy link
Contributor

@AguirreNicolas AguirreNicolas commented Jul 17, 2024

Configs : The configs collection is added where the AutoConfig document is stored. This information is used to know, for now, the max lenght of a model.
For now the tokenizer endpoint in the sidecard is still used. As for the hash, for now it is matched with the hash of the tokenizer. In the future this should be separated and generate its own signature.

EvaluatorLM: OpenAI stopped supporting the echo parameter while vLLM keeps it. We will follow vLLM for now. This change impacts the get_result function that was defined for OpenAI. We will keep the get_result function so that it is compatible with vLLM.

Random seeds: In the new version, Pytohn, NumPy, and PyTorch seeds were segregated. In our code we will generate the seeds randomly. In future versions we can include the seeds as part of the request.

Tokenizer siguature: Fix a bug when saving the tokenizer and its signature.

SQL query: Now it apply a between ranges ids, intead of a list.

Close #89

@AguirreNicolas AguirreNicolas linked an issue Jul 17, 2024 that may be closed by this pull request
@RawthiL RawthiL merged commit bd31b33 into main Jul 18, 2024
4 checks passed
@RawthiL RawthiL deleted the 89-update-to-lm-eval-harness-043 branch July 18, 2024 19:05
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.

Update to lm-eval-harness== 0.4.3
2 participants