mlfoundations / mlfoundations/evalchemy

How to change model's temperature ?

Open
#99 2 comments 4 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
HTML
Stars
610
Forks
87
PR merge metrics
No merged PRs in 30d

Description

Hello, I could not find an obvious way from the given examples on how one could change the model temparature (and number of runs, which defaults to 5) used for the generations?

For example, if i am running the script like this
python -m eval.eval\ --model vllm \ --tasks AIME24,AIME25,MATH500,GPQADiamond,AMC23 \ --model_args "pretrained=deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B" \ --batch_size 4 \ --output_path logs

It looks like the temperature of 0.7 is used (but with do_sample = False for some reason) how can I change this? passing temperature to model_args doesn't recognize it

Thanks a lot!

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start at the eval.eval command entry point and trace how model_args, temperature, do_sample, and the default number of runs are handled. Confirm the currently supported invocation and document how to set these values, or clarify the missing configuration behavior; done when the example command and resulting settings are unambiguous.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
ai, machine-learning
Issue type
Documentation
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.