mlcommons / mlcommons/inference

Dataset preprocessing code is not shared for several models

Open
#2,245 1 comment 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
1.6k
Forks
650
Avg merge
1d 22h
Merged PRs (30d)
6

Description

Hi,

https://github.com/mlcommons/inference/tree/master/language/llama2-70b provides processorca.py to generate the eval dataset, but several other models as https://github.com/mlcommons/inference/tree/master/language/deepseek-r1 and https://github.com/mlcommons/inference/tree/master/language/llama3.1-8b do not provide a reproduction script to generate the evaluation data.

This is an issue if e.g. one would like to apply the evaluation from mlperf on other models that may use other tokenizers.

Could the code used to generate this preprocessed data be shared?

Thank you

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 with language/llama2-70b/processorca.py, then compare the language/deepseek-r1 and language/llama3.1-8b directories. Determine what preprocessing is needed for each model and tokenizer. Done means the preprocessing code for the referenced models is shared and can reproduce their evaluation data.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
data, machine-learning
Issue type
Feature
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
45/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.