open-compass / open-compass/VLMEvalKit

Are images compressed in tsv files?

Open
#517 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
4.4k
Forks
768
Avg merge
1d 10h
Merged PRs (30d)
17

Description

Thanks for your contribution to providing a collection of VLM datasets and models. I'm wondering why the tsv versions of datasets in this repository are smaller than the official versions. For example, the RealworldQA dataset downloaded from the official website RealWorldQA has 677MB, while the tsv version in this repo RealWorldQA_tsv only has 175MB. You are using base64 to encode images into texts and store them directly in tsv columns, which should be lossless. So why has the data size been reduced significantly? It seems that other datasets are having the similar situation.

Contributor guide

No contributing guide indexed for this repository

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

Compare the official RealWorldQA archive linked in the issue with the repository's RealWorldQA.tsv, focusing on how images are encoded into the TSV columns. Determine why the sizes differ and document the confirmed explanation for this dataset and whether the same cause applies to the other datasets mentioned.

Written by the indexing model from the issue text.

Assessment

Domain
data
Issue type
Bug
Difficulty
3/5
Estimated time
1-2 days
Activity status
Stale
Clarity
Needs clarification
Newbie friendliness
25/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.