Each page rendered as one 150-DPI PNG. Page images are stored ONCE per distinct document
under images/<doc>/pNNNN.png and referenced by each cell question.json (image_docs,
in load order: source first, then distractors). 1350 cells, all hard negatives in [10,40] pages.
Part of MUDDLE: Measuring Understanding of Documents under Distractor and Length Effects (COLM 2026 Workshop on Context Beyond the Window).
| Piece | Where |
|---|---|
| Code | https://github.com/luoojason/muddle |
| Questions + hard-negative metadata | https://huggingface.co/datasets/luoojason/mmlongbench-text-only |
| Eval cells, PDF modality | https://huggingface.co/datasets/luoojason/muddle-eval-bundle |
| Eval cells, markdown modality | https://huggingface.co/datasets/luoojason/muddle-eval-bundle-md |
| Eval cells, page-image modality | https://huggingface.co/datasets/luoojason/muddle-eval-bundle-img |
| Paper | https://openreview.net/forum?id=9YS7tSPM9c |
52 commits
Each page rendered as one 150-DPI PNG. Page images are stored ONCE per distinct document
under images/<doc>/pNNNN.png and referenced by each cell question.json (image_docs,
in load order: source first, then distractors). 1350 cells, all hard negatives in [10,40] pages.
Part of MUDDLE: Measuring Understanding of Documents under Distractor and Length Effects (COLM 2026 Workshop on Context Beyond the Window).
| Piece | Where |
|---|---|
| Code | https://github.com/luoojason/muddle |
| Questions + hard-negative metadata | https://huggingface.co/datasets/luoojason/mmlongbench-text-only |
| Eval cells, PDF modality | https://huggingface.co/datasets/luoojason/muddle-eval-bundle |
| Eval cells, markdown modality | https://huggingface.co/datasets/luoojason/muddle-eval-bundle-md |
| Eval cells, page-image modality | https://huggingface.co/datasets/luoojason/muddle-eval-bundle-img |
| Paper | https://openreview.net/forum?id=9YS7tSPM9c |
52 commits