Papers.Data.Code
The Weekly ML Signal
Selected, not collected.
Weekly edition · Aug 31 – Sep 06 · Issue #3
Grounded evaluation gets harder
Long-document QA, scientific reasoning data, and RAG scoring.
Top 3 papers
- Repo-To-Skill: Distilling GitHub Repositories Into AI4AI Skills→
It adds verified, reusable skill graphs distilled from repositories, boosting fixed-budget research agents 134.3% on MLE-bench.
- SolarWM: Open Data and Scalable Training for Long-Horizon Video World Models→
Standardizes 10 video sources and trains 5B-33B backbones for hours-long rollouts from 5-second clips.
- Does On-Policy Distillation Really Distill? From Noisy Teacher to Self-Improvement→
It replaces noisy teacher-guided on-policy distillation with teacher-free entropy-scaled token penalties, gaining +35.41 Avg@32 on AIME24.
Top 3 repos
- Tencent/WeMM-Embedding→
Unlike single-modality embedders, it unifies five input types in one space and retains 98.7% image/video performance at 256 dimensions.
- MLO-lab/LeVJEPA→
Eliminates EMA targets, predictors, and masking yet beats V-JEPA 2 with 5.6-20.8× less pretraining compute.
- amap-cvlab/ABot-Recon→
Streams monocular 3D reconstruction causally over arbitrary-length videos with fixed 12-frame memory.
Top 3 datasets
- microsoft/XL-DocBench→
Unlike broad corpora, it tests evidence-grounded QA over documents up to 2,935 pages with page-level evidence, cross-document questions, abstention, and only 38.36% top accuracy.
- OpenDataArena/Spark-234K→
Unlike raw-text scientific corpora, it builds 234K QA pairs from paper-level reasoning skeletons with claims, evidence, quantities, assumptions, and limits.
- DISLab/Q-CARE→
Unlike larger QA datasets, it pairs 800 queries with 16 RAG outputs each for reference-free evaluation across both retrieval and generation.
One useful ML email a week.
The Weekly ML Signal.
Worth your time, not all of your time.
We track new ML work across papers, code, and data, then keep only the pieces worth a closer look. Selected, not collected.