Official Repository of "LLM × DATA" Survey Paper
-
Updated
Oct 25, 2026
Official Repository of "LLM × DATA" Survey Paper
A Python library for combining multiple datasets using customizable algorithms. It offers a flexible framework to control the contribution of each dataset in the final output.
Unofficial reproduction of Nemotron-CLIMB (CLustering-based Iterative Data Mixture Bootstrapping, NeurIPS 2025, arXiv:2504.13161) on Ascend NPUs — 5 rounds, ≈3,900 NPU-hours; learned mixtures beat every fixed-ratio baseline (+0.014–0.043 STEM); the selection mechanism is the weak link — fixed and validated live.
LLM pretraining data mixture research insights
To associate your repository with the data-mixing topic, visit your repo's landing page and select "manage topics."