AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Predicting Training Re-evaluation Curves Enables Effective Data Curriculums for LLMs

Shane Bergsma, Nolan Simran Dey, Joel Hestness

ICLR 2026 · Large language models › Language, safety and interpretability · approach: Data and evaluation methodology

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
place high-quality (code) data at the TREC minimum instead of other 10% segments (ablation)610M GPT2-style LLM · code-blend validation loss, step/10x/D2Z LR schedulesvalidation lossImproved
place HQ data at TREC minimum instead of at the very end of training (ablation)610M LLM, step-decay and decay-to-zero schedules · code-blend validation lossvalidation lossImproved
placement curriculum versus uniform training on aggregate blend (ablation)610M LLM · aggregate blend / CB validationvalidation lossImproved
place math HQ data halfway through CPT (TREC minimum) instead of at the end (ablation)3.9B LLM CPT · OpenWebMath validationvalidation lossImproved
use high CPT peak learning rate 0.015 (ablation)3.9B LLM CPT · OpenWebMath validationvalidation lossHurt
insert HQ data at predicted TREC minimum (~97%) during pre-training instead of CPT (ablation)3.9B LLM · OpenWebMath validationvalidation lossNo clear effect
increase number of MoE experts from 1 to 32 (ablation)111M MoE · TREC dropTREC depthImproved
anneal LR over final ~3 optimizer steps on GSM8k/MATH train setLlama 3 405B · GSM8k validationaccuracyNo clear effect
match AdamW timescale tau across LR, weight decay, batch size (ablation)610M LLM · TREC shapeTRECImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy