Textbooks Are All You Need

Textbooks Are All You Need — phi-1

Why it matters

Trained a 1.3B model on 7B tokens to reach 50.6% on HumanEval, demonstrating that data quality matters more than quantity

Source — Primary sources first; secondary reports are labelled “Press”

Paperarxiv.orghttps://arxiv.org/abs/2306.11644

Added in Week 35, 2026 · See that week

Same week

Hugging Face explains the MMLU score mismatch on the Open LLM LeaderboardOfficialUC Berkeley releases vLLM + PagedAttentionOfficial

Nearby on the same axis

RefinedWebPaperRedPajama-Data-1TOfficialNLP Evaluation in troublePaperRethinking Benchmark and ContaminationPaper