EleutherAI open-sources Bergson, a scalable data attribution library for LLMs, with an EMNLP 2026 oral paper
BlancheMinerva · x · 2026-09-29
EleutherAI researcher Blanche Minerva announced Bergson, an open-source Python library providing scalable, state-of-the-art data attribution methods for LLMs — estimating how removing training data points affects a model's behavior. It supports EK-FAC (2023), TrackStar (2024), SOURCE (2024), MAGIC (2025), ASTRA (2025), gradient cosine similarity, and more. Reproducibility is a core design goal: one-line run reproduction, reusable intermediate artifacts via CLI flags, and a CLI-or-code workflow. The companion paper will be an oral at EMNLP 2026.
Related event: Bergson: An Open-Source Library Unifying LLM Data Attribution Methods(2 posts)→
More from Research
- Anthropic: Claude Autonomously Trains Models to Reliably Mitigate 10 Alignment Failures — inductionheads · 2026-09-29
- Palmer Luckey backs $10M XPRIZE to decode animal communication with AI — PeterDiamandis · 2026-09-29
- Self-play on board games distills superhuman CoT, lifting Qwen3-4B math average from 24.1 to 36.6 — PMinervini · 2026-09-29
- Qualification by Calibration: New Benchmark Audits LLM Annotators Like Crowd Workers — windx0303 · 2026-09-29
- LLM-Managed Papers Are Fine, But Author Names and Titles in Citations Are Wrong — erikphoel · 2026-09-29
- Evaluation Cards, an open standard for eval reporting, wins NeurIPS 2026 Spotlight — evijit · 2026-09-29