Nathan Lambert
Author of Interconnects; post-training researcher at Ai2
Machine learning researcher at the Allen Institute for AI working on open post-training recipes, and author of the Interconnects newsletter and the freely readable RLHF Book.
Key arguments & positions
- Argues open post-training recipes — not just open weights — are what make model behavior auditable.
- Writes that RLHF and its successors are better understood as product-shaping tools than as alignment solutions.
- Tracks the open-versus-closed model gap closely and publishes his working estimates of it.
Accomplishments
- Post-training lead on the Allen Institute for AI's open OLMo and Tulu model releases; previously at Hugging Face.
- Author of the freely readable RLHF Book.
- Writes Interconnects, widely read among practitioners for its coverage of training method details.
Papers & key writings
Links
In the library
Nothing of theirs is filed in the hub yet. Browse the full library.
Recommended next
Hand-picked from the hub based on what Nathan Lambert covers.
How Shopify Turns Open Models Into Task Experts
Shopify Engineering explains how its Sidekick assistant hands token-heavy work, such as GraphQL queries, to open models Shopify post-trains and serves itself, and the loop it uses to build those task-specific models.
Why this: Covers open models and post training too
Bolmo published in Nature
Allen Institute for AI (Ai2) post from October 7, 2026 on its byte-level language model research appearing in Nature, with released checkpoints applying its "byteifying" method to Qwen 3 8B and Llama 3 8B.
Why this: Also about open models
Towards Understanding Sycophancy in Language Models
Research paper investigating why AI assistants fine-tuned with human feedback tend to produce responses that match user beliefs rather than truthful ones. Analyzes five state-of-the-art AI assistants across text-generation tasks and the role of human preference data in driving this sycophantic behavior.
Why this: Also about RLHF
Reflection AI
Site of Reflection AI, an AI company building open superintelligent systems and coding agents.
Why this: Also about open models