Lilian Weng

Channel and publications

Lil'Log's archive begins in June 2017 with "An Overview of Deep Learning for Curious People." Early posts include explainers on GANs and Wasserstein GANs, word embeddings, a multi-part series on object detection that includes a post on the R-CNN family, and a 2018 post on the multi-armed bandit problem.²

From 2023 the archive turns largely to large language models, with posts on inference optimization, a second version of "The Transformer Family," prompt engineering, LLM-powered autonomous agents, and adversarial attacks on LLMs. Posts in 2024 cover high-quality human data, diffusion models for video generation, extrinsic hallucinations, and reward hacking in reinforcement learning.²

More recent long-form posts include "Why We Think" (May 2025), "Scaling Laws, Carefully" (June 2026), and "Harness Engineering for Self-Improvement" (July 2026). Every post in the archive is credited to Lilian Weng and lists an estimated reading time.²¹

Editorial practice

The blog's FAQ says Lilian Weng updates older posts periodically and adds a one-line note at the top of each revised post. It invites readers to report errors or suggest missing papers, allows translations that keep a link to the original at the top, and offers an RSS feed for new posts.³

Work and contributions

• Lil'Log Show, channel or publication

  Writes and maintains long-form posts that review machine learning research topics.¹²³

Timeline

1. Jun 21, 2017Published the first Lil'Log post, "An Overview of Deep Learning for Curious People."²
2. Dec 31, 2017Published part 3 of the object detection series, on the R-CNN family.²
3. Jan 27, 2023Published "The Transformer Family Version 2.0."²
4. Jun 23, 2023Published "LLM Powered Autonomous Agents."²
5. Nov 28, 2024Published "Reward Hacking in Reinforcement Learning."²
6. May 1, 2025Published "Why We Think."²
7. Jul 4, 2026Published "Harness Engineering for Self-Improvement."²

Where they work

• Lil'Log 1 person in the wiki

Related people

• Nathan LambertLanguage models · Reinforcement learning
• Denny BritzLanguage models · Reinforcement learning
• Sasha RushLanguage models · Teaching and writing
• Aravind SrinivasLanguage models · Reinforcement learning
• Noam BrownReinforcement learning · Language models
• Dan JurafskyLanguage models · Teaching and writing
• AI ExplainedTeaching and writing · Language models
• Jakub PachockiLanguage models · Reinforcement learning
• Percy LiangLanguage models · Teaching and writing
• Younes BelkadaLanguage models · Reinforcement learning