Back to Discover
LILil'Log

Lil'Log

RSS Healthy

lilianweng.github.io · AI & ML · en

In-depth technical notes on machine learning, LLM agents, reinforcement learning, and AI safety by Lilian Weng.

Visit Website
Indexed Posts
20
Update Frequency
Monthly
Last Fetched
1h ago
Feed URL
https://lilianweng.github.io/index.xml
Lil'Log homepage preview
lilianweng.github.io

Lil'Log

In-depth technical notes on machine learning, LLM agents, reinforcement learning, and AI safety by Lilian Weng.

Loading snapshot...·

Recent Articles

Jul 4, 2026Lil'Log

Harness Engineering for Self-Improvement

The concept of recursive self-improvement (RSI) dates back to I. J. Good (1965), where he defined an “ultraintelligent machine” as a system that can surpass humans in all intellectual activities and design better machines to improve itself. Yudkowsky (2008) u…

Jun 24, 2026Lil'Log

Scaling Laws, Carefully

Scaling laws are one of the most critical empirical findings in deep learning. The observation is simple in form: the training loss $L$ decreases predictably as we scale up model size $N$, dataset size $D$, and compute $C$, following a power-law curve, which…

May 1, 2025Lil'Log

Why We Think

Special thanks to John Schulman for a lot of super valuable feedback and direct edits on this post. Test time compute (Graves et al. 2016, Ling, et al. 2017, Cobbe et al. 2021) and Chain-of-thought (CoT) (Wei et al. 2022, Nye et al. 2021), have led to signifi…

Nov 28, 2024Lil'Log

Reward Hacking in Reinforcement Learning

Reward hacking occurs when a reinforcement learning (RL) agent exploits flaws or ambiguities in the reward function to achieve high rewards, without genuinely learning or completing the intended task. Reward hacking exists because RL environments are often im…

Jul 7, 2024Lil'Log

Extrinsic Hallucinations in LLMs

Hallucination in large language models usually refers to the model generating unfaithful, fabricated, inconsistent, or nonsensical content. As a term, hallucination has been somewhat generalized to cases when the model makes mistakes. Here, I would like to na…

Apr 12, 2024Lil'Log

Diffusion Models for Video Generation

Diffusion models have demonstrated strong results on image synthesis in past years. Now the research community has started working on a harder task—using it for video generation. The task itself is a superset of the image case, since an image is a video of 1…

Feb 5, 2024Lil'Log

Thinking about High-Quality Human Data

[Special thank you to Ian Kivlichan for many useful pointers (E.g. the 100+ year old Nature paper “Vox populi”) and nice feedback. 🙏 ] High-quality data is the fuel for modern data deep learning model training. Most of the task-specific labeled data comes fr…

Oct 25, 2023Lil'Log

Adversarial Attacks on LLMs

The use of large language models in the real world has strongly accelerated by the launch of ChatGPT. We (including my team at OpenAI, shoutout to them) have invested a lot of effort to build default safe behavior into the model during the alignment process (…

LLM Powered Autonomous Agents
Jun 23, 2023Lil'Log

LLM Powered Autonomous Agents

Building agents with LLM (large language model) as its core controller is a cool concept. Several proof-of-concepts demos, such as AutoGPT, GPT-Engineer and BabyAGI, serve as inspiring examples. The potentiality of LLM extends beyond generating well-written c…

Mar 15, 2023Lil'Log

Prompt Engineering

Prompt Engineering, also known as In-Context Prompting, refers to methods for how to communicate with LLM to steer its behavior for desired outcomes without updating the model weights. It is an empirical science and the effect of prompt engineering methods ca…

Jan 27, 2023Lil'Log

The Transformer Family Version 2.0

Many new Transformer architecture improvements have been proposed since my last post on “The Transformer Family” about three years ago. Here I did a big refactoring and enrichment of that 2020 post — restructure the hierarchy of sections and improve many sect…

Jan 10, 2023Lil'Log

Large Transformer Model Inference Optimization

[Updated on 2023-01-24: add a small section on Distillation.] Large transformer models are mainstream nowadays, creating SoTA results for a variety of tasks. They are powerful but very expensive to train and use. The extremely high inference cost, in both tim…

Sep 8, 2022Lil'Log

Some Math behind Neural Tangent Kernel

Neural networks are well known to be over-parameterized and can often easily fit data with near-zero training loss with decent generalization performance on test dataset. Although all these parameters are initialized at random, the optimization process can co…

Jun 9, 2022Lil'Log

Generalized Visual Language Models

Processing images to generate text, such as image captioning and visual question-answering, has been studied for years. Traditionally such systems rely on an object detection network as a vision encoder to capture visual features and then produce text via a t…

Apr 15, 2022Lil'Log

Learning with not Enough Data Part 3: Data Generation

Here comes the Part 3 on learning with not enough data (Previous: Part 1 and Part 2). Let’s consider two approaches for generating synthetic data for training. Augmented data. Given a set of existing training samples, we can apply a variety of augmentation, d…

Feb 20, 2022Lil'Log

Learning with not Enough Data Part 2: Active Learning

This is part 2 of what to do when facing a limited amount of labeled data for supervised learning tasks. This time we will get some amount of human labeling work involved, but within a budget limit, and therefore we need to be smart when selecting which sampl…

Sep 24, 2021Lil'Log

How to Train Really Large Models on Many GPUs?

[Updated on 2022-03-13: add expert choice routing.] [Updated on 2022-06-10]: Greg and I wrote a shorted and upgraded version of this post, published on OpenAI Blog: “Techniques for Training Large Neural Networks”

Jul 11, 2021Lil'Log

What are Diffusion Models?

[Updated on 2021-09-19: Highly recommend this blog post on score-based generative modeling by Yang Song (author of several key papers in the references)]. [Updated on 2022-08-27: Added classifier-free guidance, GLIDE, unCLIP and Imagen. [Updated on 2022-08-31…

May 31, 2021Lil'Log

Contrastive Representation Learning

The goal of contrastive representation learning is to learn such an embedding space in which similar sample pairs stay close to each other while dissimilar ones are far apart. Contrastive learning can be applied to both supervised and unsupervised settings. W…