馃憢 Hi, all!
I am Elliot Shichao Song, a PhD candidate. My current research focuses on foundation models. I blog about my research and life.
Things I've Made
Products and small things I have made, all in one place.
)](https://ki-seki.github.io/posts/261107-billion-image-download/assets/billion-image-download-cover.png)
Efficient Large-Scale Crawling with an AI Agent: 3 Days, $175, 3 Billion Original Images
Downloading the originals behind 2.98 billion image URLs in OmniCorpus-CW, I left almost all operations to an AI agent: patrols, rate tuning, creating and deleting machines, incident handling and billing. I only made the decisions. In 79 hours it saved 99.6% of the images, 744 TB, for about $175 in machines.
)](https://ki-seki.github.io/posts/261029-captionexam/assets/captionexam-cover.png)
CaptionExam: A Challenging QA Benchmark for Captioners
On CaptionExam, even GPT-6 Astra only just passes at describing images, with a score of 61.2%.
)](https://ki-seki.github.io/posts/261025-training-data-retrieval/assets/semantic-islands-cover.png)
A Billion-Scale Retrieval System for Text-to-Image Training Data
Training a text-to-image model from scratch means finding, inspecting and filtering samples across billions of image鈥搕ext pairs. Using a retrieval system over 1.4B samples as an example, this post covers data storage and management, distributed embedding extraction, building and querying an IVF-PQ index, and online retrieval and cluster browsing.
)](https://ki-seki.github.io/posts/260531-lilian-rl-overview/RL_illustration_new.png)
Fully Annotated Guide to "A (Long) Peek into Reinforcement Learning"
This is a fully annotated guide to Lilian Weng鈥檚 post A (Long) Peek into Reinforcement Learning.