Notes & talks

Thinking out loud.

Notes from papers I’m reading, questions I’m working through, and ideas shared along the way.

Reading notes

01 entry
Paper notes

Reading DeepSeek-R1,
the Nature version.

What the multistage training pipeline reveals about reasoning, SFT, and reinforcement learning — and the questions it leaves open.

ReasoningReinforcement learning

Talks & presentations

Selected

EMNLP 2024 · Miami

RepoAgent: a repository-level documentation assistant

A demo and technical introduction to keeping code documentation in step with a changing repository.

MiraclePlus · Beijing

Introduction to XAgent and RepoAgent

LLM agents in practice: autonomous task execution, multi-agent architectures, and repository documentation.

CICC · Beijing

ChatGPT and language models in finance

A technical discussion of the GPT model family, model tuning, and possible applications of language models in the financial industry.

READ HERE · 35 SLIDES

RepoAgent · EMNLP 2024 slides

Presentation

Loading document…

Scroll through every page. Select text to copy, or use search to find a passage.

READ HERE · 1 PAGES

RepoAgent · EMNLP 2024 poster

Document

Loading document…

Scroll through every page. Select text to copy, or use search to find a passage.

WATCH HERE · SYSTEM DEMONSTRATION

RepoAgent in action

GET IN TOUCH

Let’s talk.

For research, collaboration, or a conversation. Email me or connect on LinkedIn and X.

You can also select and copy the address above.