GitHub

Post-Completion Learning for Language Models

Our code is based on open-r1, with our customized Trainer for mixed SFT+GRPO training. Some other updates focus on the white-box RL (reward function design) and post-completion training (replacement ...
CNET

I Tried Vibe Coding With Different Gemini Models. Here's What I Learned

A slower "reasoning" model might do more of the work for you -- and keep vibe coding from becoming a chore.
IEEE

Aligning Crowd-Sourced Human Feedback for Reinforcement Learning on Code Generation by Large Language Models

Abstract: This paper studies how AI-assisted programming and large language models (LLM) improve software developers' ability via AI tools (LLM agents) like Github Copilot and Amazon CodeWhisperer, ...

How do AI coding agents work? We look under the hood.

At the core of every AI coding agent is a technology called a large language model (LLM), which is a type of neural network ...
IEEE

Learning Rate-Compatible Linear Block Codes: An Auto-Encoder Based Approach

Abstract: Artificial intelligence (AI) provides an alternative way to design channel coding with affordable complexity. However, most existing studies can only learn codes for a given size and rate, ...