Pinned post
🚨New paper🚨 Self-Execution Simulation Improves Coding LLMs
Current reasoning LLMs <think> before giving a code solution.
We post-train them to explicitly simulate test execution, in order to verify and fix their own code leading to additional gains!
Details 👇🏽🧵
Current reasoning LLMs <think> before giving a code solution.
We post-train them to explicitly simulate test execution, in order to verify and fix their own code leading to additional gains!
Details 👇🏽🧵