Entity · paper

Function-Aware Fill-in-the-Middle as Mid-Training for Coding Agent Foundation Models

paperactiveprovisionalfunction-aware-fill-in-the-middle-as-mid-training-for-coding-agent-foundation-models-89e5f224·1 events·first seen 5d ago

Aliases: Function-Aware Fill-in-the-Middle as Mid-Training for Coding Agent Foundation Models

Co-occurring entities

SWE-Smith SWE-Bench Lite Qwen2.5-Coder-32B-Instruct SWE-Bench Verified Qwen3-4B LiveCodeBench BFCL R2E-Gym TAU-bench

More like this (12)

Model-Generated Agent Skills (paper)Frontier Coding Agents Use Metaprogramming to Adapt to Unfamiliar Programming Languages OpenAI Agent Builder Multi-Agent Fictitious Play Self-Compacting Language Model Agents Generative Skill Composition for LLM Agents memory-augmented agent frameworks Fill-in-the-Middle (FIM)From Atomic Actions to Standard Operating Procedures: Iterative Tool Optimization for Self-Evolving LLM Agents Role-Agent: Bootstrapping LLM Agents via Dual-Role Evolution An Agency-Transferring Model-Free Policy Enhancement Technique Korea Proprietary AI Foundation Model Program

Recent events (1)

6arXiv · cs.CL·5d ago·source ↗

Function-Aware Fill-in-the-Middle Mid-Training Improves Coding Agent Foundation Models

Researchers propose a self-supervised mid-training objective called function-aware fill-in-the-middle (FIM) that exploits the structural isomorphism between a coding agent's action-observation-continuation loop and function call sites in ordinary code. Applied to Qwen2.5-Coder-Instruct (7B/14B) and Qwen3-8B on a 2.6B-token GitHub corpus, the method yields +2.8 to +5.4 point gains on SWE-Bench-Verified and SWE-Bench-Lite across multiple post-training pipelines. Notably, the technique also mitigates capability erosion on non-agent coding and tool-use benchmarks, suggesting the function-call inductive bias generalizes beyond the training domain.

Frontier Model Releases Evaluation and Benchmarking SWE-Smith SWE-Bench Lite Qwen2.5-Coder-32B-Instruct +8 more