Function-Aware Fill-in-the-Middle as Mid-Training for Coding Agent Foundation Models
:::info Stub — Full Engineering Breakdown Coming This paper was featured on Hugging Face Daily Papers on 2026-07-14 with 104 upvotes. A full breakdown with production viability rating, implementation notes, and honest limitations is being written. Subscribe to AI Letters → :::
| Authors | Yubo Wang et al. |
| Year | 2026 |
| HF Upvotes | 104 |
| arXiv | 2607.12463 |
| Download | |
| HF Page | View on Hugging Face |
Abstract
Coding agents must integrate external tool returns into ongoing reasoning - a capability that standard left-to-right pretraining on code exposes only in its forward direction. We observe that the action-observation-continuation loop of a coding agent is structurally isomorphic to a function call site, where a caller binds arguments, a callee returns a value computed elsewhere, and downstream code consumes that value. This conditioning structure exists at internet scale in ordinary code. We exploit it through function-aware fill-in-the-middle (FIM) mid-training: a self-supervised objective that masks functions selected via program dependency graph analysis and a complexity-inferability double criterion. We mid-train Qwen2.5-Coder-Instruct (7B/14B) and Qwen3-8B on a 2.6B-token decontaminated corpus drawn from 968 GitHub repositories, then apply existing agentic post-training pipelines. Mid-training improves SWE-Bench-Verified by +2.8/+3.0 at 7B/14B and by +3.2 on Qwen3-8B; SWE-Bench-Lite gains are +3.7/+4.0/+5.4 on the same models. The improvement holds across two post-training pipelines (R2E-Gym, SWE-Smith) and on a non-Qwen2.5 base (Qwen3-8B with SWE-Lego). Beyond in-domain gains, mid-training also mitigates the capability erosion that agentic post-training otherwise inflicts on non-agent coding (e.g., LiveCodeBench) and non-coding tool-use benchmarks (tau-bench, BFCL): although the mid-training corpus contains Python code only, the function-call inductive bias survives post-training and yields consistent gains.
Engineering Breakdown
The Problem
Coding agents must integrate external tool returns into ongoing reasoning - a capability that standard left-to-right pretraining on code exposes only in its forward direction.
The Approach
We observe that the action-observation-continuation loop of a coding agent is structurally isomorphic to a function call site, where a caller binds arguments, a callee returns a value computed elsewhere, and downstream code consumes that value.
Key Results
Beyond in-domain gains, mid-training also mitigates the capability erosion that agentic post-training otherwise inflicts on non-agent coding (e.g., LiveCodeBench) and non-coding tool-use benchmarks (tau-bench, BFCL): although the mid-training corpus contains Python code only, the function-call inductive bias survives post-training and yields consistent gains.
Research Areas
This paper contributes to the following areas of AI/ML engineering:
- Machine learning
- Deep learning
- Neural networks
- Model optimization
- AI systems
- Functionaware
:::tip Subscribe Get weekly breakdowns of papers like this in AI Letters - the newsletter for engineers building production AI systems. :::