fill-in-the-middle

Tag

Cards List
#fill-in-the-middle

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

arXiv cs.AI · 2026-07-15 Cached

This paper proposes function-aware fill-in-the-middle mid-training for coding agent foundation models, exploiting the structural similarity between function calls and agent action-observation loops. The method improves performance on SWE-Bench benchmarks across multiple model sizes and post-training pipelines.

0 favorites 0 likes
#fill-in-the-middle

Memorization Dynamics of Fill-in-the-Middle Pretraining

arXiv cs.CL · 2026-05-25 Cached

This paper studies how fill-in-the-middle (FIM) pretraining affects verbatim memorization, finding that FIM more often recovers short spans while standard left-to-right training recovers long exact continuations, and that memorization under FIM grows linearly with repetitions.

0 favorites 0 likes
#fill-in-the-middle

Efficient training of language models to fill in the middle

OpenAI Blog · 2022-07-28 Cached

OpenAI presents a simple data augmentation technique that enables autoregressive language models to perform fill-in-the-middle (FIM) text generation without harming left-to-right performance, with extensive ablations and best practices provided for training such models.

0 favorites 0 likes
← Back to home

Submit Feedback