Before They Can Solve: Predicting Post-Training Coding-Agent Performance from Base Models

Paper recorded by Signals 4 on 2026-10-07 in cs.AI. Abstract reproduced from arXiv; link to the original below.

Published 2026-10-07 on arXiv · recorded by Signals 4 on 2026-10-08

Category: cs.AI · 人工智能 · first seen 2026-10-08

Abstract

How can we predict which base checkpoint is worth an expensive round of agentic post-training? End-to-end pass@$K$ tests whether successful behavior already appears in a base model's distribution, but it is a poor fit for agentic coding: many base checkpoints cannot reliably produce the well-formed tool invocation required to complete a task end-to-end. Single-shot or short-horizon tasks avoid the

Read on arXiv →