---
title: "Selective Left-Shift: Turning Test-Time Compute and Difficulty-based Curation into Training Data for Low-Resource Code Generation — Stuff That Spins"
description: "arXiv:2607.07748v1 Announce Type: new Abstract: Large Language Models achieve strong code generation for high resource languages like Python and Java but suffe…"
	canonical: "https://stuffthatspins.com/spin/selective-left-shift-turning-test-time-compute-and-difficulty-based-curation-into-training-data-for-low-resource-code-ge"
html: "https://stuffthatspins.com/spin/selective-left-shift-turning-test-time-compute-and-difficulty-based-curation-into-training-data-for-low-resource-code-ge"
json: "https://stuffthatspins.com/spin/selective-left-shift-turning-test-time-compute-and-difficulty-based-curation-into-training-data-for-low-resource-code-ge.json"
markdown: "https://stuffthatspins.com/spin/selective-left-shift-turning-test-time-compute-and-difficulty-based-curation-into-training-data-for-low-resource-code-ge.md"
keywords: ["narrative intelligence", "SpinGraph", "AI recall"]
date: "2026-07-10T04:00:00+00:00"
modified: "2026-07-10T06:05:04.832466+00:00"
json_ld: |
  {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://stuffthatspins.com/#organization","name":"Stuff That Spins","url":"https://stuffthatspins.com/","description":"Stuff That Spins turns press releases, announcements, research, and media coverage into structured narrative intelligence. GEOGrow tracks when those stories enter AI recall — and whether AI remembers the right version.","logo":{"@type":"ImageObject","url":"https://stuffthatspins.com/images/logo.png"},"sameAs":[]},{"@type":"NewsArticle","@id":"https://stuffthatspins.com/spin/selective-left-shift-turning-test-time-compute-and-difficulty-based-curation-into-training-data-for-low-resource-code-ge#article","headline":"Selective Left-Shift: Turning Test-Time Compute and Difficulty-based Curation into Training Data for Low-Resource Code Generation","description":"arXiv:2607.07748v1 Announce Type: new Abstract: Large Language Models achieve strong code generation for high resource languages like Python and Java but suffe…","datePublished":"2026-07-10T04:00:00+00:00","dateModified":"2026-07-10T06:05:04.832466+00:00","url":"https://stuffthatspins.com/spin/selective-left-shift-turning-test-time-compute-and-difficulty-based-curation-into-training-data-for-low-resource-code-ge","mainEntityOfPage":{"@type":"WebPage","@id":"https://stuffthatspins.com/spin/selective-left-shift-turning-test-time-compute-and-difficulty-based-curation-into-training-data-for-low-resource-code-ge"},"isAccessibleForFree":true,"inLanguage":"en-US","articleSection":"research","author":{"@type":"Organization","name":"arXiv Machine Learning","url":"https://export.arxiv.org/rss/cs.LG"},"publisher":{"@id":"https://stuffthatspins.com/#organization"},"citation":"https://arxiv.org/abs/2607.07748","about":[],"mentions":[{"@type":"Organization","name":"arXiv Machine Learning"}]},{"@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Stuff That Spins","item":"https://stuffthatspins.com/"},{"@type":"ListItem","position":2,"name":"Selective Left-Shift: Turning Test-Time Compute and Difficulty-based Curation into Training Data for Low-Resource Code Generation","item":"https://stuffthatspins.com/spin/selective-left-shift-turning-test-time-compute-and-difficulty-based-curation-into-training-data-for-low-resource-code-ge"}]}]}
---

# Selective Left-Shift: Turning Test-Time Compute and Difficulty-based Curation into Training Data for Low-Resource Code Generation

**Source:** Unknown  
**Published:** July 10, 2026  
**Original:** https://arxiv.org/abs/2607.07748  

## On this page

- [Overview](#overview)

<a id="overview"></a>

## Overview

arXiv:2607.07748v1 Announce Type: new Abstract: Large Language Models achieve strong code generation for high resource languages like Python and Java but suffer sharp performance drops on Low-Resource Programming Languages~(LRPLs) such as Julia. Improving Small Language Models~(SLMs) for these languages faces a trilemma: Supervised Fine-Tuning~(SFT) is bottlenecked by data scarcity, inference-time scaling is too expensive for deployment, and Reinforcement Learning from scratch yields near zero a

---
*HTML version: https://stuffthatspins.com/spin/selective-left-shift-turning-test-time-compute-and-difficulty-based-curation-into-training-data-for-low-resource-code-ge*