Hacker GitHub

← All repos

workofart/harness-training

View on GitHub ↗ ★ 24 Python

Train a harness to improve its model-agnostic and task-env-agnostic capabilities with PyTorch-like APIs

5 mentions on Hacker News by 1 person: 3 stories, 2 comments · first 2026-07-20 · latest 2026-08-14

Mentions per year 5 mentions · busiest year 2026 (5)
20082026
AllStoriesComments NewestTopOldest
  1. ▲ 4
    Show HN: Auto-train the harness, not the LLM. cross-model, cross-benchmark gains
    story by megadragon9 · 2026-08-14 · 0 comments
  2. …ypan.com/blog/2026-07-18-harness-training/ I packaged the training loop as a PyTorch-style framework. https://github.com/workofart/harness-training
    comment by megadragon9 · 2026-08-05 · in Harness engineering for self-improvement
  3. ▲ 4
    Show HN: Freeze the Model, Train the Harness
    story by megadragon9 · 2026-07-21 · 0 comments
  4. I worked on this project ( https://github.com/workofart/harness-training ) for the past few months to reframe "Agent-driven Self-improving Harness" to "Harness Training". The idea is simple, the harness is trained once with a frozen task…
    comment by megadragon9 · 2026-07-21 · in Ask HN: What Are You Working On? (July 2026)
  5. ▲ 3
    Train a Harness to improve model/env-agnostic capabilities with PyTorch-like API
    story by megadragon9 · 2026-07-20 · 0 comments