Skip to content

Thinking Machines Lab Blog company blog

LoRA Without Regret

by Thinking Machines Lab thinkingmachines.ai published 2025-09-29

How LoRA matches full training performance more broadly than expected.

Read on thinkingmachines.ai → Discuss on HN (58) Archived copy

Points: 184
HN comments: 58
Submissions: 1
Cited by: 0 commenters

AI & Machine Learning Software Engineering 2025

On Hacker News

Submitted	Title on HN	Points	Comments
Sep 29, 2025	LoRA Without Regret by grantpitt	184	58

More from Thinking Machines Lab Blog

All 8 posts →

Inkling: Our Open-Weights Model

20261,227 points291 commentscited by 6AI & Machine Learning details

Defeating Nondeterminism in LLM Inference

2025345 points130 commentscited by 43AI & Machine Learning details

Interaction Models

2026334 points53 commentscited by 7Design & UX details

Inkling 2 Small

2026282 points90 comments AI & Machine Learning details

Modular Manifolds

2025161 points54 comments AI & Machine Learning details

Announcing Tinker

2025152 points89 comments AI & Machine Learning details

Related posts from the same era

Statement on US government directive to suspend access to Fable 5 and Mythos 5

Anthropic20263,158 points2,314 commentscited by 11 details

GPT-4o

OpenAI Blog20243,138 points2,366 commentscited by 16 details

Astra can read and write weights to improve Qwen

Rikiya Takehi20262,957 points898 comments details

HackerTube: YouTube built from videos recommended by HN

karpathy20262,885 points641 comments details

Claude Fable 5

Anthropic20262,626 points2,159 commentscited by 10 details

Claude Code is steganographically marking requests

Thereallo20262,445 points750 commentscited by 3 details