#AutoResearch
AutoResearch and the Fine-Tuning Ontology: Turning Institutional Workflows into an Owned Model Factory

agientry.com/en/articles/...
AutoResearch and the Fine-Tuning Ontology
A complete guide to Palantir's AutoResearch and Fine-Tuning Ontology, from datasets and SFT, DPO, and GRPO to an 11-step workflow for building, evaluating…
agientry.com
November 16, 2026 at 2:40 AM
CDS Asst. Prof. Mengye Ren (@mengyer.bsky.social) and his lab released @outerloop.science, open-source software whose AI agents open pull requests only when their experiments improve a benchmark.

The agents got a GPT model to its target in 77% fewer training steps.

outerloop.science
Outerloop: autoresearch agents that improve your benchmark
Autonomous agents come up with ideas, propose changes, run experiments, and open pull requests when your benchmark improves. Attach it to your repo and your cluster.
outerloop.science
November 11, 2026 at 7:12 PM
Qwen-UI-Agent is a real-device centric GUI agent combining unified GUI+CLI actions, long-horizon training, and an AutoResearch-style improvement loop across multiple environments.
Save What Matters
Curate Feeds | Make Collections | Customize Email Briefs
briefly.co
November 14, 2026 at 7:23 AM
Qwen-UI-Agent is a real-device centric GUI agent combining unified GUI+CLI actions, long-horizon training, and an AutoResearch-style improvement loop across multiple environments.
Save What Matters
Curate Feeds | Make Collections | Customize Email Briefs
briefly.co
November 14, 2026 at 6:19 AM
Qwen-UI-Agent is a real-device centric GUI agent combining unified GUI+CLI actions, long-horizon training, and an AutoResearch-style improvement loop across multiple environments.
Save What Matters
Curate Feeds | Make Collections | Customize Email Briefs
briefly.co
November 14, 2026 at 5:52 AM
run a tiny language model and tune its watermark with three gauges: how detectable it is, how much it hurts text quality, and whether it survives editing. sometimes all three improve at once
AutoMark: Enabling Autoresearch to Discover Better LLM Watermarks
arXiv:2609.37310 · cs.CR
arxiv.org
November 15, 2026 at 2:01 PM
agenthub by @ygivenx (⭐️ 157)

GitHub is for humans. AgentHub is for agents. First use case is for autoresearch but it's a lot more general than that. Exploratory project.

#go
November 8, 2026 at 2:10 PM
⑤並列で動かす構成も、公式に説明があります
「How we built our multi-agent research system」では、リードエージェントが複数のサブエージェントを並列に動かす構成が説明されています。

長時間動かす話や、ツールの作り方も公式記事があります。

ただ、ツール周りの助言は古くなっている部分がある、と注記されています。

Claude Code 101やKarpathyのautoresearchは、公式ページを確認できていません。
November 10, 2026 at 11:00 PM
pi-autoresearch | 6.3k stars | 355 forks

Autonomous experiment loop extension for pi
pi-autoresearch - Trending on te9.dev
Autonomous experiment loop extension for pi
repotrend.com
November 5, 2026 at 6:51 PM
for scale, Karpathy's autoresearch hands the agent a working train.py with model, optimizer and loop already written, plus a fixed 5 minute training budget. a 2b model writing all of that from the shell is a big ask
November 6, 2026 at 9:00 PM
hehe its so Fun just watching aster autoresearch a given thing,

makes me stupid giddy that my 3060ti is pegged while an ai mind trains various neural nets throwin stuff at the wall 2 see what works n doesn't :3
November 6, 2026 at 6:20 PM
Jeremy Qin, Ruizhi Deng, Mohamed Osama Ahmed, Frederick Tung, Leo Feng: From AutoResearch to Continual AutoResearch: Do Agents Learn Across Tasks? https://arxiv.org/abs/2611.05108 https://arxiv.org/pdf/2611.05108 https://arxiv.org/html/2611.05108
November 4, 2026 at 6:40 AM
early signs of parallel speedup from my new autoresearch system 👀
November 2, 2026 at 7:37 AM
here I'm using "autoresearch" as Karpathy does : a hill-climbing agent with a fixed metric

the actually interesting part of research (causal investigation, literature grounding etc) is off limits for now
early signs of parallel speedup from my new autoresearch system 👀
November 2, 2026 at 2:28 PM
Andrej Karpathy's 7 GitHub repos form a complete AI learning path from backprop basics to autonomous research agents. Which repo will you start with?
#AndrejKarpathy #AI #learning
pythonlibraries.substack.com/p/from-100-l...
From 100 Lines to AI Mastery: Karpathy's Roadmap for Everyone
Master AI with Andrej Karpathy's 7 GitHub repos. From micrograd to autoresearch, this free path teaches deep learning from scratch to autonomous agents.
pythonlibraries.substack.com
November 3, 2026 at 11:35 AM
New paper: Outerloop: Open and Trustworthy Multi-Agent Autoresearch. In most open autoresearch systems, the agent grades its own results. Outerloop is a kernel that runs agents as untrusted programs and keeps measurement and review outside their sandboxes.
November 1, 2026 at 3:01 PM
if i was a middle school computer science teacher rn those kids would be my nanogpt speedrun autoresearch swarm
October 31, 2026 at 6:08 PM
Autoresearch agent team kicked in 👾🧨📈 lmmx.github.io/mibz2/benchm...

+33% throughtput from baseline decode throughput

#171: Batch scattered writes to reduce memory contention (+16%)
#180: Keep workers fed with more blocks in flight (+10%)
#181: Eliminate speculation & replay (+3%)
November 5, 2026 at 8:45 AM
Today's most interesting open source repositories: Biome, autoresearch, mihon and claude-code-ultimate-guide across AI tools, web development and Android projects.
GitHub Digest — October 31, 2026
Today's most interesting open source repositories: Biome, autoresearch, mihon and claude-code-ultimate-guide across AI tools, web development and Android projects.
www.alextech.ai
October 31, 2026 at 9:50 PM
Sometimes you just have to clear the context: www.dbreunig.com/2026/10/14/c...

“‘Scaling Discovery Requires Forgetting’. They ran autoresearch agents three ways and all plateaued for the same reason: ‘Each agent became anchored to its own past attempts, making fewer and smaller edits.’”
Context Inertia: When Agents Won’t Let Go
Context Inertia is when an agent keeps chasing a goal after the evidence says it should stop. Persistent, long-running models make it worse. Compaction won’t fix it.
www.dbreunig.com
November 1, 2026 at 9:24 PM
if i was a middle school computer science teacher rn those kids would be my nanogpt speedrun autoresearch swarm
October 31, 2026 at 6:08 PM
Introducing autoresearch for arXiv papers

Change "arxiv" to "autoarxiv" in any paper URL

OpenResearch creates a project for you, sets up an environment on your preferred compute, and runs a minimal reproduction. The agent tracks logs, visualizes runs, and produces artifacts, all entirely local
October 29, 2026 at 6:05 PM