Rebecca Olsen retweeted
We believe agents can produce novel ideas but they need to be pushed out of distribution. On this paper, we explore some techniques to encourage this behavior with promising results.
Can search 🔍 algorithms help autoresearch agents come up with original ideas that actually work? 🤖
We ran 6 of them on our framework, Heuresis (NeurIPS 2026): 3,222 ideas, 0 completely original.
But the right search algorithm gets closer. On every task, search beat the greedy baseline at producing minor similarity ideas, the most novel rating any idea got:
• nanoGPT: Curiosity 20, Greedy 0
• On-Policy RL: MAP-Elites 12, Greedy 0
• Model Unlearning: OMNI 29, Greedy 12
The catch is that only 1 of these ideas reached its task's top 10. Agents can't yet tell which novel ideas are worth refining.
We built Heuresis to enable research into autoresearch agents.
1. Abstractions enable plug-and-play search algorithms
2. Skills analyze runs across quality, diversity and novelty
3. HackerJudge catches reward hacking
So, can autoresearch agents come up with fundamentally new research ideas?
Not yet. But Heuresis makes it easy to build and test the search algorithms that might!
The same loop can also generate rich data for training better autoresearch models 🤓
📜 Paper: arxiv.org/abs/2606.25198
💻 Code: github.com/a-antoniades/Heur…
Rebecca Olsen retweeted
Spacecraft touches an asteroid—and the surface erupts! OSIRIS-REx’s daring Touch-and-Go maneuver at Bennu revealed how loose regolith reacts in microgravity while TAGSAM collected precious samples. incredible engineering feat brought pieces of an ancient asteroid back to Earth
Rebecca Olsen retweeted
#HurricaneTrudy gradually weakened into a tropical storm off the west-central coast of Mexico in the days leading up to this #TimeLapseTuesday. @NOAA’s #GOESWest (#GOES18) 🛰️ continuously captured this infrared imagery of the cyclone as it neared the end of its life cycle during October 16–19.
For the latest updates: nhc.noaa.gov/
Rebecca Olsen retweeted
Global Optimality of Pose Graph Optimization
Shuan Ding, Ruishuang Chen, Mengfan Cao, Zaiyue Yang
tl;dr: global optimality of PGO from its own problem structure
arxiv.org/abs/2610.14655
Rebecca Olsen retweeted
🚀 IT’S ALMOST TIME FOR #MMXLIFTOFF! 🛰️
The journey towards the Martian moon, Phobos, is finally beginning. We have come this far with the support of so many people, and now we take the first step in this long voyage!
#MMXLIFTOFF: 10.20 at 04:41:03 JST 🚀
Rebecca Olsen retweeted
SAE features don't simply add up
We introduce SAE Additivity Residuals (SAR), quantifying the behavioral variation that individual features cannot explain. Across 18 settings, learned coordinates beat random rotations. Pairwise terms recover much of what's missing — boosting steering rank fidelity from 0.59 to 0.78.
Rebecca Olsen retweeted
Testing my Starlink V5 magnetic mount.
I don't fully trust these magnets, especially on the uneven surface of the air conditioning unit, so I'll secure the mount with a steel safety cable as an extra precaution.
Surrounding buildings partially obstruct the view of the sky, and the weather is wet and foggy today. Despite these less-than-ideal conditions, the V5 is performing very well.
This will serve as my backup internet connection in case the ground-based infrastructure is damaged or destroyed.
Rebecca Olsen retweeted
Talk about diving into your work! ✈️🔧
Keeping NASA’s X-59 flight-ready sometimes means going headfirst into its engine inlet. See how technicians use “duct diving” to inspect critical components of our quiet supersonic aircraft.
go.nasa.gov/4dJiosH
Rebecca Olsen retweeted
🚀 3 days until #MMXLIFTOFF 🛰️
Today’s countdown comes from #JAXA ISAS Director General Fujimoto Masaki! MMX embodies the challenge and aspirations of so many people 😊 We are now just THREE DAYS FROM LAUNCH!
#MMXLIFTOFF: 10.20 at 04:41:03 JST 🚀
Rebecca Olsen retweeted
7:39pm CT
We were asked to put earplugs on. Raptor engines igniting – at McGregor, TX
📣 CALL FOR DELEGATES
SGAC is sponsoring 100 young Africans to attend SG[Nigeria] x ASEIC 2026 in Abuja (Nov 23-27).
Submit your CV + a 500-word Impact Pitch by Oct 23, 2026 (23:59 WAT).
Apply: spacegeneration.org/sgnigeri…
#SGNigeria #ASEIC2026 #AfricaInSpace
Rebecca Olsen retweeted
NVIDIA just released its DFlash 2 draft model for Muse-Spark-Open on Hugging Face
Uses block diffusion for parallel speculative decoding, enabling lossless acceleration for the target model
Rebecca Olsen retweeted
.
Corporate sustainability assessment covers 179 companies with data-center operations, as of 1 April 2026
About two-thirds of that sample report a net-zero commitment.
The bars split those pledges into two groups: targets that are SBTi-verified or otherwise science-based, and targets that are not. The remainder have no net-zero target at all.
Asia-Pacific is the largest regional slice (82 firms). Europe follows with 50, the United States and Canada with 36. Latin America and the Middle East are too small to read as regional patterns. Africa is omitted because the sample is only two companies.The headline share is a commitment count, not a measure of delivery.
Source: S&P Global Energy
Rebecca Olsen retweeted
An LLM robot agent’s success rate goes up as it practices. Did it learn anything, or did it just try more often? And do the skills and memories it saves still help on new tasks, in a new simulator, on a real robot?
We built SEAR to answer both. To our knowledge, it’s the first large scale controlled study of self-evolving robot agents. The paper is now on arXiv 📄
The agent improves a robot in one of three ways: it writes code that becomes the policy, it acts live in the control loop, or it trains its own VLA.
7 frontier LLMs, 330 sim tasks, 5 real-robot tasks, 4+ years of agent runtime, with a matched control for every comparison and a sealed judge on unseen scenes.
A few findings surprised us 🧵 1/7
arxiv.org/abs/2610.18792
Rebecca Olsen retweeted
Simulations Of The Evolving Ozone Layer: Implications For The Early Earth, Exoplanets, And The Search For Life
astrobiology.com/2026/10/15/… #astrobiology #atmosphere #climate

































