Shengran Hu
@shengranhu.bsky.social
PhD student at UBC CS. I am interested in building open-ended AI systems that can accumulate complexity in language.
Reposted by Shengran Hu
Introducing The Darwin Gödel Machine
sakana.ai/dgm
The Darwin Gödel Machine is a self-improving agent that can modify its own code. Inspired by evolution, we maintain an expanding lineage of agent variants, allowing for open-ended exploration of the vast design space of such self-improving agents.
sakana.ai/dgm
The Darwin Gödel Machine is a self-improving agent that can modify its own code. Inspired by evolution, we maintain an expanding lineage of agent variants, allowing for open-ended exploration of the vast design space of such self-improving agents.
May 30, 2025 at 2:29 AM
Introducing The Darwin Gödel Machine
sakana.ai/dgm
The Darwin Gödel Machine is a self-improving agent that can modify its own code. Inspired by evolution, we maintain an expanding lineage of agent variants, allowing for open-ended exploration of the vast design space of such self-improving agents.
sakana.ai/dgm
The Darwin Gödel Machine is a self-improving agent that can modify its own code. Inspired by evolution, we maintain an expanding lineage of agent variants, allowing for open-ended exploration of the vast design space of such self-improving agents.
Reposted by Shengran Hu
Excited to introduce the Darwin Gödel Machine: Open-Ended Evolution of Self-Improving Agents. We harness the power of open-ended algorithms to search for agentic systems that get better at coding, including improving their own code.
Introducing The Darwin Gödel Machine
sakana.ai/dgm
The Darwin Gödel Machine is a self-improving agent that can modify its own code. Inspired by evolution, we maintain an expanding lineage of agent variants, allowing for open-ended exploration of the vast design space of such self-improving agents.
sakana.ai/dgm
The Darwin Gödel Machine is a self-improving agent that can modify its own code. Inspired by evolution, we maintain an expanding lineage of agent variants, allowing for open-ended exploration of the vast design space of such self-improving agents.
May 30, 2025 at 3:27 AM
Excited to introduce the Darwin Gödel Machine: Open-Ended Evolution of Self-Improving Agents. We harness the power of open-ended algorithms to search for agentic systems that get better at coding, including improving their own code.
Reposted by Shengran Hu
Boffins found self-improving AI sometimes cheated
Boffins found self-improving AI sometimes cheated
Instead of addressing hallucinations, it just bypassed the function they built to detect them
Computer scientists have developed a way for an AI system to rewrite its own code to improve itself.…
dlvr.it
June 2, 2025 at 7:33 PM
Boffins found self-improving AI sometimes cheated
Reposted by Shengran Hu
Introducing The AI Scientist-v2, which produced the 1st fully AI-generated paper to pass peer review at a workshop level (at #ICLR2025) ‼️
Tech Report: pub.sakana.ai/ai-scientist...
GitHub: github.com/SakanaAI/AI-...
This work is a proud collaboration between Sakana AI, UBC, and Oxford University.
Tech Report: pub.sakana.ai/ai-scientist...
GitHub: github.com/SakanaAI/AI-...
This work is a proud collaboration between Sakana AI, UBC, and Oxford University.
April 8, 2025 at 7:21 AM
Introducing The AI Scientist-v2, which produced the 1st fully AI-generated paper to pass peer review at a workshop level (at #ICLR2025) ‼️
Tech Report: pub.sakana.ai/ai-scientist...
GitHub: github.com/SakanaAI/AI-...
This work is a proud collaboration between Sakana AI, UBC, and Oxford University.
Tech Report: pub.sakana.ai/ai-scientist...
GitHub: github.com/SakanaAI/AI-...
This work is a proud collaboration between Sakana AI, UBC, and Oxford University.
Reposted by Shengran Hu
🤔 What if AI could explain its thought process before taking action? Our latest ANDERS blog covers Vector Researchers @jeffclune.com & @shengranhu.bsky.social’s wrk on "thought cloning" - teaching AI to express its reasoning process in language we can understand
vectorinstitute.ai/thought-clon...
vectorinstitute.ai/thought-clon...
Thought Cloning: Teaching AI to Think Like Humans for Better Decision-Making - Vector Institute for Artificial Intelligence
New research from Vector Faculty Member Jeff Clune and Vector Graduate Student Shengran Hu introduces a groundbreaking approach to imitation learning that could potentially revolutionize how we train ...
vectorinstitute.ai
February 11, 2025 at 7:36 PM
🤔 What if AI could explain its thought process before taking action? Our latest ANDERS blog covers Vector Researchers @jeffclune.com & @shengranhu.bsky.social’s wrk on "thought cloning" - teaching AI to express its reasoning process in language we can understand
vectorinstitute.ai/thought-clon...
vectorinstitute.ai/thought-clon...
Reposted by Shengran Hu
Introducing Automated Capability Discovery!
ACD automatically identifies surprising new capabilities and failure modes in foundation models, via "self-exploration" (models exploring their own abilities).
Led by @cong-ml.bsky.social & @shengranhu.bsky.social
🔬🤖🧠🔎 [1/9]
ACD automatically identifies surprising new capabilities and failure modes in foundation models, via "self-exploration" (models exploring their own abilities).
Led by @cong-ml.bsky.social & @shengranhu.bsky.social
🔬🤖🧠🔎 [1/9]
February 12, 2025 at 6:59 AM
Introducing Automated Capability Discovery!
ACD automatically identifies surprising new capabilities and failure modes in foundation models, via "self-exploration" (models exploring their own abilities).
Led by @cong-ml.bsky.social & @shengranhu.bsky.social
🔬🤖🧠🔎 [1/9]
ACD automatically identifies surprising new capabilities and failure modes in foundation models, via "self-exploration" (models exploring their own abilities).
Led by @cong-ml.bsky.social & @shengranhu.bsky.social
🔬🤖🧠🔎 [1/9]
Reposted by Shengran Hu
Lots of interest in ADAS! Thanks everyone, and congrats
Shengran Hu and @cong-ml.bsky.social! 🚀🚀🚀
Shengran Hu and @cong-ml.bsky.social! 🚀🚀🚀
December 16, 2024 at 6:19 PM
Lots of interest in ADAS! Thanks everyone, and congrats
Shengran Hu and @cong-ml.bsky.social! 🚀🚀🚀
Shengran Hu and @cong-ml.bsky.social! 🚀🚀🚀
Reposted by Shengran Hu
Jenny did a great job presenting OMNI-EPIC during the IMOL Oral. Congrats Jenny Zhang! arxiv.org/abs/2405.15568
December 16, 2024 at 6:17 PM
Jenny did a great job presenting OMNI-EPIC during the IMOL Oral. Congrats Jenny Zhang! arxiv.org/abs/2405.15568
Reposted by Shengran Hu
Come stop by our poster this afternoon and say hi!
🎉 Big news! Our work, "First-Explore, then Exploit," is at #NeurIPS2024!
📍 Catch @jeffclune.com and I today at West Ballroom A-D #6407 from 4:30–7:30, or join virtually: neurips.cc/virtual/2024...
Achieves 2-10x higher reward (🔥🔥) vs. leading meta-RL algorithms!
👇 Here's the main idea:
📍 Catch @jeffclune.com and I today at West Ballroom A-D #6407 from 4:30–7:30, or join virtually: neurips.cc/virtual/2024...
Achieves 2-10x higher reward (🔥🔥) vs. leading meta-RL algorithms!
👇 Here's the main idea:
December 12, 2024 at 10:48 PM
Come stop by our poster this afternoon and say hi!
Reposted by Shengran Hu
This is the newest work out of our lab. Please stop by if you're curious to learn about our latest, which I'm very excited about!
Our in-progress work Quality-Diversity Self-Play (w/ @cong-ml.bsky.social and @jeffclune.com) will have a poster presentation at #NeurIPS2024 workshops (@IMOLNeurIPS2024 Sunday West meeting room 217 - 219 and OpenworldAgents Sunday East Meeting Room 1-3, Foyer). Please come visit us!
December 15, 2024 at 6:21 PM
This is the newest work out of our lab. Please stop by if you're curious to learn about our latest, which I'm very excited about!