Leon Lang
@leon-lang.bsky.social
PhD Candidate at the University of Amsterdam. AI Alignment and safety research. Formerly multivariate information theory and equivariant deep learning. Masters degrees in both maths and AI. https://langleon.github.io/
Reposted by Leon Lang
⚠️ The Perils of Optimizing Learned Reward Functions: Low Training Error Does Not Guarantee Low Regret
By Lukas Fluri*, @leon-lang.bsky.social *, Alessandro Abate, Patrick Forré, David Krueger, Joar Skalse
📜 arxiv.org/abs/2406.15753
🧵6 / 8
By Lukas Fluri*, @leon-lang.bsky.social *, Alessandro Abate, Patrick Forré, David Krueger, Joar Skalse
📜 arxiv.org/abs/2406.15753
🧵6 / 8
May 6, 2025 at 2:53 PM
⚠️ The Perils of Optimizing Learned Reward Functions: Low Training Error Does Not Guarantee Low Regret
By Lukas Fluri*, @leon-lang.bsky.social *, Alessandro Abate, Patrick Forré, David Krueger, Joar Skalse
📜 arxiv.org/abs/2406.15753
🧵6 / 8
By Lukas Fluri*, @leon-lang.bsky.social *, Alessandro Abate, Patrick Forré, David Krueger, Joar Skalse
📜 arxiv.org/abs/2406.15753
🧵6 / 8
Brief paper announcement (longer thread might follow):
In our new paper "Modeling Human Beliefs about AI behavior for Scalable Oversight", I propose to model a human evaluator's beliefs to better interpret the feedback, which might help for scalable oversight. (1/4)
In our new paper "Modeling Human Beliefs about AI behavior for Scalable Oversight", I propose to model a human evaluator's beliefs to better interpret the feedback, which might help for scalable oversight. (1/4)
March 3, 2025 at 3:44 PM
Brief paper announcement (longer thread might follow):
In our new paper "Modeling Human Beliefs about AI behavior for Scalable Oversight", I propose to model a human evaluator's beliefs to better interpret the feedback, which might help for scalable oversight. (1/4)
In our new paper "Modeling Human Beliefs about AI behavior for Scalable Oversight", I propose to model a human evaluator's beliefs to better interpret the feedback, which might help for scalable oversight. (1/4)
Reposted by Leon Lang
If you are attending #NeurIPS2024🇨🇦, make sure to check out AMLab's 11 accepted papers ...and to have a chat with our members there! 👩🔬🍻☕
Submissions include generative modelling, AI4Science, geometric deep learning, reinforcement learning and early exiting. See the thread for the full list!
🧵1 / 12
Submissions include generative modelling, AI4Science, geometric deep learning, reinforcement learning and early exiting. See the thread for the full list!
🧵1 / 12
December 9, 2024 at 1:24 PM
If you are attending #NeurIPS2024🇨🇦, make sure to check out AMLab's 11 accepted papers ...and to have a chat with our members there! 👩🔬🍻☕
Submissions include generative modelling, AI4Science, geometric deep learning, reinforcement learning and early exiting. See the thread for the full list!
🧵1 / 12
Submissions include generative modelling, AI4Science, geometric deep learning, reinforcement learning and early exiting. See the thread for the full list!
🧵1 / 12
Reposted by Leon Lang
First UAI conference in Latin America!! 🔥🔥🔥
North America and Europe you are nice, but sometimes I also want to visit somewhere else 😅
North America and Europe you are nice, but sometimes I also want to visit somewhere else 😅
The 41st Conference on #Uncertainty in #AI will be held in Rio de Janeiro 🇧🇷, July 21-25!
The CfP is out 👉 www.auai.org/uai2025/call...
🚨 Feb 10: Paper submission
🗣️ Apr 3-10: rebuttal period
🎉/💀 May 6: Author notification
#UAI2025 #ML #stats #learning #reasoning #uncertainty
The CfP is out 👉 www.auai.org/uai2025/call...
🚨 Feb 10: Paper submission
🗣️ Apr 3-10: rebuttal period
🎉/💀 May 6: Author notification
#UAI2025 #ML #stats #learning #reasoning #uncertainty
December 3, 2024 at 5:30 PM
First UAI conference in Latin America!! 🔥🔥🔥
North America and Europe you are nice, but sometimes I also want to visit somewhere else 😅
North America and Europe you are nice, but sometimes I also want to visit somewhere else 😅
I just completed "Historian Hysteria" - Day 1 - Advent of Code 2024 #AdventOfCode adventofcode.com/2024/day/1
December 1, 2024 at 5:19 PM
I just completed "Historian Hysteria" - Day 1 - Advent of Code 2024 #AdventOfCode adventofcode.com/2024/day/1
I notice more “big” accounts here that follow a lot of people. The same accounts follow almost no one on twitter. Is this motivated by a difference in the algorithms of these platforms?
December 1, 2024 at 11:04 AM
I notice more “big” accounts here that follow a lot of people. The same accounts follow almost no one on twitter. Is this motivated by a difference in the algorithms of these platforms?
Reposted by Leon Lang
Yet another safety researcher has left OpenAI.
Rosie Campbell says she has been “unsettled by some of the shifts over the last ~year, and the loss of so many people who shaped our culture”.
She says she “can’t see a place” for her to continue her work internally.
Rosie Campbell says she has been “unsettled by some of the shifts over the last ~year, and the loss of so many people who shaped our culture”.
She says she “can’t see a place” for her to continue her work internally.
December 1, 2024 at 12:48 AM
Yet another safety researcher has left OpenAI.
Rosie Campbell says she has been “unsettled by some of the shifts over the last ~year, and the loss of so many people who shaped our culture”.
She says she “can’t see a place” for her to continue her work internally.
Rosie Campbell says she has been “unsettled by some of the shifts over the last ~year, and the loss of so many people who shaped our culture”.
She says she “can’t see a place” for her to continue her work internally.
Reposted by Leon Lang
We are taking on a mission to track progress in AI capabilities over time.
Very proud of our team!
Very proud of our team!
We've just launched our AI Benchmarking Hub!
This is a new platform for rigorous, independent evaluations of AI model capabilities, featuring interactive visualizations and in-depth analysis. (1/8)
epoch.ai/blog/introdu...
This is a new platform for rigorous, independent evaluations of AI model capabilities, featuring interactive visualizations and in-depth analysis. (1/8)
epoch.ai/blog/introdu...
November 27, 2024 at 8:38 PM
We are taking on a mission to track progress in AI capabilities over time.
Very proud of our team!
Very proud of our team!
Reposted by Leon Lang
Hey hey,
I am around in the Bay area for the next few weeks. Bay area folks hit me up if you want to meet up for coffee/ vegan food in and around SF ☕🌯 🥟
Got a major weather upgrade☀️ from Amsterdam's insanity last week 🌀🌩️
I am around in the Bay area for the next few weeks. Bay area folks hit me up if you want to meet up for coffee/ vegan food in and around SF ☕🌯 🥟
Got a major weather upgrade☀️ from Amsterdam's insanity last week 🌀🌩️
November 24, 2024 at 9:54 PM
Hey hey,
I am around in the Bay area for the next few weeks. Bay area folks hit me up if you want to meet up for coffee/ vegan food in and around SF ☕🌯 🥟
Got a major weather upgrade☀️ from Amsterdam's insanity last week 🌀🌩️
I am around in the Bay area for the next few weeks. Bay area folks hit me up if you want to meet up for coffee/ vegan food in and around SF ☕🌯 🥟
Got a major weather upgrade☀️ from Amsterdam's insanity last week 🌀🌩️
Thanks for highlighting our paper! :)
Great paper. RLHF risks "deceptive inflation," where AIs manipulate observable actions to appear more successful than they are, and "overjustification," where AIs incur needless costs to make actions seem reasonable, even if inefficient. arxiv.org/abs/2402.17747
November 25, 2024 at 7:33 PM
Thanks for highlighting our paper! :)
Reposted by Leon Lang
Meet our Lab's members: staff, postdocs and PhD students! :)
With this starter pack you can easily connect with us and keep up to date with all the member's research and news 🦋
go.bsky.app/8EGigUy
With this starter pack you can easily connect with us and keep up to date with all the member's research and news 🦋
go.bsky.app/8EGigUy
November 21, 2024 at 9:22 PM
Meet our Lab's members: staff, postdocs and PhD students! :)
With this starter pack you can easily connect with us and keep up to date with all the member's research and news 🦋
go.bsky.app/8EGigUy
With this starter pack you can easily connect with us and keep up to date with all the member's research and news 🦋
go.bsky.app/8EGigUy
Reposted by Leon Lang
MIT undergrads from families earning less than $200K will pay no tuition fees from 2025, and undergrads from families earning less than $100K will have everything covered, including housing, dining, and a personal allowance.
news.mit.edu/2024/mit-tui...
news.mit.edu/2024/mit-tui...
November 20, 2024 at 8:22 PM
MIT undergrads from families earning less than $200K will pay no tuition fees from 2025, and undergrads from families earning less than $100K will have everything covered, including housing, dining, and a personal allowance.
news.mit.edu/2024/mit-tui...
news.mit.edu/2024/mit-tui...
Reposted by Leon Lang
Does anyone understand why it’s so easy to clone twitter with no IP issues?
It’s hard to understand qualitative legal thresholds, but the UI looking ~exactly the same both here and on threads intuitively seems like the kind of thing that could violate a copyright if twitter had pursued one
It’s hard to understand qualitative legal thresholds, but the UI looking ~exactly the same both here and on threads intuitively seems like the kind of thing that could violate a copyright if twitter had pursued one
November 20, 2024 at 9:41 PM
Does anyone understand why it’s so easy to clone twitter with no IP issues?
It’s hard to understand qualitative legal thresholds, but the UI looking ~exactly the same both here and on threads intuitively seems like the kind of thing that could violate a copyright if twitter had pursued one
It’s hard to understand qualitative legal thresholds, but the UI looking ~exactly the same both here and on threads intuitively seems like the kind of thing that could violate a copyright if twitter had pursued one
Reposted by Leon Lang
Hi everyone! This is AMLab :)
Looking forward to share our research here on 🦋 !
Looking forward to share our research here on 🦋 !
November 19, 2024 at 4:00 PM
Hi everyone! This is AMLab :)
Looking forward to share our research here on 🦋 !
Looking forward to share our research here on 🦋 !
Reposted by Leon Lang
the success of Bluesky with an absolutely tiny team does sorta prove Musk had a point about Twitter having way too many staff
November 19, 2024 at 3:00 PM
the success of Bluesky with an absolutely tiny team does sorta prove Musk had a point about Twitter having way too many staff