Eugene Vinitsky 🍒
@eugenevinitsky.bsky.social
Anti-cynic. Towards a weirder future. Reinforcement Learning, Autonomous Vehicles, transportation systems, the works. Asst. Prof at NYU
https://emerge-lab.github.io
https://www.admonymous.co/eugenevinitsky
https://emerge-lab.github.io
https://www.admonymous.co/eugenevinitsky
Lab meeting presentations are beginning to compete for most unhinged
November 7, 2025 at 6:25 PM
Lab meeting presentations are beginning to compete for most unhinged
This labeler is such a clever way of making spam visible. Bluesky wins yet again: bsky.app/profile/stec...
November 1, 2025 at 7:38 PM
This labeler is such a clever way of making spam visible. Bluesky wins yet again: bsky.app/profile/stec...
I'm thinking more of stuff like this:
October 26, 2025 at 6:35 PM
I'm thinking more of stuff like this:
A core value Elon-aligned folks are trying to imbue into LLMs is that racism towards Indians is acceptable and is based on some fact. Just so sickening
October 25, 2025 at 6:17 PM
A core value Elon-aligned folks are trying to imbue into LLMs is that racism towards Indians is acceptable and is based on some fact. Just so sickening
I can't find the original template I made!
October 23, 2025 at 2:29 AM
I can't find the original template I made!
What if we did a single run and declared victory
October 23, 2025 at 2:28 AM
What if we did a single run and declared victory
We're going to replace all recommendation systems heuristics with grok
*seven minutes later*
We regret to inform you that twitter is now bankrupt and takes 4 minutes to load
*seven minutes later*
We regret to inform you that twitter is now bankrupt and takes 4 minutes to load
October 18, 2025 at 3:39 PM
We're going to replace all recommendation systems heuristics with grok
*seven minutes later*
We regret to inform you that twitter is now bankrupt and takes 4 minutes to load
*seven minutes later*
We regret to inform you that twitter is now bankrupt and takes 4 minutes to load
Ah, posting about AI on bluesky?
October 17, 2025 at 5:07 AM
Ah, posting about AI on bluesky?
this is going to be a huge blow to OpenAI's valuation if it checks out
October 15, 2025 at 6:55 PM
this is going to be a huge blow to OpenAI's valuation if it checks out
I just took a picture of my local data center and unfortunately everything they've told you is real
September 26, 2025 at 2:51 AM
I just took a picture of my local data center and unfortunately everything they've told you is real
RL is a solution to some problems but probably not all problems
September 25, 2025 at 12:37 AM
RL is a solution to some problems but probably not all problems
Simulation theory is real
September 19, 2025 at 3:39 AM
Simulation theory is real
I want to print it out giant and put it everywhere
September 18, 2025 at 12:05 AM
I want to print it out giant and put it everywhere
did not quite expect these worlds to collide
September 15, 2025 at 5:12 AM
did not quite expect these worlds to collide
Coming to terms with the fact that this is my politics now
@paperbackparadise.bsky.social
@paperbackparadise.bsky.social
September 10, 2025 at 4:36 AM
Coming to terms with the fact that this is my politics now
@paperbackparadise.bsky.social
@paperbackparadise.bsky.social
A plot of the tech job-pocalypse. Wait, sorry, I misread that is that...a hiring rebound after a post-pandemic downturn?
August 27, 2025 at 10:52 PM
A plot of the tech job-pocalypse. Wait, sorry, I misread that is that...a hiring rebound after a post-pandemic downturn?
This is clearly where this is all going
August 17, 2025 at 10:00 PM
This is clearly where this is all going
Using MARL, we enabled decentralized, cooperative drones to manipulate cable-suspended loads comparably to full centralization but with better latency, scaling, and some generalization! CORL paper from Jack Zeng and Andreu Matoses Gimenez
Paper: arxiv.org/abs/2508.01522
Paper: arxiv.org/abs/2508.01522
August 11, 2025 at 3:18 PM
Using MARL, we enabled decentralized, cooperative drones to manipulate cable-suspended loads comparably to full centralization but with better latency, scaling, and some generalization! CORL paper from Jack Zeng and Andreu Matoses Gimenez
Paper: arxiv.org/abs/2508.01522
Paper: arxiv.org/abs/2508.01522
Something I like about RLC is you can write these type of negative empirical papers
August 8, 2025 at 4:16 PM
Something I like about RLC is you can write these type of negative empirical papers
Dale Schuurmans going in on RL at RLC
August 6, 2025 at 11:02 PM
Dale Schuurmans going in on RL at RLC
An RLC paper or two a day until the conference!
First off: openreview.net/forum?id=HUT...
There isn't really a canonical benchmark for partial observability in RL. This one is:
1) Written in Jax
2) Covers clear variants of partial observability
3) has checks that memory is helpful for each env
First off: openreview.net/forum?id=HUT...
There isn't really a canonical benchmark for partial observability in RL. This one is:
1) Written in Jax
2) Covers clear variants of partial observability
3) has checks that memory is helpful for each env
August 2, 2025 at 9:13 PM
An RLC paper or two a day until the conference!
First off: openreview.net/forum?id=HUT...
There isn't really a canonical benchmark for partial observability in RL. This one is:
1) Written in Jax
2) Covers clear variants of partial observability
3) has checks that memory is helpful for each env
First off: openreview.net/forum?id=HUT...
There isn't really a canonical benchmark for partial observability in RL. This one is:
1) Written in Jax
2) Covers clear variants of partial observability
3) has checks that memory is helpful for each env