Neil Chowdhury @ChowdhuryNeil

@TransluceAI, previously @OpenAI nchowdhury.com San Francisco Joined June 2016

Tweets

353
Followers

3K
Following

411
Likes

711

Neil Chowdhury @ChowdhuryNeil

a day ago

Docent has been really useful for understanding the outputs of my RL training runs -- glad it's finally open-source!

Transluce @TransluceAI

a day ago

Docent has been really useful for understanding the outputs of my RL training runs -- glad it's finally open-source!

1 13 66 7K 12

0 0 8 982 1

Elizabeth Barnes @BethMayBarnes

3 days ago

METR is a non-profit research organization, and we are actively fundraising! We prioritise independence and trustworthiness, which shapes both our research process and our funding options. To date, we have not accepted funding from frontier AI labs.

3 33 299 25K 35

Neil Chowdhury @ChowdhuryNeil

a week ago

they parted disclaim marinade they parted illusions

1 1 10 1K 1

Neil Chowdhury @ChowdhuryNeil

a week ago

Stop by on Thursday if you're at MIT 🙂

Transluce @TransluceAI

a week ago

Stop by on Thursday if you're at MIT 🙂

2 3 33 4K 11

0 0 10 874 0

Sayash Kapoor @sayashk

2 weeks ago

Agent benchmarks lose *most* of their resolution because we throw out the logs and only look at accuracy. I’m very excited that HAL is incorporating @TransluceAI’s Docent to analyze agent logs in depth. Peter’s thread is a simple example of the type of analysis this enables,…

Peter Kirgis @PKirgis

2 weeks ago

1 12 38 20K 20

Download Image

3 10 68 15K 26

Neil Chowdhury @ChowdhuryNeil

2 weeks ago

Very cool work -- points toward AI being one of the rare cases in tech where governments can be on the cutting edge

Xander Davies @alxndrdavies

2 weeks ago

Very cool work -- points toward AI being one of the rare cases in tech where governments can be on the cutting edge

8 63 291 52K 121

Download Image

0 0 8 1K 7

Neil Chowdhury @ChowdhuryNeil

4 weeks ago

Looks like somebody added safeguards for best-of-N jailbreaking

1 0 3 752 1

Download Image

Neil Chowdhury @ChowdhuryNeil

4 weeks ago

Very happy to see this! I hope other AI developers follow (Anthropic created a collective constitution a couple years ago, perhaps it needs updating), and that we as a community develop better rubrics & measurement tools for model behavior :)

Tyna Eloundou @ThankYourNiceAI

4 weeks ago

81 128 619 173K 129

0 0 4 662 0

Transluce @TransluceAI

a month ago

Docent, our tool for analyzing complex AI behaviors, is now in public alpha! It helps scalably answer questions about agent behavior, like “is my model reward hacking” or “where does it violate instructions.” Today, anyone can get started with just a few lines of code!