Desh Raj's Avatar

Desh Raj

@rdesh26.bsky.social

Research Scientist @ Meta GenAI in NYC. Working on audio/speech for LLaMA. Previously: PhD @ JHU CLSP desh2608.github.io

534 Followers  |  193 Following  |  8 Posts  |  Joined: 19.11.2024  |  1.7555

Latest posts by rdesh26.bsky.social on Bluesky

UT Austin campus

UT Austin campus

Extremely excited to announce that I will be joining
@utaustin.bsky.social Computer Science in August 2025 as an Assistant Professor! πŸŽ‰

05.05.2025 20:28 β€” πŸ‘ 43    πŸ” 9    πŸ’¬ 5    πŸ“Œ 2

Congrats to UT Austin on securing an incredible human!

05.05.2025 21:28 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0
Breaking News: Johns Hopkins University, a leader in scientific research, said it would eliminate more than 2,000 workers because of federal budget cuts.

Breaking News: Johns Hopkins University, a leader in scientific research, said it would eliminate more than 2,000 workers because of federal budget cuts.

Johns Hopkins is the largest private employer in Maryland.

13.03.2025 21:48 β€” πŸ‘ 2414    πŸ” 867    πŸ’¬ 43    πŸ“Œ 68
We're not done yet | 18F

I got laid off today, with the rest of 18F.

18F was an elite federal software shop. We made gov't websites work better, more efficiently for the American people. We saved taxpayers from getting screwed over by contractors. And were fired for it.

We made this website to tell our story:
18f.org

01.03.2025 22:38 β€” πŸ‘ 93472    πŸ” 31167    πŸ’¬ 3204    πŸ“Œ 1412
Preview
The foundations of America’s prosperity are being dismantled Federal scientists warn that Americans could feel the effects of the new administration's devastating cuts for decades to come

For decades, the US government has painstakingly kept American science #1 globallyβ€”and every facet of American life has improved because of it. The internet? Flu shot? Ozempic? All grew out of federally-funded research. Now all that's being dismantled. 1/ www.technologyreview.com/2025/02/21/1...

21.02.2025 13:00 β€” πŸ‘ 3029    πŸ” 1516    πŸ’¬ 79    πŸ“Œ 116
i sensed anxiety and frustration at NeurIPS’24 – Kyunghyun Cho

Insightful post by Kyunghyun Cho on the job prospects of PhD students finishing right now. AI has seen so rapid change that the world isn't the same it used to be when they started their studies.

kyunghyuncho.me/i-sensed-anx...

24.12.2024 09:44 β€” πŸ‘ 10    πŸ” 3    πŸ’¬ 1    πŸ“Œ 0
Video thumbnail

Entropy is one of those formulas that many of us learn, swallow whole, and even use regularly without really understanding.

(E.g., where does that β€œlog” come from? Are there other possible formulas?)

Yet there's an intuitive & almost inevitable way to arrive at this expression.

09.12.2024 22:44 β€” πŸ‘ 546    πŸ” 129    πŸ’¬ 22    πŸ“Œ 12
Post image

Inventors of flow matching have released a comprehensive guide going over the math & code of flow matching!

Also covers variants like non-Euclidean & discrete flow matching.

A PyTorch library is also released with this guide!

This looks like a very good read! πŸ”₯

arxiv: arxiv.org/abs/2412.06264

10.12.2024 08:35 β€” πŸ‘ 109    πŸ” 26    πŸ’¬ 1    πŸ“Œ 1
Post image

🚨I too am on the job marketβ€ΌοΈπŸ€―

I'm searching for faculty positions/postdocs in multilingual/multicultural NLP, vision+language models, and eval for genAI!

I'll be at #NeurIPS2024 presenting our work on meta-evaluation for text-to-image faithfulness! Let's chat there!

Papers in🧡, see more: saxon.me

06.12.2024 01:44 β€” πŸ‘ 49    πŸ” 9    πŸ’¬ 1    πŸ“Œ 2
Post image

🚨 I am on the faculty job market this year 🚨
I will be presenting at #NeurIPS2024 and am happy to chat in-person or digitally!

I work on developing AI agents that can collaborate and communicate robustly with us and each other.

More at: esteng.github.io and in thread below

πŸ§΅πŸ‘‡

05.12.2024 19:00 β€” πŸ‘ 47    πŸ” 14    πŸ’¬ 2    πŸ“Œ 6

Congratulations!

05.12.2024 00:14 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image Post image

This is my first official post at Bluesky with great news :)

We got the best paper award at IEEE SLT'24! This work elegantly and straightforwardly solves contextual biasing issues with dynamic vocabulary arxiv.org/abs/2405.13344. Congrats, Yui, Yosuke, Shakeel, and Yifan!
! I'm super happy!

04.12.2024 14:15 β€” πŸ‘ 40    πŸ” 7    πŸ’¬ 2    πŸ“Œ 1
Jonathan Berant (Tel Aviv University / Google) / Towards Robust Language Model Post-training
YouTube video by Yoav Artzi Jonathan Berant (Tel Aviv University / Google) / Towards Robust Language Model Post-training

I am seriously behind uploading Learning Machines videos, but I did want to get @jonathanberant.bsky.social's out sooner than later. It's not only a great talk, it also gives a remarkably broad overview and contextualization, so it's an excellent way to ramp up on post-training
youtu.be/2AthqCX3h8U

02.12.2024 03:45 β€” πŸ‘ 53    πŸ” 12    πŸ’¬ 1    πŸ“Œ 0
Preview
CTCLoss gradient is incorrect Β· Issue #52241 Β· pytorch/pytorch πŸ› Bug Hi, While working on some CTC extensions, I noticed that torch's CTCLoss was computing incorrect gradient. At least when using CPU (I have not tested on GPU yet). I observed this problem on b...

I just learned that Torch ctc_loss calculates the wrong gradient (but when there was log_softmax before, it does not matter).

For the grad ctc_loss w.r.t. log_probs, it calculates exp(log_probs) - y, but correct would be -y. Some workaround: github.com/pytorch/pyto...

PS: First Bluesky post.

26.11.2024 23:16 β€” πŸ‘ 11    πŸ” 2    πŸ’¬ 0    πŸ“Œ 0

Can someone recommend good resources for deep dive into RLHF methods? I have a basic understanding mostly from blogs and mainstream papers (like InstructGPT), but would like to gain more insights + understand current research. Thanks in advance!

27.11.2024 02:51 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

I've started putting together a starter pack with people working on Speech Technology and Speech Science: go.bsky.app/BQ7mbkA

(Self-)nominations welcome!

19.11.2024 11:13 β€” πŸ‘ 82    πŸ” 34    πŸ’¬ 44    πŸ“Œ 3

Does everyone in your community agree on some folk knowledge that isn’t published anywhere? Put it in a paper! It’s a pretty valuable contribution

26.11.2024 22:31 β€” πŸ‘ 202    πŸ” 26    πŸ’¬ 24    πŸ“Œ 10
Preview
Research Scientist Intern, AI Research - Speech & Audio (PhD) Meta's mission is to build the future of human connection and the technology that makes it possible.

Our GenAI-Speech team at Meta is hiring RS interns for summer 2025 to work on speech, LLMs, dialog generation, and other exciting stuff! Check out the job posting here: www.metacareers.com/jobs/3841154...

22.11.2024 03:41 β€” πŸ‘ 10    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
Preview
CLSP Join the conversation

Putting together a JHU Center for Language and Speech Processing starter pack!

Please reply or DM me if you're doing research at CLSP and would like to be added - I'm still trying to find out which of us are on here so far.

go.bsky.app/JtWKca2

19.11.2024 15:37 β€” πŸ‘ 22    πŸ” 10    πŸ’¬ 2    πŸ“Œ 1

I am glad you found them useful!

19.11.2024 16:39 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Can you add me too? Thanks!

19.11.2024 14:00 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

Hi, I'm Desh! I work on speech and LLMs at Meta NY (but you may also know me from my H1B rant going viral).

19.11.2024 13:11 β€” πŸ‘ 6    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0

Thanks for this super useful starter kit. Can you add me as well?

19.11.2024 11:40 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

@rdesh26 is following 20 prominent accounts