Devvrit
@Devvrit_Khatri
Followers
230
Following
2K
Media
11
Statuses
119
GradStudent@UTCompSci. Large Scale ML - Scalability and Efficiency
Joined December 2019
RT @dvsaisurya: 📢 Thrilled to share our new paper, LASER: Attention with Exponential Transformation, accepted at ICML2025, work done at Goo….
0
4
0
RT @JaydevTonde: Google DeepMind just launched #Gemma3n! 🚀 They've released E2B & E4B, with more sizes coming for your specific hardware ne….
0
2
0
As I've said previously, one of the best research teams to work with. Please don't loose the opportunity. :).
We are hiring Research Scientists for our Machine Learning and Optimization team at Google DeepMind Bangalore. If you're passionate about cutting-edge AI research and building efficient, elastic, customized, and safe LLMs, we'd love to hear from you. We are looking for.
1
0
13
RT @Tim_Dettmers: MatFormers are very powerful alternatives to transformers. Similar to a regular transformer, but after training, you can….
0
45
0
Gains for free, as promised 🙂.
Pocket powerhouse admist I/O awesomeness!. Gemma 3n E4B & E2B are insane models, optimized for on-device while rivaling frontier models. It's a 🪆Matryoshka Transformer (MatFormer)🪆: Natively elastic b/w 4B & 2B pareto-optimally!. ⭐️: free models with ZERO training cost!. 🧵👇
0
0
10
Gain for free yet again? Now in quantization! It’s amazing to see int2 performing so good! Amazing work.
Announcing Matryoshka Quantization! A single Transformer can now be served at any integer precision!! In addition, our (sliced) int2 models outperform the baseline by 10%. Work co-led w/ @puranjay1412, in colab w/ @JeffDean, @jainprateek_ & @adityakusupati. 1/7
0
0
6
Check out the NeurIPS spotlight work on the time reversed training of LLMs, by @varunyer . The use cases ranging from jailbreaking, reranking, to general performance improvements are impressive. @varunyer is applying for PhD - highly recommend having him in your lab!.
✨ Presenting Time-Reversed LLMs, a class of LLMs capable of generating feedback on the responses of forward LLMs. (NeurIPS Spotlight) 📢.📄 Arxiv: Read on for more details!. 🧵(1/n).
0
1
7
Check out the amazing work by @gaganjain1582 and co. on adaptive expert usage. Also he’s on job market. Grab him before someone else does!.
It's todayyy!.If you are at NeurIPS, come to our poster (#2502) in the East Exhibit Hall at 4.30 PM. I'll also be at the Google DeepMind booth from 3 to 4 PM. #NeurIPS #NeurIPS2024.
0
1
10
If you want free performance gains (for real, no catch here) AND elastic inference, please check our poster and come talk to us. We promise you’ll benefit using MatFormer training methodology if training multiple models, treat on us if you don’t 🙂.
@adityakusupati, @Devvrit_Khatri and I will be at poster #2507 in East Exhibit Hall A-C at #NeurIPS2024 chatting about MatFormer and elastic models today at 4.30pm!.
1
1
8
Just boarded my flight to Vancouver for #NeurIPS2024 ! Excited to meet everyone and chat about efficiency and scalability in LLMs. Hit me up if you’re around! 😁.
0
0
19
💯.
Prateek is the best combination of raw smarts, creativity, and kindness . - his team's success in pushing the frontier of AI reflects it! I've been his (biggest) fan for years - I can't recommend joining his team enough!.
0
0
4
I’ve been working with MLO for the past couple of years and it’s the best research lab that I’ve worked at. The team is super helpful and friendly, and the work is highly, highly impactful!. Would strongly recommend applying 💻.
Excited to share that the Machine Learning and Optimization team at @GoogleDeepMind India is hiring Research Scientists and Research Engineers! If you're passionate about cutting-edge AI research and building efficient, elastic, and safe LLMs, we'd love to hear from you. Check.
0
0
19
RT @gaganjain1582: Your models about to get a MoNE-y makeover! 💸 We're thrilled to unveil MoNE (Mixture of Nested Experts), a game-changer….
0
15
0
RT @gaganjain1582: 👀A sneak-peek at our recent work accepted at #ECCV2024 @eccvconf ! .📜: Want a scalable and robu….
0
24
0
Average bangalore traffic. And of course we’d be an hour in this, not 10 mins.
0
0
3
Shoutout to @RedHornBrew for surprising me with this sweet gesture. Highly recommend their vanilla lavender iced latte!
1
0
3