Varun Gumma (@VarunGumma23)

V

Varun Gumma@VarunGumma23 · Jul 13

Peeps stop using IndicTrans2. Use the Rotary IndicTrans2 which supports longer context, and it just got better for inference. Or use Sarvam-translate (IndicTrans3) which can translate academic papers and documents extremely well.

VVarun Gumma@VarunGumma23 · Jul 13

[1/N] Thanks to @Adalat_AI , the IndicTrans2-RoPE models are now CT2 compatible. Big shoutout to their entire team for finding our work interesting and porting them, which definitely wasn't easy (huggingface.co/collections/ad…)

0

1

8

1

419

Varun Gumma Retweeted

S

Sanchit Ahuja@SanchitAhuja7 · Jun 26

A friend of mine at Adalat AI is looking for a 3-month research intern to work on legal benchmarking—running experiments across LLMs, analyzing results, and co-authoring a paper targeting an A* conference. DM me or @orgho12 on Twitter if you're interested.

2

17

5

2.0K

Varun Gumma Retweeted

A

Aditya Vashistha@imadityav · Jun 10

Thank you Sunayana and @kalikabali for inviting me to the NLP lunch! Good food and even better conversations with the NLP crew at @MSFTResearch India!

0

1

31

0

1.0K

Varun Gumma Retweeted

A

Ashwin Sankar (✈️ ACL)@_iunravel · May 31

🚨 I'm releasing Notebook 0 a week earlier than planned! 🚨 This is the soft intro to our modern Text-to-Speech (TTS) tutorial series. It’s designed to ground you in signal processing concepts & why they matter for speech! 👉 colab.research.google.com/drive/15NQ10Lu…

13

94

849

1.0K

64.0K

Varun Gumma Retweeted

A

Arnav Goel@_goel_arnav · Apr 20

✈️I will be @iclr_conf in Singapore🇸🇬 next week to present our work on attributing the cultural knowledge of a LLM to its memorization or generalization of it's pre-training corpora. Looking forward to chatting with people 🙂 #ICLR2025 📜: arxiv.org/abs/2412.20760

2

7

44

6

5.0K

V

Varun Gumma@VarunGumma23 · Apr 15

Request for participation from people from Kerala and Tamil Nadu in this super interesting study ⬇️

DDeepthi Sudharsan@deepthi_sudhan · Apr 15

🎯 𝐂𝐚𝐥𝐥 𝐟𝐨𝐫 𝐏𝐚𝐫𝐭𝐢𝐜𝐢𝐩𝐚𝐧𝐭𝐬 (𝐊𝐞𝐫𝐚𝐥𝐚 𝐨𝐫 𝐓𝐚𝐦𝐢𝐥 𝐍𝐚𝐝𝐮 𝐨𝐧𝐥𝐲) 𝐌𝐢𝐜𝐫𝐨𝐬𝐨𝐟𝐭 𝐑𝐞𝐬𝐞𝐚𝐫𝐜𝐡 𝐈𝐧𝐝𝐢𝐚 is inviting community members from Kerala or Tamil Nadu to take part in a user study to evaluate AI-generated images of cultural artifacts.

0

4

5

1

2.0K

Varun Gumma Retweeted

D

Deepthi Sudharsan@deepthi_sudhan · Apr 15

🎯 𝐂𝐚𝐥𝐥 𝐟𝐨𝐫 𝐏𝐚𝐫𝐭𝐢𝐜𝐢𝐩𝐚𝐧𝐭𝐬 (𝐊𝐞𝐫𝐚𝐥𝐚 𝐨𝐫 𝐓𝐚𝐦𝐢𝐥 𝐍𝐚𝐝𝐮 𝐨𝐧𝐥𝐲) 𝐌𝐢𝐜𝐫𝐨𝐬𝐨𝐟𝐭 𝐑𝐞𝐬𝐞𝐚𝐫𝐜𝐡 𝐈𝐧𝐝𝐢𝐚 is inviting community members from Kerala or Tamil Nadu to take part in a user study to evaluate AI-generated images of cultural artifacts.

2

7

11

2

3.0K

Varun Gumma Retweeted

S

Sanchit Ahuja@SanchitAhuja7 · Mar 12

Our team at Microsoft Research India is looking for a Research Intern for a 6 month position. The position will be on-site in BLR. You will get to work on multilingual data, modelling and evals. Please DM me with a short blurb about yourself and your CV/Resume.

8

19

209

83

16.0K

Varun Gumma Retweeted

R

Raj Dabre@prajdabre · Feb 17

TFW people with multiple A* papers are struggling to get PhD admits in the USA.

8

2

174

12

12.0K

Varun Gumma Retweeted

N

Nagarajan Natarajan@naga86 · Jan 13

Microsoft Research India is excited to announce applications are open for our Research Fellow program (deadline 15th Feb 2025). Details of the program and the application are here: 🔗 Research Fellow program: aka.ms/msrirf @MSFTResearch

3

48

141

88

38.0K

V

Varun Gumma@VarunGumma23 · Jan 14

MSR India is accepting applications for the 2025 Research Fellow program

NNagarajan Natarajan@naga86 · Jan 13

Microsoft Research India is excited to announce applications are open for our Research Fellow program (deadline 15th Feb 2025). Details of the program and the application are here: 🔗 Research Fellow program: aka.ms/msrirf @MSFTResearch

1

12

104

20

12.0K

Varun Gumma Retweeted

R

Raj Dabre@prajdabre · Jan 13

Overall, this paper was responsible for a major shift in my research work because, let's face it, compact models FTW. It greatly benefitted @VarunGumma23's master thesis and prolly will be his PhD focus. If you have not read this paper, then please do. It's a life changer and…

1

2

5

2

1.0K

V

Varun Gumma@VarunGumma23 · Nov 24

After consulting @VarunGumma23 and @pranjalchitale, in my free time, I have replicated their work, and I am releasing the 4 IndicTrans2 models which use RoPE instead of learned PEs. This effectively enables IndicTrans2 models to handle long context. You can now process longer…

VVarun Gumma@VarunGumma23 · Oct 19

A v2 of our work on Interchangeability of Positional Embeddings (PEs) for NMT System is out: arxiv.org/abs/2408.11382. In this paper, we mainly pitch towards inducing Document-Level abilities in NMT systems and show that you can efficiently switch out PEs to achieve this.

3

2

36

12

4.0K

Varun Gumma Retweeted

R

Raj Dabre@prajdabre · Nov 14

The #EMNLP2024 social event feels like a disaster. Terrible crowd management! I'm hungry! Feed me!

5

3

82

5

10.0K

Varun Gumma Retweeted

H

Hamna_Abid@ham_na_abi · Nov 2

Gen AI models can generate compelling stories, but do they truly reflect all cultures? 🎉Excited to share our work, ‘Kahani’—a visual storytelling pipeline designed to create culturally grounded visual stories for non-Western cultures! 1/6

2

16

9

8.0K