Tanmoy Chakraborty

@Etihad @EtihadHelp I was charged TWICE for a seat booking on May 5 (PNR: 7****4). The duplicate charge was supposed to be refunded. It's been over 2 MONTHS and I've followed up numerous times with zero resolution. Every follow-up gets the same line: "escalated to the backend team." No refund, no update, no accountability. @gcaauae @ADMediaOffice. This repeated runaround is customer harassment. Refund my money immediately. @Economyae @consumerforum_

Among all the podcasts I've been on, this one is especially close to my heart. It covers my childhood journey and explores many (AI-related) questions that people rarely ask. youtu.be/IvYRL_u5-ho?si…

@EtihadHelp I followed you now.

@Tanmoy_Chak I can see your DM but we are unable to reply to your DM due to a profile restriction, kindly remove the restriction that would allow profiles you do not follow to DM you, or follow us back. Thanks. *Aya

ACL season is here! Sharing our papers once again. Unfortunately, I wont be able to attend ACL this year, but I hope you enjoy reading our work. @aclmeeting #ACL2026

Prof @Tanmoy_Chak delivered a keynote at the DHOW-MiLLA: Joint Workshop on Diffusion of Harmful Content on Online Web and Countering #Misinformation in the Age of #LLMs and #Agents, held during ACM #TheWebConf2026

Last week, Prof @Tanmoy_Chak was invited as a distinguished speaker at #ETEdgeAISummit2026. He highlighted that "Success in #AI will not come from building the biggest model, but from building the smartest and most relevant one for the problem being solved." 🚀

@Tanmoy_Chak #indiagovt keeps announcing '#India will lead the #AI revolution,' but #IIT professors literally can't get a GPU to do research- India building AI for Twitter PR or for actual ground-level innovation? Full Episode link-youtu.be/5q-ePMMYGNc?si… #indiaai #indiatech

For years, I have been arguing that SLMs are the future and that more power comes when SLMs are allowed to "collaborate." Our recent paper, published in #TMLR @TmlrOrg, provides strong empirical evidence for this hypothesis. We introduce INFORM -- an interpretability framework that treats model coordination as an explicit, analyzable computation, helping us understand how collaborating models work together to solve complex tasks. Paper, code and video here👇

We are pleased to welcome @Tanmoy_Chak, Associate Professor, Dept of Electrical Engineering and Yardi School of Artificial Intelligence at Indian Institute of Technology Delhi (@iitdelhi) as a distinguished speaker at the ET Edge AI Summit & Awards 2026 as one of the voices shaping the future of AI in India.​ ET Edge AI Summit & Awards 2026 is bringing together AI leaders, policymakers, innovators, and industry stakeholders, it creates space for conversations that connect enterprise transformation with broader societal impact, while keeping the focus on building intelligence that is responsible, competitive, and inclusive. ​ 23rd June 2026 | New Delhi​ Register Now: ai-summit.et-edge.com/register-now/?… #ArchitectTheIntelligence #ETEdgeAISummit2026 #AIAtScale #BuiltByIntelligence

What makes a good teacher? On-policy distillation has spent the year reinventing loss functions to fix problems that come from one source: the teacher doesn't know the student. New article on why every popular OPD loss has an unbounded advantage and why the fix isn't another loss. Read the full article at - x.com/ayans007/statu… w/ @Tanmoy_Chak @lcs2lab #LLM #KnowledgeDistillation

Markovian ODE-guided scoring can assess the quality of offline reasoning traces in language models Arghodeep Nandi, Ojasva Saxena, Tanmoy Chakraborty arxiv.org/abs/2603.01580 [𝚌𝚜.𝙲𝙻]

A reporting checklist for large language models in behavioural science dlvr.it/TSxtMM

Check out our recent work published in Nature Human Behaviour on responsible LLM checklist: nature.com/articles/s4156… @NatureHumBehav

This is huge. Our PEFT method, MonteCLoRA, has been merged with @huggingface. Do use it. Believe me. It is much much better than LoRA in terms of efficiency and stability.

Time to celebrate acceptance of two papers in 𝐈𝐂𝐌𝐋'26, including one 𝐒𝐩𝐨𝐭𝐥𝐢𝐠𝐡𝐭 (top 2.2%) 🎉 👉 Polaris: Coupled Orbital Polar Embeddings for Hierarchical Concept Learning 📔 arxiv.org/pdf/2605.00265 ✨ Introduces Polaris -- a hyperspherical embedding framework that decouples semantics from hierarchy using orbital geometry, uncertainty-aware learning, and efficient retrieval. 👉 Linguistic Properties and Model Scale in Brain Encoding: From Small to Compressed Language Models (#𝐒𝐩𝐨𝐭𝐥𝐢𝐠𝐡𝐭) 📔 arxiv.org/pdf/2602.07547 ✨ Shows that compact ~3B models can match much larger LLMs in brain alignment, with robustness even under compression. Grateful to all collaborators and students for the amazing work! 🚀 @icmlconf @lcs2lab @iitdelhi #ICML26