Tanmoy Chakraborty

🌟 𝐀 𝐍𝐞𝐰 T𝐞𝐱𝐭𝐛𝐨𝐨𝐤 -- 𝐈𝐧𝐭𝐫𝐨𝐝𝐮𝐜𝐭𝐢𝐨𝐧 𝐭𝐨 𝐋𝐚𝐫𝐠𝐞 𝐋𝐚𝐧𝐠𝐮𝐚𝐠𝐞 𝐌𝐨𝐝𝐞𝐥𝐬 🌟 I am excited to share the release of my new textbook, 𝘐𝘯𝘵𝘳𝘰𝘥𝘶𝘤𝘵𝘪𝘰𝘯 𝘵𝘰 𝘓𝘢𝘳𝘨𝘦 𝘓𝘢𝘯𝘨𝘶𝘢𝘨𝘦 𝘔𝘰𝘥𝘦𝘭𝘴 (#LLMs) -- Perhaps the first textbook on LLMs. Target Audience: 👉 Students/beginners, Looking for a structured starting point to learn LLMs 👉 Teachers, planning to offer a course on LLMs 👉 Industry professional, seeking to deepen their understanding of LLMs Explore the Book: 🔗 Book Website: tanmoychak.com/llmbook/ 📑 Table of Contents: tanmoychak.com/llmbook/toc.pdf 🛒 Available on Amazon: amazon.in/dp/936386474X/ Enhance Your Learning Experience: 👉 Slides & Lecture Videos: Chapter-wise resources -- lcs2-iitd.github.io/ELL881-AIL821-… 👉 Exercises & Solutions: Practice with detailed chapter exercises (solutions available on request). 👉 Upcoming @nptel_official Course: Starting January 2025! Preview here: onlinecourses.nptel.ac.in/noc25_cs45/pre… Book Endorsement: 📖 Foreword by Prof. Tim Baldwin @eltimster 👏 Endorsements from Prof. Iryna Gurevych @IGurevych and Prof. Pushpak Bhattacharyya #LLMs #Textbook @iitdelhi @WileyIndiaPL @lcs2lab

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

🇧🇷 #LCS2 goes to #Rio 🇧🇷 Presenting our paper where we move beyond memoryless personalization → modeling user preferences as action-conditioned geometric walks with memory for better, user-aligned summaries. See you at #Riocentro 🚀 #Personalization #RepresentationLearning

🚨 CLEF 2026 - CheckThat! Lab We are excited to announce that we are organising a task at this year’s CheckThat! Lab, which extends the fact-checking pipeline with a new task focused on an important step in professional fact-checking: generating full fact-checking articles 📰

🚨 Submissions are now open for the Conference for AI Scientists (CAISc) 2026, co-organised by Lossfunk and @bitspilaniindia. Submit to probe what happens when AI systems drive scientific discovery. Submissions are open until May 15! Here is everything you need to know 🧵

An AI system MUST be the primary author: that's the only rule! Thrilled to be co-organizing this pioneering conference CAISc 2026! Send in your AI-driven research by May 15th.. @bitspilaniindia @ramgopal_rao @murari_ai @Tanmoy_Chak @palashiitkgp

Six papers from our lab have been accepted for publication in #ACL2026. The papers cover topics including Interpretability, empowering small VLMs with advanced tool calling, LLM personalisation, and different benchmarking. #nlproc @aclmeeting

@manojbalaji1 My paper was not rejected becasue of this :)

I completely understand your frustration, and honestly, the reasoning you were given is really difficult to accept. Sometimes in academia, we just run into unfortunate situations that feel entirely out of our control. I can definitely relate to what you are going through. Last year, my paper was rejected by EMNLP simply because one of my co-authors was flagged as an unresponsive reviewer. Unfortunately, the rest of the team only found out at the final decision stage. I actually reached out to the EMNLP 2025 organizers to ask why they didn't notify the co-authors beforehand. NeurIPS has a great system where they nudge co-authors so we can help remind the reviewer, which I think is a highly logical and effective approach. However, the EMNLP team explained that they had reminded the reviewer directly multiple times and that their policies simply differ from those of NeurIPS. It is incredibly disheartening when administrative policies, rather than the quality of the research itself, negatively impact a paper's outcome. The peer review system certainly has its flaws, and as a community, we need to keep advocating to improve them. Hang in there, and please don't let this discourage you! Many of us have faced similar hurdles, and your work is still valuable!

I strongly condemn and protest against rejecting a paper from ACL with such a justification. If I am not mistaken, "Findings" started with the motivation of accommodating such borderline "good" papers. I don’t see any reason behind such a justification, given that ACL does not have any venue constraints (runs in hybrid mode). #ACL2026 #NLProc @aclmeeting

@ANRFIndia Awesome. Thanks a lot @ANRFIndia

@Tanmoy_Chak Please check @Tanmoy_Chak

@ANRFIndia I'm trying to upload my MAHA proposal, but the portal appears to have issues: 1. The detailed proposal uploaded under the "Details" tab - "Upload Other Technical Details" is not reflected in the PDF under the under "Preview and Submit" tab. 2. My updated CV is not showing in the final PDF (the old version appears instead). Could this be checked urgently? The deadline is tomorrow.

Our newly introduced 𝐆𝐔𝐈𝐃𝐄-𝐋𝐋𝐌 -- A reporting checklist for using LLMs in behavioral & social science. Massive collaborative effort led by @stfeuerriegel.

2/ The organising committee for CAISc 2026 is led by @paraschopra, @dhruvtrehan9, and @gargdhruv36. We are glad to have @Tanmoy_Chak (IIT Delhi), Palash Goyal (Google Research), Dr Mohan Kankanhalli (NUS AI Institute), Shirish Karande (TCS Research) on our steering committee, and @murari_ai and Pratik Narang as our Program Committee Chairs. Additionally, our program committee for final human review spans CS, Mathematics, electrical engineering, and not just ML.

@flyspicejet @flyspicejet I received the refund today. Thank you very much.

@Tanmoy_Chak We sincerely regret any inconvenience caused, Tanmoy. We are in receipt of your email, and have updated our team to check and revert soon.

@flyspicejet Refund of ~₹1.97L for a failed booking on 6 Mar is still pending despite multiple follow-ups (even after the promised 5 working days). Case no: 10509519. No response to email and unsatisfactory customer support. If the refund is not processed immediately, I will be forced to escalate this to consumer protection authorities and financial regulators. Please don't expect people to have infinite bandwidth to follow up regularly.

@flyspicejet our flight from FJR to Delhi got cancelled again today -- SJ9085 at 1305. No rescheduled notification. This is the second time we booked tickets. My 4.5L of ticket fees is on hold. I am accompanied by my 3 yrs old son and wife. Pls reschedule the flight asap. We are not in a position to book other flight due to financial constraints. Pls understand the situation.

@Tanmoy_Chak Hi Tanmoy, we have responded to you via DM.

@flyspicejet We booked tickets from Fujairah to Delhi for 7th Mar at 13:05. The payment was successful. But we did not receive tickets. Please check asap.

@flyspicejet Alright. Please consider this very urgent. I will not book tickets further until you message me. @flyspicejet

@Tanmoy_Chak We shall reply to your DM shortly.

@flyspicejet @flyspicejet pls reply asap to my DM. We are waiting. Either you send us the tickets or cancel the transaction and initiate a refund so that we can book it again.

@Tanmoy_Chak Hi Tanmoy, we have responded to you via DM.

Our new study on interpretability explains -- 𝐭𝐡𝐞 𝐏𝐡𝐲𝐬𝐢𝐜𝐬 𝐨𝐟 𝐊𝐕 𝐂𝐚𝐜𝐡𝐞 𝐂𝐨𝐦𝐩𝐫𝐞𝐬𝐬𝐢𝐨𝐧 𝐟𝐨𝐫 𝐋𝐋𝐌𝐬 Pre-print: arxiv.org/abs/2603.01426 As context lengths continue to grow, the KV cache has become the primary memory bottleneck during inference. While many compression techniques report impressive memory savings with minimal drops in benchmark accuracy, we asked a more structural question: 👉 𝘞𝘩𝘢𝘵 𝘢𝘤𝘵𝘶𝘢𝘭𝘭𝘺 𝘩𝘢𝘱𝘱𝘦𝘯𝘴 𝘵𝘰 𝘢𝘵𝘵𝘦𝘯𝘵𝘪𝘰𝘯 𝘢𝘯𝘥 𝘳𝘦𝘢𝘴𝘰𝘯𝘪𝘯𝘨 𝘸𝘩𝘦𝘯 𝘸𝘦 𝘤𝘰𝘮𝘱𝘳𝘦𝘴𝘴 𝘵𝘩𝘦 𝘒𝘝 𝘤𝘢𝘤𝘩𝘦? We frame KV compression as a 𝐜𝐨𝐧𝐭𝐫𝐨𝐥𝐥𝐞𝐝 𝐩𝐞𝐫𝐭𝐮𝐫𝐛𝐚𝐭𝐢𝐨𝐧 𝐨𝐟 𝐭𝐨𝐤𝐞𝐧-𝐥𝐞𝐯𝐞𝐥 𝐫𝐨𝐮𝐭𝐢𝐧𝐠 𝐢𝐧 𝐬𝐞𝐥𝐟-𝐚𝐭𝐭𝐞𝐧𝐭𝐢𝐨𝐧. Rather than evaluating only final task accuracy, we design synthetic datasets to probe: (1) Multi-entity tracking, (2) Coreference resolution, and (3) Multi-hop reasoning. This setup allows us to disentangle three critical dimensions: Information Retention, Accessibility, and Utilisation. Our findings reveal an interesting pattern: 👉 𝐌𝐨𝐝𝐞𝐫𝐚𝐭𝐞 𝐜𝐨𝐦𝐩𝐫𝐞𝐬𝐬𝐢𝐨𝐧 often preserves surface-level accuracy despite substantial internal representational degradation — suggesting significant redundancy in current models. 👉 𝐍𝐞𝐚𝐫 𝐞𝐱𝐭𝐫𝐞𝐦𝐞 𝐜𝐨𝐦𝐩𝐫𝐞𝐬𝐬𝐢𝐨𝐧, we observe a sharp "safety cliff" in hallucinations, driven by global erasure of answer-critical tokens. 👉 We also uncover a second failure mode -- 𝐫𝐞𝐩𝐫𝐞𝐬𝐞𝐧𝐭𝐚𝐭𝐢𝐨𝐧𝐚𝐥 𝐫𝐢𝐠𝐢𝐝𝐢𝐭𝐲 -- where tokens remain present, but routing flexibility collapses. These results suggest that evaluating compression solely through downstream accuracy can mask stronger structural effects on reasoning. Understanding these internal dynamics is crucial as we move toward longer-context and more memory-efficient LLMs. Brilliant work by Ayan Sengupta and Samhruth Ananthanarayanan. #ScienceofLLMs #Interpretability #KVCache #ModelCompression