Siba Smarak Panigrahi
@sibasmarak
PhD in @ICepfl @EPFL_en | MSc in CS @mcgillu and @Mila_Quebec | UG in CSE @IITKgp | Prev. @ServiceNowRSRCH @ProseMsft @Adobe @USCViterbi @mitidss | 🇨🇭🇨🇦🇮🇳
I am excited to share our #CVPR2025 paper on combining two topics close to my heart: equivariance with canonicalization and state space models. In this work, we provide a recipe to make VisionMamba equivariant/invariant by canonicalizing the patch token representations before…
Symmetry is the fundamental property of crystals, yet generative models don't yield crystals with realistic symmetries We solved that with SymmCD and can get crystals from any of the 230 space groups Learn more at our #ICLR poster w/@dnllvy @sibasmarak arxiv.org/abs/2502.03638 🧵
I'll be at hashtag #ICLR2025 in Singapore this week! 🎉🇸🇬 1️⃣ BigDocs: An Open Dataset for Training Multimodal Models on Document and Code Tasks 🗓️ Thu, Apr 24 | 10:00 a.m.–12:30 p.m. SGT 📍 Hall 3 + Hall 2B | Poster 280 (continual pre-training dataset to improve code and…
Thrilled to share LUNA🌕 – our new generative AI model that reassembles tissues from dissociated cells! Think of it as AlphaFold for cells 🧬 From the MERFISH mouse brain atlas to the scRNA-seq atlas & Slide-tags datasets, LUNA is widely applicable. Bringing us closer to virtual…
Tweets that make your day! 😁
This symmetric diffusion paper at ICLR is nice (simple idea in retrospect): SymmCD: Symmetry-Preserving Crystal Generation with Diffusion Models We'd actually implemented this idea internally at Orbital, and it works nicely even for very large crystal structures:
I'm excited that 2/2 accepted to @iclr_conf in Singapore! For the first time, 2 out of 2! 🎉 1️⃣ "BigDocs: An Open and Permissively-Licensed Dataset for Training Multimodal Models on Document and Code Tasks" with collaborators from @ServiceNowRSRCH 2️⃣ "SymmCD:…

We are excited to announce MLRC 2025, the eighth iteration of MLRC, which will also be its first in-person edition @ Princeton University, NJ, USA on August 21st, 2025 reproml.org/blog/announcin… Thanks to Princeton AI Lab & Arvind Narayanan @random_walker for facilitating the venue!
Announcing SGI 2025! Undergrads and MS students: Apply for 6 weeks of paid summer geometry processing research. No experience needed: 1 week tutorials + 5 weeks of projects. Mentors are top researchers in this emerging branch of graphics/computing/math. sgi.mit.edu
How can you efficiently design snowflakes? 🚨 Another paper at #NeurIPS2024 workshop - this time at AI4 Accelerated Materials design! In this work, to generate symmetric crystals (you can create snowflakes also 😉), we learn 1⃣ The element coordinates and types within the…
I'm at #NeurIPS2024 this week, presenting our work SymmCD at the #ai4mat workshop! openreview.net/forum?id=V7x2K… Come check it out, and feel free to message me to meet up, especially if you want to chat about generative modelling for materials, crystals, and symmetry!
Finally! 🥳 It was a wild ride to work on BigDocs as a visiting researcher at @ServiceNowRSRCH, Montreal, this summer. I learnt a lot of skills, especially how to make multi-node, multi-GPU training go brrrr on H100s. 🌐 Website: bigdocs.github.io ✈️ Catch at…
🎉 Excited to introduce BigDocs! An open, transparent multimodal dataset designed for: 📄 Documents 🌐 Web content 🖥️ GUI understanding 👨💻 Code generation from images We’re also launching BigDocs-Bench, featuring 10 tasks to test models on: ➡️ Document, Web, GUI Visual…
🚨 Exciting Opportunity! 🚨 @GoogleDeepMind India is hiring for multiple open positions in its Pre-doctoral Researcher (AI Resident) program! 🎓 If you're graduating (Bachelor’s/Master’s) by mid-2025 & dream of advancing AI research, this is your launchpad! 🚀 Build a strong…
Did this happen? Yes! ✅ I'm officially an alumnus of @mcgillu and @Mila_Quebec! The past two years have been an unforgettable growth journey, filled with learning and discovering a home away from home. Thanks to the unwavering support of my family, supervisor, and friends, I’m…


Hi @iclr_conf , how about making it mandatory for reviewers to provide a conclusion and adjusted scores? If they don't, it should be seen as an irresponsible review, leading to a desk rejection. We’re spending a lot of time on these rebuttals!
How much resource? 10,000 GPUs. So you have an excellent chance to make them go brrrrrrr at @EPFL_en beside the Alps and the lake! 🏔️🌊 Our lab is leading the science vertical of the SwissAI initiative. Want to know more?🔗swiss-ai.org
🚀 We're #Hiring #PhD students at MLBio Lab at @EPFL_en through the EDIC Program! 📅 Deadline: Dec 15 Are you passionate about generative AI, unsupervised learning and solving problems in biomedicine? 🧬 Apply & mention your interest in your application. 💥 We're part of…
I remember back in the day, in my early chess tournaments, my mother bought Minute Maid orange juice because Magnus used to keep orange juice in his matches. I miss those days. 100 percent. 🥺
Great opportunity for those applying to #grad schools! You'll get (free) feedback from #PhD at @ICepfl in your domain of interest. The form is filling fast, so you know what to do 😉
Applying for a CS-related PhD soon? The PhD volunteers from EPFL's Computer & Communication Sciences department can help review and provide feedback on the SoPs and resumes of people who lack mentorship. Register in this form forms.gle/qDMRypmzfxosWL… Created by @shail_hash