Meta tags:
author=Rahul Ramachandran;

Headings (most frequently used words):

education, research,

Text of the page (most frequently used words):
and (7), models (6), vision (6), computer (5), rahul (5), ramachandran (5), tasks (4), the (3), for (3), science (3), prof (3), code (2), datasets (2), description (2), arxiv (2), vineeth (2), balasubramanian (2), using (2), multimodal (2), gpt (2), standard (2), particularly (2), amir (2), zamir (2), research (2), iit (2), hyderabad (2), bachelor (2), university (2), illinois (2), urbana (2), champaign (2), education (2), with (2), about (2), stolen, from, jon, barron, source, 2024, use, item, response, theory, irt, assess, model, calibration, select, informative, data, subsets, demonstrate, usefulness, various, latent, parameters, analyzing, comparing, dmlr, icml, deepak, vijaykeerthy, charchit, sharma, tejal, kulkarni, evaluation, human, competency, frameworks, 2026, benchmark, top, like, gemini, prompt, based, framework, while, these, are, strong, generalists, especially, semantic, they, still, trail, behind, specialized, geometry, website, iclr, oğuzhan, fatih, kar, andrei, atanov, roman, bachmann, ali, garjani, how, well, does, understand, evaluating, foundation, 2025, 2030, expected, previously, finished, had, privilege, working, iith, epfl, first, year, student, where, fortunate, advised, excited, robot, learning, currently, interested, sim, real, transfer, manipulation, non, prehensile, saurabh, gupta, ɹɑːmʌtʃʌnðɹʌn, ɹɑːhuːl,

Text of the page (random words):
rahul ramachandran rahul ramachandran about education research rahul ɹɑːhuːl ramachandran ɹɑːmʌtʃʌnðɹʌn hi i m a first year ph d student at the university of illinois urbana champaign where i m fortunate to be advised by prof saurabh gupta i m excited about robot learning and am currently interested in sim to real transfer for manipulation particularly for non prehensile tasks previously i finished my bachelor s in computer science at iit hyderabad i ve had the privilege of working with prof vineeth n balasubramanian at iith and with prof amir zamir at epfl education ph d in computer science 2030 expected university of illinois urbana champaign bachelor s in computer science 2025 iit hyderabad research how well does gpt 4o understand vision evaluating multimodal foundation models on standard computer vision tasks rahul ramachandran ali garjani roman bachmann andrei atanov oğuzhan fatih kar amir zamir iclr 2026 arxiv website code description we benchmark top multimodal models like gpt 4o and gemini on standard vision tasks using a prompt based framework while these models are strong generalists especially on semantic tasks they still trail behind specialized vision models particularly in geometry on evaluation of vision datasets and models using human competency frameworks rahul ramachandran tejal kulkarni charchit sharma deepak vijaykeerthy vineeth n balasubramanian dmlr icml 2024 arxiv description we use item response theory irt to assess model calibration select informative data subsets and demonstrate the usefulness of various latent parameters for analyzing and comparing models and datasets in computer vision stolen from jon barron s source code