akshit kumar
komikat.bsky.social
akshit kumar
@komikat.bsky.social
mfw tfw
akshit.one
Reposted by akshit kumar
Interviewer: Can you explain this gap in your resume?

Assyriologist: Du[ring] grad sch[ool] I [took?] a position ... [13 lines illegible] ... now.
December 20, 2024 at 8:36 PM
Reposted by akshit kumar
Transformer LMs get pretty far by acting like ngram models, so why do they learn syntax? A new paper by sunnytqin.bsky.social, me, and @dmelis.bsky.social illuminates grammar learning in a whirlwind tour of generalization, grokking, training dynamics, memorization, and random variation. #mlsky #nlp
Sometimes I am a Tree: Data Drives Unstable Hierarchical Generalization
Language models (LMs), like other neural networks, often favor shortcut heuristics based on surface-level patterns. Although LMs behave like n-gram models early in training, they must eventually learn...
arxiv.org
December 20, 2024 at 5:56 PM
Reposted by akshit kumar
Neutrino Modem xkcd.com/3017
November 28, 2024 at 9:55 PM
#intro doing btech cs + ms (research) cl at iiit hyderabad. have worked w dependency parsing, interested in lm reasoning and interpretability
November 21, 2024 at 7:44 AM
Reposted by akshit kumar
when you try to convert your text into smaller pieces but all it gives you is Elvish, that’s a tolkienizer
November 20, 2024 at 5:51 PM