🚨 NEW PAPER: "Optimizing Test-Time Compute via Meta Reinforcement Fine-Tuning"!
🤔 With all these long-reasoning LLMs, what are we actually optimizing for? Length penalties? Token budgets? We needed a better way to think about it!
Website: https://t.co/G5JTmryx0d
🧵[1/9]
The paper I'll be presenting is on the work we did previously at Microsoft Research to understand the digital news consumption patterns of blind and visually impaired individuals in India.
Video: https://t.co/zQjUpiyNSe
Paper: https://t.co/h33CkK5Zd2
@mohit088@Iameghnaa
The power language models hold blew my mind. @github copilot (autocompletion) model could fit a linear model to the input data and extrapolate to test cases. That’s so neat!
It was indeed a memorable day! Winning
#sih2020 motivated our team to make efforts for brining a positive change in the country!
Kudos to the team working 24x7
@SIH_2020
#smartindiahackathon2020#ivlabs#VNIT#Nagpur
@abhayjere@AICTE_INDIA@adsahasrabudhe@PMOIndia@Persistentsys@HRDMinistry I have an internship planned around the estimated dates, which is the case with other pre-final year students everywhere. Scheduling the competition in this time frame will definitely bring down the participation. I kindly request you to reconsider the decision.