Shashwat Goel (@shashwatgoel7) 's Twitter Profile
Shashwat Goel

@shashwatgoel7

Scaling supervision for AI.

PhD student @ELLISInst_Tue @MPI_IS

Here for the aha and haha moments

ID: 1277988007304261632

linkhttp://shash42.github.io calendar_today30-06-2020 15:31:10

474 Tweet

567 Takipçi

652 Takip Edilen

Shashwat Goel (@shashwatgoel7) 's Twitter Profile Photo
Shashwat Goel
@shashwatgoel7

11 days ago

Confused about recent LLM RL results where models improve without any ground-truth signal? We were too. Until we looked at the reported numbers of the Pre-RL models and realized they were serverely underreported across papers. We compiled discrepancies in a blog below🧵👇

Confused about recent LLM RL results where models improve without any ground-truth signal? We were too. Until we looked at the reported numbers of the Pre-RL models and realized they were serverely underreported across papers. We compiled discrepancies in a blog below🧵👇