Latent Space Paper Club: AIEWF Special Edition (Test of Time, DeepSeek R1/V3) — VIbhu Sapra
AI Engineer
In this podcast, Vibhu Sapra reviews the past year and a half of Paper Club, highlighting its consistent weekly sessions with authors from NVIDIA, Meta, and Amazon, averaging 100 attendees and reaching 300 for DeepSeek V3. He announces the launch of "Test of Time Paper Club," a curriculum-based V2 focused on foundational AI papers, running from July to December with both in-person (San Francisco) and remote options, covering core topics and featuring presentations on 2-4 papers weekly. The session also includes a review of the DeepSeek model, including the May 28 update, which shows significant improvements in reasoning and performance, as well as a new distillation model based on Quen 3 8B.
Part 1: Paper Club Overview
Part 2: DeepSeek V3 Analysis
Part 3: DeepSeek Training and Performance
Part 4: Conclusion and Acknowledgements
Sign in to continue reading, translating and more.
Open full episode in Podwise