Summer-22B: A Systematic Approach to Dataset Engineering and Training at Scale for Video Foundation Model

Simo Ryu, Chunghwan Han · 2026

Read Summer-22B: A Systematic Approach to Dataset Engineering and Training at Scale for Video Foundation Model by Simo Ryu, Chunghwan Han (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2603-00173.txt.

Description

Ryu and Han describe training Summer-22B, a video foundation model built from scratch on about 50 million clips. They combine metadata curation, multi-stage filtering, μP parameterisation and hypersphere-constrained optimisation, run via the Lavender Data system with inference-aware architecture. Dataset work took most of the effort, design variants mattered less than hoped, and μP transfer worked even under geometric constraints. The report shares lessons for others scaling similar projects.

Catalog record

CREATORSimo Ryu, Chunghwan Han
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE59382
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.00173

Browse the full library · About the Tower of Babel