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

📅 2026-02-26
📈 Citations: 0
Influential: 0
📄 PDF

career value

195K/year
🤖 AI Summary
This work proposes a systematic data engineering methodology to address the engineering challenges of processing and managing raw video data for large-scale video foundation model training. By leveraging metadata-driven data cleaning, multi-stage filtering, and an inference-aware architecture—combined with the Lavender Data management system, μP hyperparameter transfer, and hyperspherical geometric constraint optimization—the authors efficiently construct a high-quality training set from approximately 50 million video clips. Using this curated dataset, they successfully train Summer-22B, a 22-billion-parameter video foundation model. This study provides the first empirical validation that systematic data engineering plays a dominant role in enhancing model performance, underscoring the critical importance of data quality and structural integrity in large-scale video modeling.

Technology Category

Application Category

📝 Abstract
We describe our experience training Summer-22B, a video foundation model developed from scratch. This report documents the engineering challenges, design decisions, and lessons learned while scaling from raw footage collection to a functional model trained on approximately 50 million clips. We outline our approach combining metadata-driven dataset curation, multi-stage filtering, $μ$P parameterization, and hypersphere-constrained optimization. We developed the Lavender Data system for dataset management and adopted inference-aware architectural choices. We share observations on what worked in our setting: dataset engineering consumed the majority of effort, architectural variants showed smaller differences than we expected, and $μ$P hyperparameter transfer appeared effective even under geometric constraints. We hope this account proves useful to others undertaking similar projects.
Problem

Research questions and friction points this paper is trying to address.

video foundation model
dataset engineering
large-scale training
data curation
model scaling
Innovation

Methods, ideas, or system contributions that make the work stand out.

dataset engineering
μP parameterization
hypersphere-constrained optimization
video foundation model
inference-aware architecture
🔎 Similar Papers
2024-02-20International Conference on Machine LearningCitations: 30