ContentV: Efficient Training of Video Generation Models with Limited Compute

TL;DR

ContentV is a highly efficient text-to-video model that achieves state-of-the-art results with significantly reduced computational resources by leveraging innovative architecture, training strategies, and reinforcement learning techniques.

Contribution

The paper introduces ContentV, a novel 8B-parameter text-to-video model that combines a minimalist architecture, multi-stage training, and reinforcement learning to enable high-quality video generation with limited compute.

Findings

Achieves 85.14 on VBench with 4 weeks of training on 256 NPUs

Generates diverse, high-quality videos across multiple resolutions and durations

Reduces computational costs significantly compared to previous models

Abstract

Recent advances in video generation demand increasingly efficient training recipes to mitigate escalating computational costs. In this report, we present ContentV, an 8B-parameter text-to-video model that achieves state-of-the-art performance (85.14 on VBench) after training on 256 x 64GB Neural Processing Units (NPUs) for merely four weeks. ContentV generates diverse, high-quality videos across multiple resolutions and durations from text prompts, enabled by three key innovations: (1) A minimalist architecture that maximizes reuse of pre-trained image generation models for video generation; (2) A systematic multi-stage training strategy leveraging flow matching for enhanced efficiency; and (3) A cost-effective reinforcement learning with human feedback framework that improves generation quality without requiring additional human annotations. All the code and models are available at: https://contentv.github.io.