← Latest products

NeuroFlow 55.8x video inference speedup for Vision Transformers PyTorch

Official PyTorch implementation of NeuroFlow: EMA-Gated Temporal Sequence Compression for Vision Transformers. Achieves up to 55.8x wall-clock speedup for video inference via semantic surprise routing and a training-free Dual-Memory Reconstruction Protocol.

Open source

19 stars · GitHubexternal site

Visit github.comexternal site

Reactions (1)

Similar products

by description similarity

  1. DeepDream for Video with Temporal Consistency

    github.com

    DeepDream for video with temporal consistency. Features RAFT optical flow estimation and occlusion masking to prevent ghosting. A PyTorch implementation.

    Hacker News72 pts

  2. Graphlearn-for-PyTorch

    github.com

    distributed graph learning on PyTorch

    Hacker News9 pts

  3. Compress vision-language and unimodal AI models by structured pruning

    github.com

    [ICML 2023] UPop: Unified and Progressive Pruning for Compressing Vision-Language Transformers

    Hacker News7 pts

  4. Curated Transformers

    github.com

    PyTorch LLMs with less code duplication

    Hacker News10 pts

  5. Reimplementing PyTorch from scratch (MLP

    github.com

    CNN) to learn the internals

    Hacker News5 pts