PulseAugur / Brief
EN
LIVE 02:33:27

Brief

last 24h
[2/2] 221 sources

Multi-source AI news clustered, deduplicated, and scored 0–100 across authority, cluster strength, headline signal, and time decay.

  1. TOOL · Mastodon — sigmoid.social 日本語(JA) ·

    NVIDIA releases real-time video generation AI "LongLive-2.0", achieving lightweight and high-quality generation through training assuming FP4 quantization – GIGAZINE https://www.yayafa.com/2807728/ # AgenticAi # AI # ArtificialGeneralIntelligence

    NVIDIA has unveiled LongLive-2.0, a real-time AI video generation model. This model is designed to be lightweight and high-quality by incorporating FP4 quantization into its training process. This approach allows for efficient generation of video content. AI

    NVIDIA releases real-time video generation AI "LongLive-2.0", achieving lightweight and high-quality generation through training assuming FP4 quantization – GIGAZINE https://www.yayafa.com/2807728/ # AgenticAi # AI # ArtificialGeneralIntelligence

    IMPACT This model's focus on efficient, real-time video generation could accelerate creative workflows and new applications in media and entertainment.

  2. TOOL · arXiv cs.CV English(EN) ·

    LongLive-2.0: An NVFP4 Parallel Infrastructure for Long Video Generation

    Researchers have developed LongLive-2.0, a parallel infrastructure designed to optimize the training and inference of long video generation models. This system utilizes NVFP4 precision and sequence-parallel autoregressive training to reduce memory requirements and accelerate computations. For inference, LongLive-2.0 employs techniques like W4A4 NVFP4 inference and asynchronous streaming VAE decoding to achieve high throughput, demonstrating up to a 2.15x speedup in training and 1.84x in inference. AI

    LongLive-2.0: An NVFP4 Parallel Infrastructure for Long Video Generation

    IMPACT Enables more efficient training and faster inference for long video generation models, potentially leading to wider adoption and new applications.