Hugging Face's logo Hugging Face
  • Models
  • Datasets
  • Spaces
  • Buckets
  • Docs
  • Enterprise
  • Pricing
    • Website
      • Tasks
      • HuggingChat
      • Collections
      • Languages
      • Organizations
    • Community
      • Blog
      • Posts
      • Daily Papers
      • Hardware
      • Learn
      • Discord
      • Forum
      • GitHub
    • Solutions
      • Team & Enterprise
      • Hugging Face PRO
      • Enterprise Support
      • Inference Providers
      • Inference Endpoints
      • Storage Buckets

  • Log In
  • Sign Up

saaivignesh20
/
MAGI-2-preview-FP8

Text-to-Video
Safetensors
video-generation
audio-video
fp8
Model card Files Files and versions
xet
Community
MAGI-2-preview-FP8
199 GB
Ctrl+K
Ctrl+K
  • 1 contributor
History: 11 commits
saaivignesh20's picture
saaivignesh20
Full model mirror noted; refiner on Blackwell (FlexAttention window attention, FA2 stand-ins), warm worker, configs writer
f485501 verified 15 days ago
  • configs
    Upload folder using huggingface_hub 15 days ago
  • magi2_fp8
    Full model mirror noted; refiner on Blackwell (FlexAttention window attention, FA2 stand-ins), warm worker, configs writer 15 days ago
  • preview
    FP8 routed experts for MAGI-2 Preview 15 days ago
  • refiner
    Mirror SandAI's MAGI-2 Preview refiner (rev 2dea51b6) 15 days ago
  • stable-audio-open-1.0
    Mirror SandAI's MAGI-2 Preview stable-audio-open-1.0 (rev 2dea51b6) 15 days ago
  • text_encoder
    Mirror SandAI's MAGI-2 Preview text_encoder (rev 2dea51b6) 15 days ago
  • turbo_vae
    Mirror SandAI's MAGI-2 Preview turbo_vae (rev 2dea51b6) 15 days ago
  • vae
    Mirror SandAI's MAGI-2 Preview vae (rev 2dea51b6) 15 days ago
  • .gitattributes
    1.58 kB
    Mirror SandAI's MAGI-2 Preview text_encoder (rev 2dea51b6) 15 days ago
  • LICENSE
    10.8 kB
    Upload folder using huggingface_hub 15 days ago
  • README.md
    7.87 kB
    Full model mirror noted; refiner on Blackwell (FlexAttention window attention, FA2 stand-ins), warm worker, configs writer 15 days ago
  • quantization_config.json
    974 Bytes
    W8A8 expert kernel (FP8 x FP8 tensor cores, K-major weights at load): 11.4 ms vs 17.7 bf16 per call; A/B results 15 days ago