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

  • Log In
  • Sign Up

apozz
/
WorldStereo-fp8

Image-to-Video
Safetensors
fp8
quantized
video
wan
worldstereo
Model card Files Files and versions
xet
Community
WorldStereo-fp8
20.4 GB
Ctrl+K
Ctrl+K
  • 1 contributor
History: 10 commits
apozz's picture
apozz
fix: re-cook (Linear-only) -- previous version cast 2 Conv3d weights as fp8 which broke load_state_dict
af612e1 verified 12 days ago
  • .gitattributes
    1.52 kB
    initial commit 12 days ago
  • LICENSE
    1.36 kB
    add LICENSE 12 days ago
  • NOTICE
    1.21 kB
    add NOTICE 12 days ago
  • README.md
    6.28 kB
    update: 549 fp8 (Linear-only), Conv weights stay bf16 12 days ago
  • WorldStereo-memory-dmd-fp8_e4m3fn_scaled.safetensors
    20.4 GB
    xet
    fix: re-cook (Linear-only) -- previous version cast 2 Conv3d weights as fp8 which broke load_state_dict 12 days ago
  • config.json
    3.91 kB
    add config.json 12 days ago
  • cook_worldstereo_fp8.py
    7.48 kB
    fix: is_castable_weight now requires len(shape)==2 (Linear only); Conv weights stay bf16 12 days ago