Full CLIP-L (FP8 Vision) for Hunyan Video

Full CLIP-L (FP8 Vision) for Hunyan Video

This model is a full scaled version of Vit-CLIP-L (Un-pruned full Vision Model)

• Works in Hunyan Video

• Works in FLUX/SDXL (Vision Tree will likely be pruned or ignored)

For those who are unfamiliar the CLIP-L we commonly use is just a VIT-CLIP with the vision blocks pruned.

For video models the vision blocks can be utilized.