Hugging Face's logo

amd-shark
/
sdxl-quant-int8

Transformers
Model card Files
xet
 Community
2
sdxl-quant-int8 / math_model.py

Commit History

  1. Update quant params structure (#2)⁠

    nickfraser
    Verified signature
    6b62ce4

  1. Remove potential overflow / saturation error.

    nickfraser
    161df88

  2. Added comments - highlight possible overflow situation

    nickfraser
    3f5851c

  3. Updated math model to target int8 x int8 kernels.

    nickfraser
    4024f9d

  1. Updated QOp model to fuse SmoothQuant scales with input quantization

    nickfraser
    dca9b6e

  1. Fix: set `keepdim=True`

    nickfraser
    9ab1060

  2. [math_model/test] Added "QOp" implementation and basic tests.

    nickfraser
    eb5a5f6

  1. Upload math_model.py with huggingface_hub

    GiusFra
    Verified signature
    049c65f

  2. Upload math_model.py with huggingface_hub

    GiusFra
    Verified signature
    d5dfd96