CalamitousFelicitousness 9f91726020 fix(ideogram4): resolve activation dtype for quantized transformers ...

transformer.dtype reports the storage dtype for fp8-quantized models
(fp8 params are floating, int8 params are not), so activations were
cast to fp8 and the forward crashed. Read the compute dtype from the
SDNQ dequantizer instead.

2026-07-11 06:40:07 +01:00

..

__init__.py

feat(model): secondary unet override slot and ideogram4 native loading

2026-07-11 06:40:07 +01:00

ideogram4.py

fix(ideogram4): resolve activation dtype for quantized transformers

2026-07-11 06:40:07 +01:00

patch_qwen.py

fix(ideogram4): scope qwen3-vl forward patch to the enhance window

2026-06-07 00:30:45 +01:00