DoubleXKing

## Description I transfer my model to tensorrt engine using tftrt in IN8. However, the speed is same as FP32 even FP16. I also change minimum_segment_size to 2, 3, 5,...

load_ip_adapter cannot load ip-adapter-faceid-plusv2_sdxl

1

My code: `pipe.load_ip_adapter("h94/IP-Adapter-FaceID", subfolder="", weight_name="ip-adapter-faceid-plusv2_sdxl.bin")` It gives error: ip-adapter-faceid-plusv2_sdxl.bin: 100%|█████████████████████████████████████████████████████████| 1.49G/1.49G [01:47