CUDA out of memory

[nichellehouston]

GPU:L40S 48GB
CPU:16
RAM:128

!python generate.py --task t2v-14B --size "832*480" --ckpt_dir ./Wan2.1-T2V-14B --offload_model True --t5_cpu --sample_shift 8 --sample_guide_scale 6 --prompt "xxx"

torch.OutOfMemoryError: CUDA out of memory. Tried to allocate 100.00 MiB. GPU 0 has a total capacity of 44.53 GiB of which 91.25 MiB is free. Process 3579258 has 44.43 GiB memory in use. Of the allocated memory 43.93 GiB is allocated by PyTorch, and 18.76 MiB is reserved by PyTorch but unallocated.

[pikachurus]

use this (#80) PR