why the memeory occupy still high even after the generating finish

Nobody has claimed this yet.

running the cmdlet
close the cmdlet

Currently enabled native sdp backends: ['flash', 'math', 'mem_efficient', 'cudnn']
Xformers is installed!
Flash Attn is installed!
Sage Attn is installed!
Namespace(share=False, server='127.0.0.1', port=None, inbrowser=True)
Free VRAM 5.03515625 GB
High-VRAM Mode: False
Downloading shards: 100%|████████████████████████████████████████████████████████████████████████| 4/4 [00:00
Traceback (most recent call last):
File "asyncio\events.py", line 80, in _run
File "asyncio\proactor_events.py", line 162, in _call_connection_lost
ConnectionResetError: [WinError 10054] 远程主机强迫关闭了一个现有的连接。
100%|██████████████████████████████████████████████████████████████████████████████████| 25/25 [07:42<00:00, 18.49s/it]
Offloading DynamicSwap_HunyuanVideoTransformer3DModelPacked from cuda:0 to preserve memory: 8 GB
Loaded AutoencoderKLHunyuanVideo to cuda:0 as complete.
Unloaded AutoencoderKLHunyuanVideo as complete.
Decoded. Current latent shape torch.Size([1, 16, 18, 60, 104]); pixel shape torch.Size([1, 3, 69, 480, 832])
latent_padding_size = 9, is_last_section = False
Moving DynamicSwap_HunyuanVideoTransformer3DModelPacked to cuda:0 with preserved memory: 6 GB
100%|██████████████████████████████████████████████████████████████████████████████████| 25/25 [07:12<00:00, 17.32s/it]
Offloading DynamicSwap_HunyuanVideoTransformer3DModelPacked from cuda:0 to preserve memory: 8 GB
Loaded AutoencoderKLHunyuanVideo to cuda:0 as complete.
Unloaded AutoencoderKLHunyuanVideo as complete.
Decoded. Current latent shape torch.Size([1, 16, 27, 60, 104]); pixel shape torch.Size([1, 3, 105, 480, 832])
latent_padding_size = 0, is_last_section = True
Moving DynamicSwap_HunyuanVideoTransformer3DModelPacked to cuda:0 with preserved memory: 6 GB
100%|██████████████████████████████████████████████████████████████████████████████████| 25/25 [07:05<00:00, 17.00s/it]
Offloading DynamicSwap_HunyuanVideoTransformer3DModelPacked from cuda:0 to preserve memory: 8 GB
Loaded AutoencoderKLHunyuanVideo to cuda:0 as complete.
Unloaded AutoencoderKLHunyuanVideo as complete.
Decoded. Current latent shape torch.Size([1, 16, 37, 60, 104]); pixel shape torch.Size([1, 3, 145, 480, 832])

No source files or tests are identified. Start by reproducing the reported command and checking GPU memory after the final “Decoded” log, alongside the unload and offload messages. Done means identifying why memory remains high after generation and verifying the behavior after the relevant change.

Written by the indexing model from the issue text.