Jack

Thank you for your prompt reply! ---Original--- From: "Edward Z. ***@***.***> Date: Mon, Jul 22, 2024 09:18 AM To: ***@***.***>; Cc: ***@***.******@***.***>; Subject: Re: [pytorch/pytorch] dynamo supports Tensor.tolist but notTensor.item...

Would you mind to share more details?

> Do you have any intuition on why this happens? Or any suggestion to mitigate it? I don't know whether if you set lower “strength” will help. My situation is...

@TJU-WEIHAO Have you find anyways to findtune the TCD lora?

> Is it necessary to run garment_net multi-times to get reference_features in [L1784](https://github.com/yisol/IDM-VTON/blob/80817aecbcd4836fdffe508c1528190f60396acd/src/tryon_pipeline.py#L1783) ? I think it is ok to only use the reference_features of timestep=0 in all denoising steps...

> Hello, thanks for your recognition, and sorry for the late reply. The code for computing cross-attention difference code is based on some internal code base, for which we do...

> Yeah, EulerA sampler can be used on TCD-LoRA, and EulerA sampler can also be used on LCM-LoRA. It gives decent results, but for better work it may require a...

> Additional Debugging > > When running the test program with `PYTHONMALLOC=debug` environment variable, Python will check the status of the GIL when performing memory allocation and deallocation: > >...

@strint Would you mind to give a loot at the issue ? `Fatal Python error: _PyMem_DebugFree: Python memory allocator called without holding the GIL` Deeply grateful! ``` Stack trace (most...

> > Hi @yingchingl Were you able to solve it? Would you mind kindly to give some advice about it? > > Hi @bigmover , No, I can't solve it...