out of memory

torch.OutOfMemoryError: CUDA out of memory. Tried to allocate 16.00 MiB. GPU 0 has a total capacity of 31.47 GiB of which 16.62 MiB is free. Including non-PyTorch memory, this process has 31.44 GiB memory in use. Of the allocated memory 31.11 GiB is allocated by PyTorch, and 39.01 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables)

the GPU memory is 32GB,and I set PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True also, i cannot work

The report names no files, tests, entry points, training command, or model configuration. Start by reproducing the failing training or fine-tuning workload with the reported 32 GB GPU and PYTORCH_CUDA_ALLOC_CONF setting, then inspect the memory-management documentation linked in the issue. The issue needs reproduction details and an agreed success condition before completion can be verified.

Written by the indexing model from the issue text.