CUDA out of memory

[simba0626]

Hi authors,

MEMIT is an interesting work.
When I run: python -m experiments.evaluate --alg_name=MEMIT --model_name=EleutherAI/gpt-j-6B --hparams_fname=EleutherAI_gpt-j-6B.json --num_edits=10 --use_cache

There is error about CUDA out of memory:
File "\memit-main\memit\memit_main.py", line 97, in
weights_copy = {k: v.detach().clone() for k, v in weights.items()}
RuntimeError: CUDA out of memory. Tried to allocate 256.00 MiB (GPU 0; 24.00 GiB total capacity; 23.15 GiB already allocated; 0 bytes free; 23.16 GiB reserved in total by PyTorch) If reserved memory is >> allocated memory try setting max_split_size_mb to avoid fragmentation. See documentation for Memory Management and PYTORCH_CUDA_ALLOC_CONF

My local is 24GB 3090 GPU.
Would you help me how to run the MEMIT code ? How to revise the configure file (EleutherAI_gpt-j-6B.json) , in order to reduce memory?

Thank you very much.
MEMIT is very nice.