Out of memory error

[ropde]

Hello,

Tried to run ribbonfold on a sequence of 325 a.a (5 chains) and got the out of memory error below. Is there anything I could try to overcome this or is this sequence just too big for my hardware? The example from the repo runs fine on the same machine.

Rounds: 10
Recycles: 5
MSA Random Sampling Mode: one_cluster
Use Dropout: True
Use Initial Structure: True
load model checkpoint success: ./ckpt/model_ckpt_001.pt
run sample 0...
full sequence length: 1625
monomer chain length: 325
number of chains: 5
2025-11-05 12:17:07.453391: I tensorflow/core/platform/cpu_feature_guard.cc:142] This TensorFlow binary is optimized with oneAPI Deep Neural Network Library (oneDNN) to use the following CPU instructions in performance-critical operations: AVX2 AVX512F FMA
To enable them in other operations, rebuild TensorFlow with the appropriate compiler flags.
num_recycle: 0
...
torch.cuda.OutOfMemoryError: CUDA out of memory. Tried to allocate 646.00 MiB (GPU 0; 11.72 GiB total capacity; 9.97 GiB already allocated; 262.00 MiB free; 10.45 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

Thank you

[liangyue1133]

Hello, and thank you for your question.
The total length of this protein—over 1,600 residues—is indeed quite large and can lead to GPU memory issues. The memory consumption of our algorithm is comparable to that of AlphaFold2, and such sequence length is likely beyond the capacity of typical GPU setups.
Thank you