openfold

openfold copied to clipboard

Output issue on H100 with memory-efficient kernel

13

Hello, As per [this line](https://github.com/aqlaboratory/openfold/blob/main/openfold/model/evoformer.py#L649), when neither Deepspeed nor LMA is selected, the custom memory-efficient [kernel](https://github.com/aqlaboratory/openfold/tree/main/openfold/utils/kernel) is used for the attention layer. When running inference with this option on H100,...

psohani

I am trying to fine-tune OpenFold multimer. I see that the OpenFold training scripts now contain a multimer option, but there are a few open issues in this forum (e.g....

ecalfee

Trying to run the example training from here: https://openfold.readthedocs.io/en/latest/Training_OpenFold.html Command: ``` python3 train_openfold.py $DATA_DIR/pdb_data/mmcif_files \ $DATA_DIR/alignment_data/ $DATA_DIR/pdb_data/mmcif_files $OUTPUT_DIR 2021-10-10 \ --template_release_dates_cache_path ../../pdb_data/data_caches/mmcif_cache.json \ --train_chain_data_cache_path ../../pdb_data/data_caches/chain_data_cache.json \ --seed 1 --num_nodes 1...

RJ3

That direct clone of branch url does not work. Replaced it with a proper clone command with -b.

vaclavhanzl

vaclavhanzl

When I installed teh ESM,I can easy fllow this code: ``` import torch import esm # Load ESM-2 model model, alphabet = esm.pretrained.esm2_t33_650M_UR50D() batch_converter = alphabet.get_batch_converter() model.eval() # disables dropout...

cpplyqz

Random crop during training

1

Hello, This question is regarding the 'random_crop_to_size' transform [here](https://github.com/aqlaboratory/openfold/blob/f37d0d964d24b20494776b7f5643113dd584126c/openfold/data/data_transforms.py#L1183). Is the protein cropped at different locations across the different epochs during training or are they cropped the same way? Thanks!

VarunUllanat

I am trying to train a multimer model resuming from jax weights and get the following error: Traceback (most recent call last): File "/work/10110/abhinav22/ls6/openfold/train_openfold.py", line 706, in main(args) File "/work/10110/abhinav22/ls6/openfold/train_openfold.py",...

abhinavb22

tests.test_primitives.TestLMA: torch.cuda.OutOfMemoryError: CUDA out of memory

3

I've just run `scripts/install_third_party_dependencies.sh` and `python3 setup.py install`. No erros during the instalation. `resouces` folder was created inside openfold folder, but: ``` ./scripts/run_unit_tests.sh [2024-07-09 18:14:45,945] [INFO] [real_accelerator.py:161:get_accelerator] Setting ds_accelerator to...

rafaeltiveron

Hi, I am trying to install openfold on my work station that has a TU117GL. I manage to install everything, but when I run unit test I get RuntimeError: FlashAttention...

fapirro