Gaurav Diwan

The error was indeed due to the format of my COGs file. Only one sequence per organism is permitted. Changing the format to this aspect solved the issue, although the...

CUDA version 11.6 jax 0.3.25 jaxlib 0.3.25+cuda11.cudnn82

I see, then it must be something else. I tried again with the latest version of cudatoolkit (11.8) and cudnn (8.4.1), but it still fails. I am also running the...

Unfortunately that does not work either. I get the following error ``` I0302 12:01:53.762366 22486551396480 templates.py:857] Using precomputed obsolete pdbs ../alphafold_data/pdb_mmcif/obsolete.dat. I0302 12:01:54.058316 22486551396480 xla_bridge.py:353] Unable to initialize backend 'tpu_driver':...

I seemed to have solved the issue with the version of jax and jaxlib. Now I do not get the rocm and plugin errors. But the run still gets killed...

I was finally only able to run the python script `run_alphafold.py` using the following parameters ``` python run_alphafold.py \ --fasta_paths=../alphafold_input/IFT57.fasta \ --output_dir=../alphafold_output \ --parameter_path=../alphafold_data/params/ \ --uniref90_database_path=../alphafold_data/uniref90/uniref90.fasta \ --mgnify_database_path=../alphafold_data/mgnify/mgy_clusters_2018_12.fa \ --template_mmcif_dir=../alphafold_data/pdb_mmcif/mmcif_files/...