in_silico_perturbation memory requirement

#5
by DYXDAVE - opened

While running the in_silico_perturbation ipynb, it seems cannot load the model. What is a recommend memory for doing the perturbation? Thanks

Thank you for your interest in Geneformer. This is the dataset repository for the pretraining corpus Genecorpus-30M. Are you asking about loading the model (Geneformer) or the pretraining dataset (Genecorpus-30M)?

If you are asking about the pretraining dataset, Genecorpus-30M is a large dataset so loading it for the first time may take some time. The Hugging Face Datasets format is designed to cache the data though so future usage will be faster in that this cache allows Datasets to avoid re-processing the entire dataset with each use. Hugging Face Datasets are in the Apache Arrow format so datasets are processed with zero-copy reads to avoid any memory constraints. Therefore, loading the dataset should not be memory-limited.

If you are instead asking about the model, please open the issue in the Geneformer model repository so the answer can be associated with that repository instead of the dataset one.

ctheodoris changed discussion status to closed

Sign up or log in to comment