Hi guys, we really enjoyed your Nicheformer paper and were excited to try it out. When running it on a dataset of 10,000 cells and 958 genes, it took about 2 hours to generate embeddings. Is this runtime expected? Do you have any tips for speeding it up?
We are using an H100 GPU. We used the get_embeddings() function from the experiments folder in the git repo, as suggested by Alejandro.
Hi guys, we really enjoyed your Nicheformer paper and were excited to try it out. When running it on a dataset of 10,000 cells and 958 genes, it took about 2 hours to generate embeddings. Is this runtime expected? Do you have any tips for speeding it up?
We are using an H100 GPU. We used the get_embeddings() function from the experiments folder in the git repo, as suggested by Alejandro.