HelloWorldLTY
HelloWorldLTY
Hi, would you please fix this issue? Thanks.
Hi, I think you should run k-means with n_cluster=50 to initiliaze it. In their version of ICLR submission, they use PRCL with k-means to initiliaze the cluster.
Hi, did you resolve this problem? Do you know what is the minimial file requirement for using this model for inference? Thanks
Hi, for their example tcga_luadlusc dataset, I find a similar conclusion. The acc or other metrics are all 1 for UNI v2 and over 0.96 for UNI v1. {'lin_acc': 1.0,...
Moreover, I found that I met nan when training GEARs with own datasets What could be the possible reasons? Thanks. ``` Start Training... Epoch 1 Step 1 Train Loss: 0.2541...
Hi, I used datasets from this paper (https://www.biorxiv.org/content/10.1101/2024.07.03.601903v1.abstract) which contains two large-scale perturbation datasets. I follow the recommended pre-processing step and verify that there are no nan values. Thanks.
I have the same question here, did you try to use other versions of pytorch and make it compilable with h100? ---update---- I have figured out the approach to use...
Actually autocast cannot handle the mode correctly, which is really wired. Any updates about this problem?
Same question here, I also wonder if it is possible to access datasets and paired labels used in this figure:  Thanks a lot!!!
Hi, I think this idea is helpful and I do face many OOM issues when testing it with gpt-4o-mini.