Divye Gala

icon location New York, NY icon location Software Engineer at NVIDIA RAPIDS. Enjoy working with accelerated Machine Learning and Graph Analytics, real-time inference, and high throughput systems.

Results 120 comments of Divye Gala

CUDA-MEMCHECK reports Uninitialized access

Just confirmed this behavior on CUDA 10.2 for `thrust::min_max`

`allow_single_cluster=True` gets overwritten for `cluster_selection_method=leaf`

@lmcinnes what is the expected behavior? I'm happy to submit a patch/fix

[QST] Why does scikit HDBSCAN return different results when we compare with CuML's HDBSCAN?

@jcfaracco the first intuition I have is that your `min_samples` is really low. Can you try increasing it? If your data is really dense, it is possible that the first...

[FEA] Add Cross Validators to cuml

@heytanay you should be able to use scikit-learn's cross validators directly with a cuML model. Can you try that and let me know if it works?

[FEA] Add Cross Validators to cuml

@beckernick correct me if I am wrong, but shouldn't we be able to get away by using CuPy arrays, since they have the same mechanisms as NumPy arrays?

[FEA] Add Cross Validators to cuml

@heytanay we would absolutely welcome a contribution from your side here. Let me know if I can be a resource to you in any way during your PR process, be...

[FEA] Add Cross Validators to cuml

@heytanay yep, I don't foresee any need of CUDA here. You should be able to leverage features from cuML or our dependencies to directly build this feature out in Python.

CPU/GPU interoperability POC

rerun tests

[BUG] LabelEncoder returns cudf array even when output_type="numpy"

@tvdboom in the meantime, could you tell what version of RAPIDS you are using?

[BUG] Building cuml encounters failure due to dependency requirement on cumlprims_mg

@farhanrahman I believe you only have to install those dependencies if you are setting up the dependency tree yourself. Installing the dev environment is enough to get all needed dependencies,...