Nick Becker

icon indicating a website link www.linkedin.com/in/nicholasebecker/

icon company @NVIDIA icon location United States icon location Building an accelerated data science and data engineering ecosystem.

Results 180 comments of Nick Becker

Adding support for cuML's nn-descent with UMAP

> @beckernick So if I'm not mistaken, going to fit_transform should already solve a big part of the problem, right? If so, then this should be a quick fix. Nice!...

[BUG]: Nx-cuGraph caching not working with documentation example

It looks like [our docs](https://docs.rapids.ai/api/cugraph/nightly/nx_cugraph/nx_cugraph/) are wrong (screenshot below). We [should be using](https://networkx.org/documentation/stable/reference/backends.html) `NETWORKX_CACHE_CONVERTED_GRAPHS` not `CACHE_CONVERTED_GRAPH` to configure caching behavior.

[BUG] n_neighbors should be smaller than the graph degree computed by nn descent

Thanks for reporting. Would you be able to give this a try with the stable v24.10 release that just came out? I'm not able to reproduce it in the current...

[BUG] n_neighbors should be smaller than the graph degree computed by nn descent

Closing, but please feel free to re-open if this reproduces in v24.10+

[BUG] n_neighbors should be smaller than the graph degree computed by nn descent

@cjnolet , do you expect this is as straighforward as adding a `graph_degree = max(graph_degree, n_neighbors)` for this codepath?

[BUG] n_neighbors should be smaller than the graph degree computed by nn descent

Thanks Corey. @saucechef , for now, you can do this manually on your side if you're using more neighbors than the default graph degree of 64 by using the `build_kwds`...

[FEA]: nx-cugraph should intercept built-in constructors like `from_pandas_edgelist` if `NETWORKX_BACKEND_PRIORITY=cugraph`

Seems like this would be an expansion of https://github.com/networkx/networkx/pull/6876

[DOC] In-depth documentation for Handles and Streams

This discussion came up again today, as DBSCAN Python docs indicate it's possible to use Handles/streams for concurrency. We noted several points: - There wasn't a handy example of how...

Compile times are growing significantly

@jrhemstad @harrism , is this still a relevant issue?

Slower performance than expected for TPC-H SF100 Q01

Sounds good. Wondering if perhaps I'm hitting some thread contention during IO? If I limit DuckDB to a single thread, I see similar performance. ```python import duckdb import chdb duckdb_con...