core
core copied to clipboard
NVIDIA-Merlin
Core Utilities for NVIDIA Merlin
Ref: https://github.com/NVIDIA-Merlin/NVTabular/issues/1875 This adds a relatively-aggressive upper pin for now (right before the dask-expr migration that will certainly break most of Merlin).
This PR changes how we determine if cuda is available on the system. We move from numba to using HAS_GPU which uses nvml device count. If there are no devices,...
As pointed out by @oliverholworthy in https://github.com/NVIDIA-Merlin/core/pull/274#discussion_r1160187901, `cuda_isavailable()` [is used in `merlin.core.compat`](https://github.com/NVIDIA-Merlin/core/blob/6e52b48140615708b59926b5f9c3601f8feeab93/merlin/core/compat/__init__.py#L24) to check for cuda support. Unfortunately, [this is a known problem](https://github.com/NVIDIA-Merlin/core/issues/144) for dask-cuda. This most likely means that...
Add workflow to create dev tag on creation of release branch Depends on: https://github.com/NVIDIA-Merlin/.github/pull/19
Enable TagSet to be used in Schema selection methods. ```python tags = TagSet(["a", "b"]) schema.select_by_tag(tags) ``` This is motivated by schema selection functionality in the Merlin Models PyTorch API, where...
### Bug description When training the all recommender systems models with NVIDIA Merlin at: ... history = model.fit( train_data, epochs=3, batch_size=512, pre=predict_last ) ... **runtime error: "ValueError: high is out...
Add GitHub Actions Workflow job to run NVTabular unit tests on GPU. _Motivation_: To help us catch issues with GPU device functionality in NVTabular that relies on 'core' before a...
https://github.com/NVIDIA-Merlin/core/blob/e1eaf2694b402bf2ea99a78d3953d59e8a00e35c/requirements.txt#L16 This project [depends on pynvml](https://github.com/NVIDIA-Merlin/core/blob/e1eaf2694b402bf2ea99a78d3953d59e8a00e35c/requirements.txt#L16), where the nvidia-blessed bindings to nvml are in the [nvidia-ml-py](https://pypi.org/project/nvidia-ml-py/) package instead. This is causing problems for the [gpustat project](https://github.com/wookayin/gpustat/issues/153#issuecomment-1551036105), which needs some internals...
Split Dimension into fixed and ragged to support specifying variable sized dense tensors