https://github.com/nvidia-merlin/dataloader

The merlin dataloader lets you rapidly load tabular data for training deep leaning models with TensorFlow, PyTorch or JAX
https://github.com/nvidia-merlin/dataloader

deep-learning jax pytorch recommender-systems tensorflow

Last synced: 4 days ago
JSON representation

The merlin dataloader lets you rapidly load tabular data for training deep leaning models with TensorFlow, PyTorch or JAX

• Host: GitHub
• URL: https://github.com/nvidia-merlin/dataloader
• Owner: NVIDIA-Merlin
• License: apache-2.0
• Created: 2022-07-08T20:10:04.000Z (over 2 years ago)
• Default Branch: main
• Last Pushed: 2024-04-16T17:25:23.000Z (7 months ago)
• Last Synced: 2024-11-01T13:46:51.085Z (12 days ago)
• Topics: deep-learning, jax, pytorch, recommender-systems, tensorflow
• Language: Python
• Homepage:
• Size: 28.7 MB
• Stars: 406
• Watchers: 18
• Forks: 25
• Open Issues: 21
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# [Merlin Dataloader](https://github.com/NVIDIA-Merlin/dataloader)

![PyPI - Python Version](https://img.shields.io/pypi/pyversions/merlin-dataloader)

[![PyPI version shields.io](https://img.shields.io/pypi/v/merlin-dataloader.svg)](https://pypi.python.org/pypi/merlin-dataloader/)

![GitHub License](https://img.shields.io/github/license/NVIDIA-Merlin/dataloader)

[![Documentation](https://img.shields.io/badge/documentation-blue.svg)](https://nvidia-merlin.github.io/dataloader/stable/README.html)

The merlin-dataloader lets you quickly train recommender models for TensorFlow, PyTorch and JAX. It eliminates the biggest bottleneck in training recommender models, by providing GPU optimized dataloaders that read data directly into the GPU, and then do a 0-copy transfer to TensorFlow and PyTorch using [dlpack](https://github.com/dmlc/dlpack).

The benefits of the Merlin Dataloader include:

- Over 10x speedup over native framework dataloaders

- Handles larger than memory datasets

- Per-epoch shuffling

- Distributed training

## Installation

Merlin-dataloader requires Python version 3.7+. Additionally, GPU support requires CUDA 11.0+.

To install using Conda:

```

conda install -c nvidia -c rapidsai -c numba -c conda-forge merlin-dataloader python=3.7 cudatoolkit=11.2

```

To install from PyPi:

```

pip install merlin-dataloader

```

There are also [docker containers on NGC](https://nvidia-merlin.github.io/Merlin/stable/containers.html) with the merlin-dataloader and dependencies included on them

## Basic Usage

```python

# Get a merlin dataset from a set of parquet files

import merlin.io

dataset = merlin.io.Dataset(PARQUET_FILE_PATHS, engine="parquet")

# Create a Tensorflow dataloader from the dataset, loading 65K items

# per batch

from merlin.dataloader.tensorflow import Loader

loader = Loader(dataset, batch_size=65536)

# Get a single batch of data. Inputs will be a dictionary of columnname

# to TensorFlow tensors

inputs, target = next(loader)

# Train a Keras model with the dataloader

model = tf.keras.Model( ... )

model.fit(loader, epochs=5)

```