https://github.com/novasush/parallelization-with-tfds
We are going to use CPU for Extract , Transform and Load, and GPU for training model parallelly
https://github.com/novasush/parallelization-with-tfds
tensorflow tensorflow-data-pipeline tensorflow-dataset tensorflow-dataset-api tensorflow-datasets tensorflow-examples tensorflow-experiments tensorflow-models tensorflow-tutorials tensorflow2
Last synced: 3 months ago
JSON representation
We are going to use CPU for Extract , Transform and Load, and GPU for training model parallelly
- Host: GitHub
- URL: https://github.com/novasush/parallelization-with-tfds
- Owner: novasush
- License: mit
- Created: 2020-03-07T02:29:57.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2020-03-09T14:58:03.000Z (over 5 years ago)
- Last Synced: 2025-04-07T16:19:53.452Z (3 months ago)
- Topics: tensorflow, tensorflow-data-pipeline, tensorflow-dataset, tensorflow-dataset-api, tensorflow-datasets, tensorflow-examples, tensorflow-experiments, tensorflow-models, tensorflow-tutorials, tensorflow2
- Language: Jupyter Notebook
- Size: 22.5 KB
- Stars: 2
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# ETL data on CPU and simultaneously Training model on GPU using TFDS
### We are going to use CPU for Extract , Transform and Load data from Tensorflow Dataset, and parallelly GPU for training model parallelly.
## The code is available for the same in notebook file present in this notebook.
* This code will not work on Google Colab as you cannot use both cpu and gpu in colab.
* Make sure you are running this code on Machine containing both CPU and GPU and tensorflow with GPU support. You can find tensorflow gpu installation documentation [here](https://towardsdatascience.com/how-to-install-tensorflow-gpu-on-ubuntu-18-04-1c1d2d6d6fd2).
* If you are getting
```python
ValueError: No data provided for "input_1". Need data for each key in: ['input_1']
```
Then you are either running code on google colab or on a machine which has either cpu or gpu only.
You can also refer my blog from [here](https://medium.com/@novasush/data-pipelining-for-effective-use-of-cpu-and-gpu-with-tensorflow-and-tfds-3c9cad39a03e).