Flash CLI and Flash Zero (#611)

* Use the LightningCLI in the image classification example * FlashCLI * Finetune support * Port LightningCLI * Update requirements * Initial commit * Updates * Updates * Updates * Temp fill reqs * Bump PL req * Test * Update * Update * Remove debug code * [pre-commit.ci] auto fixes from pre-commit.com hooks for more information, see https://pre-commit.ci * Updates * Add speech recognition * Add text classification * Add tabular and seq2seq * Remove extra reqs * Fix test * Add pointcloud * Add graph * Fix test * A fix * Try fix * Try fix * Try fix * Add tests * Add click CLI * Add click CLI * Add some docs * Update docs * Punctuation * Add some tests * Update CHANGELOG.md * Fix test * Add some tests * Test * Updates * Try fix Co-authored-by: Carlos Mocholi <[email protected]> Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>
Lightning-Universe · Aug 4, 2021 · 8e42d39 · 8e42d39
1 parent f9d3348
commit 8e42d39
Show file tree

Hide file tree

Showing 78 changed files with 3,101 additions and 54 deletions.
diff --git a/.gitignore b/.gitignore
@@ -161,3 +161,5 @@ jigsaw_toxic_comments
 flash_examples/serve/tabular_classification/data
 logs/cache/*
 flash_examples/data
+flash_examples/cli/*/data
+timit/
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -32,6 +32,8 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 - Added a `SpeechRecognition` task for speech to text using Wav2Vec ([#586](https://github.com/PyTorchLightning/lightning-flash/pull/586))
 
+- Added Flash Zero, a zero code command line ML platform built with flash ([#611](https://github.com/PyTorchLightning/lightning-flash/pull/611))
+
 ### Changed
 
 - Changed how pretrained flag works for loading weights for ImageClassifier task ([#560](https://github.com/PyTorchLightning/lightning-flash/pull/560))
@@ -46,6 +48,8 @@ The format is based on [Keep a Changelog](http://keepachangelog.com/en/1.0.0/).
 
 - Fixed a bug where an uncaught ValueError could be raised when checking if a module is available ([#615](https://github.com/PyTorchLightning/lightning-flash/pull/615))
 
+- Fixed a bug where some tasks were not compatible with PyTorch 1.7 due to use of `torch.jit.isinstance` ([#611](https://github.com/PyTorchLightning/lightning-flash/pull/611))
+
 ## [0.4.0] - 2021-06-22
 
 ### Added

diff --git a/docs/source/general/flash_zero.rst b/docs/source/general/flash_zero.rst
@@ -0,0 +1,56 @@
+.. _flash_zero:
+
+**********
+Flash Zero
+**********
+
+Flash Zero is a zero-code machine learning platform built directly into lightning-flash.
+To get started and view the available tasks, run:
+
+.. code-block:: bash
+
+    flash --help
+
+Customize Trainer and Model arguments
+_____________________________________
+
+Flash Zero is built on top of the
+`lightning CLI <https://pytorch-lightning.readthedocs.io/en/stable/common/lightning_cli.html>`_, so the trainer and
+model arguments can be configured either from the command line or from a config file.
+For example, to run the image classifier for 10 epochs with a `resnet50` backbone you can use:
+
+.. code-block:: bash
+
+    flash image-classification --trainer.max_epochs 10 --model.backbone resnet50
+
+To view all of the available options for a task, run:
+
+.. code-block:: bash
+
+    flash image-classification --help
+
+Using Custom Data
+_________________
+
+Flash Zero works with your own data through subcommands. The available subcommands for each task are given at the bottom
+of their help pages (e.g. when running :code:`flash image-classification --help`). You can then use the required
+subcommand to train on your own data. Let's look at an example using the Hymenoptera data from the
+:ref:`image_classification` guide. First, download and unzip your data:
+
+.. code-block:: bash
+
+    curl https://pl-flash-data.s3.amazonaws.com/hymenoptera_data.zip -o hymenoptera_data
+    unzip hymenoptera_data.zip
+
+Now train with Flash Zero:
+
+.. code-block:: bash
+
+    flash image-classification from_folders --train_folder ./hymenoptera_data/train
+
+You can view the help page for each subcommand. For example, to view the options for training an image classifier from
+folders, you can run:
+
+.. code-block:: bash
+
+    flash image-classification from_folders --help
diff --git a/docs/source/index.rst b/docs/source/index.rst
@@ -26,6 +26,7 @@ Lightning Flash
    general/jit
    general/data
    general/registry
+   general/flash_zero
    general/serve
 
 .. toctree::

diff --git a/docs/source/reference/audio_classification.rst b/docs/source/reference/audio_classification.rst
@@ -71,3 +71,22 @@ Here's the full example:
 .. literalinclude:: ../../../flash_examples/audio_classification.py
     :language: python
     :lines: 14-
+
+------
+
+**********
+Flash Zero
+**********
+
+The audio classifier can be used directly from the command line with zero code using :ref:`flash_zero`.
+You can run the above example with:
+
+.. code-block:: bash
+
+    flash audio-classification
+
+To view configuration options and options for running the audio classifier with your own data, use:
+
+.. code-block:: bash
+
+    flash audio-classification --help
diff --git a/docs/source/reference/graph_classification.rst b/docs/source/reference/graph_classification.rst
@@ -31,3 +31,22 @@ Here's the full example:
 .. literalinclude:: ../../../flash_examples/graph_classification.py
     :language: python
     :lines: 14-
+
+------
+
+**********
+Flash Zero
+**********
+
+The graph classifier can be used directly from the command line with zero code using :ref:`flash_zero`.
+You can run the above example with:
+
+.. code-block:: bash
+
+    flash graph-classifier
+
+To view configuration options and options for running the graph classifier with your own data, use:
+
+.. code-block:: bash
+
+    flash graph-classifier --help
diff --git a/docs/source/reference/image_classification.rst b/docs/source/reference/image_classification.rst
@@ -57,6 +57,25 @@ Here's the full example:
 
 ------
 
+**********
+Flash Zero
+**********
+
+The image classifier can be used directly from the command line with zero code using :ref:`flash_zero`.
+You can run the hymenoptera example with:
+
+.. code-block:: bash
+
+    flash image-classification
+
+To view configuration options and options for running the image classifier with your own data, use:
+
+.. code-block:: bash
+
+    flash image-classification --help
+
+------
+
 *******
 Serving
 *******

diff --git a/docs/source/reference/image_classification_multi_label.rst b/docs/source/reference/image_classification_multi_label.rst
@@ -49,6 +49,27 @@ Here's the full example:
     :language: python
     :lines: 14-
 
+
+------
+
+**********
+Flash Zero
+**********
+
+The multi-label image classifier can be used directly from the command line with zero code using :ref:`flash_zero`.
+You can run the movie posters example with:
+
+.. code-block:: bash
+
+    flash image-classification from_movie_posters
+
+To view configuration options and options for running the image classifier with your own data, use:
+
+.. code-block:: bash
+
+    flash image-classification --help
+
+
 ------
 
 *******

diff --git a/docs/source/reference/object_detection.rst b/docs/source/reference/object_detection.rst
@@ -47,3 +47,22 @@ Here's the full example:
 .. literalinclude:: ../../../flash_examples/object_detection.py
     :language: python
     :lines: 14-
+
+------
+
+**********
+Flash Zero
+**********
+
+The object detector can be used directly from the command line with zero code using :ref:`flash_zero`.
+You can run the above example with:
+
+.. code-block:: bash
+
+    flash object-detection
+
+To view configuration options and options for running the object detector with your own data, use:
+
+.. code-block:: bash
+
+    flash object-detection --help
diff --git a/docs/source/reference/pointcloud_object_detection.rst b/docs/source/reference/pointcloud_object_detection.rst
@@ -76,7 +76,24 @@ Here's the full example:
     :language: python
     :lines: 14-
 
-
-
 .. image:: https://raw.githubusercontent.com/intel-isl/Open3D-ML/master/docs/images/visualizer_BoundingBoxes.png
    :width: 100%
+
+------
+
+**********
+Flash Zero
+**********
+
+The point cloud object detector can be used directly from the command line with zero code using :ref:`flash_zero`.
+You can run the above example with:
+
+.. code-block:: bash
+
+    flash pointcloud-detection
+
+To view configuration options and options for running the point cloud object detector with your own data, use:
+
+.. code-block:: bash
+
+    flash pointcloud-detection --help
diff --git a/docs/source/reference/pointcloud_segmentation.rst b/docs/source/reference/pointcloud_segmentation.rst
@@ -67,7 +67,24 @@ Here's the full example:
     :language: python
     :lines: 14-
 
-
-
 .. image:: https://raw.githubusercontent.com/intel-isl/Open3D-ML/master/docs/images/getting_started_ml_visualizer.gif
    :width: 100%
+
+------
+
+**********
+Flash Zero
+**********
+
+The point cloud segmentation task can be used directly from the command line with zero code using :ref:`flash_zero`.
+You can run the above example with:
+
+.. code-block:: bash
+
+    flash pointcloud-segmentation
+
+To view configuration options and options for running the point cloud segmentation task with your own data, use:
+
+.. code-block:: bash
+
+    flash pointcloud-segmentation --help
diff --git a/docs/source/reference/semantic_segmentation.rst b/docs/source/reference/semantic_segmentation.rst
@@ -44,6 +44,27 @@ Here's the full example:
     :language: python
     :lines: 14-
 
+
+------
+
+**********
+Flash Zero
+**********
+
+The semantic segmentation task can be used directly from the command line with zero code using :ref:`flash_zero`.
+You can run the above example with:
+
+.. code-block:: bash
+
+    flash semantic-segmentation
+
+To view configuration options and options for running the semantic segmentation task with your own data, use:
+
+.. code-block:: bash
+
+    flash semantic-segmentation --help
+
+
 ------
 
 *******

diff --git a/docs/source/reference/speech_recognition.rst b/docs/source/reference/speech_recognition.rst
@@ -49,6 +49,25 @@ Here's the full example:
 
 ------
 
+**********
+Flash Zero
+**********
+
+The speech recognition task can be used directly from the command line with zero code using :ref:`flash_zero`.
+You can run the above example with:
+
+.. code-block:: bash
+
+    flash speech-recognition
+
+To view configuration options and options for running the speech recognition task with your own data, use:
+
+.. code-block:: bash
+
+    flash speech-recognition --help
+
+------
+
 *******
 Serving
 *******

diff --git a/docs/source/reference/style_transfer.rst b/docs/source/reference/style_transfer.rst
@@ -33,3 +33,22 @@ Here's the full example:
 .. literalinclude:: ../../../flash_examples/style_transfer.py
     :language: python
     :lines: 14-
+
+------
+
+**********
+Flash Zero
+**********
+
+The style transfer task can be used directly from the command line with zero code using :ref:`flash_zero`.
+You can run the above example with:
+
+.. code-block:: bash
+
+    flash style-transfer
+
+To view configuration options and options for running the style transfer task with your own data, use:
+
+.. code-block:: bash
+
+    flash style-transfer --help
diff --git a/docs/source/reference/summarization.rst b/docs/source/reference/summarization.rst
@@ -49,6 +49,25 @@ Here's the full example:
 
 ------
 
+**********
+Flash Zero
+**********
+
+The summarization task can be used directly from the command line with zero code using :ref:`flash_zero`.
+You can run the above example with:
+
+.. code-block:: bash
+
+    flash summarization
+
+To view configuration options and options for running the summarization task with your own data, use:
+
+.. code-block:: bash
+
+    flash summarization --help
+
+------
+
 *******
 Serving
 *******

diff --git a/docs/source/reference/tabular_classification.rst b/docs/source/reference/tabular_classification.rst
@@ -48,6 +48,25 @@ Here's the full example:
 
 ------
 
+**********
+Flash Zero
+**********
+
+The tabular classifier can be used directly from the command line with zero code using :ref:`flash_zero`.
+You can run the above example with:
+
+.. code-block:: bash
+
+    flash tabular-classifier
+
+To view configuration options and options for running the tabular classifier with your own data, use:
+
+.. code-block:: bash
+
+    flash tabular-classifier --help
+
+------
+
 *******
 Serving
 *******