Teddy/final mac examples

* finetuning image classification final

* finetuning tabular classification final

* finetuning text classification final

* predict image final

flash/core/data/datamodule.py

@@ -51,10 +51,11 @@ def __init__(
 
         self.batch_size = batch_size
 
+        # TODO: figure out best solution for setting num_workers
+        # if num_workers is None:
+        #    num_workers = os.cpu_count()
         if num_workers is None:
-            num_workers = os.cpu_count()
-        if num_workers is None:
-            warnings.warn("Could not infer cpu count automatically, setting it to zero")
+        #    warnings.warn("Could not infer cpu count automatically, setting it to zero")
             num_workers = 0
         self.num_workers = num_workers
 

flash_examples/finetuning/image_classification.py

@@ -1,11 +1,11 @@
 # import our libraries
-import pytorch_lightning as pl
 import torch
+import flash
 from flash.vision import ImageClassificationData, ImageClassifier
-from flash.vision.classification.dataset import hymenoptera_data_download
+from flash.core.data import download_data
 
 # 1. Download data
-hymenoptera_data_download('data/')
+download_data("https://pl-flash-data.s3.amazonaws.com/hymenoptera_data.zip", 'data/')
 
 # 2. Organize our data
 datamodule = ImageClassificationData.from_folders(
@@ -17,10 +17,10 @@
 model = ImageClassifier(num_classes=datamodule.num_classes)
 
 # 4. Create trainer
-trainer = pl.Trainer(max_epochs=1)
+trainer = flash.Trainer(max_epochs=1)
 
 # 5. Train the model
-trainer.fit(model, datamodule=datamodule)
+trainer.finetune(model, datamodule=datamodule)
 
 # 6. Save model
 torch.save(model, "image_classification_model.pt")

flash_examples/finetuning/tabular_classification.py

@@ -1,18 +1,17 @@
 # import our libraries
-import pandas as pd
-import pytorch_lightning as pl
 import torch
 from pytorch_lightning.metrics.classification import Accuracy, Precision, Recall
 
+import flash
 from flash.tabular import TabularClassifier, TabularData
-from flash.tabular.classification.data.dataset import titanic_data_download
+from flash.core.data import download_data
 
 # 1. Download data
-titanic_data_download("./data/titanic")
+download_data("https://pl-flash-data.s3.amazonaws.com/titanic.zip", 'data/')
 
 # 2. Organize our data - create a LightningDataModule
-datamodule = TabularData.from_df(
-    pd.read_csv("./data/titanic/titanic.csv"),
+datamodule = TabularData.from_csv(
+    "./data/titanic/titanic.csv",
     categorical_input=["Sex", "Age", "SibSp", "Parch", "Ticket", "Cabin", "Embarked"],
     numerical_input=["Fare"],
     target="Survived",
@@ -23,7 +22,7 @@
 model = TabularClassifier.from_data(datamodule, metrics=[Accuracy(), Precision(), Recall()])
 
 # 4. Create trainer
-trainer = pl.Trainer(max_epochs=1)
+trainer = flash.Trainer(max_epochs=1)
 
 # 5. Train model
 trainer.fit(model, datamodule=datamodule)

flash_examples/finetuning/text_classification.py

@@ -1,7 +1,7 @@
 # import our libraries
-import pytorch_lightning as pl
 import torch
 
+import flash
 from flash.core.data import download_data
 from flash.text import TextClassificationData, TextClassifier
 
@@ -20,10 +20,10 @@
 model = TextClassifier(num_classes=datamodule.num_classes)
 
 # 4. Create trainer - Make training slightly faster for demo.
-trainer = pl.Trainer(max_epochs=1, limit_train_batches=8, limit_val_batches=8)
+trainer = flash.Trainer(max_epochs=1, limit_train_batches=8, limit_val_batches=8)
 
 # 5. Finetune the model
-trainer.fit(model, datamodule=datamodule)
+trainer.finetune(model, datamodule=datamodule)
 
 # 6. Save model
 torch.save(model, "text_classification_model.pt")

flash_examples/predict/image_classification.py

@@ -1,7 +1,7 @@
 # import our libraries
 import torch
-from pytorch_lightning import Trainer
 
+from flash import Trainer
 from flash.vision import ImageClassificationData
 
 # 1. Load model from checkpoint