Restructure iris example

cli/cmd/get.go

@@ -490,7 +490,7 @@ func describeAPI(name string, resourcesRes *schema.GetResourcesResponse, flagVer
 	apiEndpoint := urls.Join(resourcesRes.APIsBaseURL, anyAPIStatus.Path)
 
 	out := "\nURL:      " + apiEndpoint + "\n"
-	out += fmt.Sprintf("cURL:     curl -k -X POST -H \"Content-Type: application/json\" %s -d @<json_file_path>\n", apiEndpoint)
+	out += fmt.Sprintf("cURL:     curl -k -X POST -H \"Content-Type: application/json\" %s -d @samples.json\n", apiEndpoint)
 	out += "\n"
 	out += fmt.Sprintf("Status:              %s\n", groupStatus.Message())
 	out += fmt.Sprintf("Available replicas:  %s\n", s.Int32(groupStatus.Available()))

docs/apis/packaging-models.md

@@ -47,10 +47,10 @@ with open("sklearn.onnx", "wb") as f:
 
 Here are examples of converting models from some of the common ML frameworks to ONNX:
 
-* [PyTorch](https://github.com/cortexlabs/cortex/blob/master/examples/iris/pytorch/model.py)
-* [Sklearn](https://github.com/cortexlabs/cortex/blob/master/examples/iris/sklearn/model.py)
-* [XGBoost](https://github.com/cortexlabs/cortex/blob/master/examples/iris/xgboost/model.py)
-* [Keras](https://github.com/cortexlabs/cortex/blob/master/examples/iris/keras/model.py)
+* [PyTorch](https://github.com/cortexlabs/cortex/blob/master/examples/iris/models/pytorch_model.py)
+* [Sklearn](https://github.com/cortexlabs/cortex/blob/master/examples/iris/models/sklearn_model.py)
+* [XGBoost](https://github.com/cortexlabs/cortex/blob/master/examples/iris/models/xgboost_model.py)
+* [Keras](https://github.com/cortexlabs/cortex/blob/master/examples/iris/models/keras_model.py)
 
 Upload your trained model in ONNX format to Amazon S3:
 

examples/iris/cortex.yaml

@@ -8,19 +8,19 @@
 - kind: api
   name: pytorch
   model: s3://cortex-examples/iris/pytorch.onnx
-  request_handler: pytorch/handler.py
+  request_handler: handlers/pytorch.py
 
 - kind: api
   name: xgboost
   model: s3://cortex-examples/iris/xgboost.onnx
-  request_handler: xgboost/handler.py
+  request_handler: handlers/xgboost.py
 
 - kind: api
   name: sklearn
   model: s3://cortex-examples/iris/sklearn.onnx
-  request_handler: sklearn/handler.py
+  request_handler: handlers/sklearn.py
 
 - kind: api
   name: keras
   model: s3://cortex-examples/iris/keras.onnx
-  request_handler: keras/handler.py
+  request_handler: handlers/keras.py

examples/iris/keras/handler.py → examples/iris/handlers/keras.py

@@ -3,6 +3,17 @@
 iris_labels = ["Iris-setosa", "Iris-versicolor", "Iris-virginica"]
 
 
+def pre_inference(sample, metadata):
+    return {
+        metadata[0].name: [
+            sample["sepal_length"],
+            sample["sepal_width"],
+            sample["petal_length"],
+            sample["petal_width"],
+        ]
+    }
+
+
 def post_inference(prediction, metadata):
     probabilites = prediction[0][0]
     predicted_class_id = int(np.argmax(probabilites))

examples/iris/handlers/xgboost.py

@@ -0,0 +1,19 @@
+import numpy as np
+
+iris_labels = ["Iris-setosa", "Iris-versicolor", "Iris-virginica"]
+
+
+def pre_inference(sample, metadata):
+    return {
+        metadata[0].name: [
+            sample["sepal_length"],
+            sample["sepal_width"],
+            sample["petal_length"],
+            sample["petal_width"],
+        ]
+    }
+
+
+def post_inference(prediction, metadata):
+    predicted_class_id = prediction[0][0]
+    return {"class_label": iris_labels[predicted_class_id], "class_index": predicted_class_id}

examples/iris/models/requirements.txt

@@ -0,0 +1,9 @@
+keras
+keras2onnx
+tensorflow
+torch
+pandas
+skl2onnx
+sklearn
+onnxmltools
+xgboost