Remove Inputs Key from Truss Templates (#244)

* the actual change

* oops rm extra files

* create docs

* fix up all tests for new approach

* update select examples to demonstrate input-free approach

* tests and pre/post processing

* rm any

Author: philipkiely-baseten

README.md

@@ -71,7 +71,7 @@ rfc.fit(data_x, data_y)
 tr = truss.create(rfc, target_directory="iris_rfc_truss")
 
 # Serve a prediction from the model
-tr.predict({"inputs": [[0, 0, 0, 0]]})
+tr.predict([[0, 0, 0, 0]])
 ```
 
 ### Package your model
@@ -80,6 +80,7 @@ The `truss.create()` command can be used with any supported framework:
 
 * [Hugging Face](https://truss.baseten.co/create/huggingface)
 * [LightGBM](https://truss.baseten.co/create/lightgbm)
+* [MLflow](https://truss.baseten.co/create/mlflow)
 * [PyTorch](https://truss.baseten.co/create/pytorch)
 * [scikit-learn](https://truss.baseten.co/create/sklearn)
 * [Tensorflow](https://truss.baseten.co/create/tensorflow)
@@ -98,7 +99,7 @@ truss run-image iris_rfc_truss
 Then, as long as the container is running, you can invoke the model as an API as follows:
 
 ```
-curl -X POST http://127.0.0.1:8080/v1/models/model:predict -d '{"inputs": [[0, 0, 0, 0]]}'
+curl -X POST http://127.0.0.1:8080/v1/models/model:predict -d '[[0, 0, 0, 0]]'
 ```
 
 ### Configure your model for deployment

docs/create/huggingface.md

@@ -13,6 +13,7 @@ If you're using a Jupyter notebook, add a line to install the `transformers` and
 {% hint style="warning" %}
 Truss officially supports `transformers` version 4.21.0 or higher. Especially if you're using an online notebook environment like Google Colab or a bundle of packages like Anaconda, ensure that the version you are using is supported. If it's not, use the `--upgrade` flag and pip will install the most recent version.
 {% endhint %}
+
 ### Create an in-memory model
 
 This is the part you want to replace with your own code. Using a Hugging Face transformer, build a machine learning model and keep it in-memory. In this example we're using [bert-base-uncased](https://huggingface.co/bert-base-uncased), which will fill in the missing word in a sentence.
@@ -44,7 +45,7 @@ Check the target directory to see your new Truss!
 To get a prediction from the Truss, try running:
 
 ```python
-tr.predict({"inputs": ["Donatello is a teenage mutant [MASK] turtle"]})
+tr.predict("Donatello is a teenage mutant [MASK] turtle")
 ```
 
 For more on running the Truss locally, see [local development](../develop/localhost.md).

docs/create/lightgbm.md

@@ -66,7 +66,7 @@ Check the target directory to see your new Truss!
 To get a prediction from the Truss, try running:
 
 ```python
-tr.predict({"inputs": [[0, 0, 0, 0, 0, 0]]})
+tr.predict([[0, 0, 0, 0, 0, 0]])
 ```
 
 For more on running the Truss locally, see [local development](../develop/localhost.md).

docs/create/mlflow.md

@@ -53,8 +53,7 @@ To get a prediction from the Truss, try running:
 
 ```python
 data = np.array([-4, 1, 0, 10, -2, 1]).reshape(-1, 1)
-predictions = tr.predict({"inputs": data})
-print(predictions)
+tr.predict(data)
 ```
 
 For more on running the Truss locally, see [local development](../develop/localhost.md).

docs/create/pytorch.md

@@ -165,7 +165,7 @@ inputs = datasets.MNIST("../data", train=False, transform=transform)
 dataset = torch.utils.data.DataLoader(inputs, batch_size=1)
 
 import numpy as np
-print(tr.predict({"inputs": np.array(next(iter(dataset))[0])}))
+print(tr.predict(np.array(next(iter(dataset))[0])))
 ```
 
 For information on running the Truss locally, see [local development](../develop/localhost.md).

docs/create/sklearn.md

@@ -45,7 +45,7 @@ Check the target directory to see your new Truss!
 To get a prediction from the Truss, try running:
 
 ```python
-tr.predict({"inputs": [[0, 0, 0, 0]]})
+tr.predict([[0, 0, 0, 0]])
 ```
 
 For more on running the Truss locally, see [local development](../develop/localhost.md).

docs/create/tensorflow.md

@@ -44,19 +44,24 @@ Check the target directory to see your new Truss!
 
 ### Serve the model
 
-The TensorFlow model requires [pre- and post-processing functions](../develop/processing.md) to run. These functions go in `model/model.py`:
+In your newly created Truss, open `model/model.py` and add pre- and post-processing functions as follows.
+
+First, add the following imports at the top of the file:
+
 
 ```python
-#Preprocess and Postprocess Functions
 import requests
 import tempfile
-import numpy as np
-
+import tensorflow as tf
 from scipy.special import softmax
+```
+
+Then, update the pre-processing function to:
 
-def preprocess(url):
+```python
+def preprocess(self, model_input: Any) -> Any:
     """Preprocess step for ResNet"""
-    request = requests.get(url)
+    request = requests.get(model_input)
     with tempfile.NamedTemporaryFile() as f:
         f.write(request.content)
         f.seek(0)
@@ -65,22 +70,32 @@ def preprocess(url):
         tf.image.resize([input_image], (224, 224))
     )
     return np.array(preprocessed_image)
+```
+
+Finally, update the post-processing function to:
 
-def postprocess(predictions, k=5):
+```python
+def postprocess(self, model_output: Dict, k=5) -> Dict:
     """Post process step for ResNet"""
-    class_predictions = predictions[0]
+    class_predictions = model_output["predictions"][0]
     LABELS = requests.get(
-        'https://storage.googleapis.com/download.tensorflow.org/data/ImageNetLabels.txt'
-    ).text.split('\n')
+        "https://storage.googleapis.com/download.tensorflow.org/data/ImageNetLabels.txt"
+    ).text.split("\n")
     class_probabilities = softmax(class_predictions)
     top_probability_indices = class_probabilities.argsort()[::-1][:k].tolist()
-    return {LABELS[index]: 100 * class_probabilities[index].round(3) for index in top_probability_indices}
+    return {
+        LABELS[index]: 100 * class_probabilities[index].round(3)
+        for index in top_probability_indices
+    }
 ```
 
 With these functions in place, you can invoke the model and pass it a URL, as in:
 
 ```python
-tr.predict({"inputs": "https://github.com/pytorch/hub/raw/master/images/dog.jpg"})
+from truss import load
+
+tr = load("tensorflow_truss")
+tr.predict("https://github.com/pytorch/hub/raw/master/images/dog.jpg")
 ```
 
 For information on running the Truss locally, see [local development](../develop/localhost.md).

docs/create/xgboost.md

@@ -61,7 +61,7 @@ Check the target directory to see your new Truss!
 To get a prediction from the Truss, try running:
 
 ```python
-tr.predict({"inputs": [[0, 0, 0, 0, 0, 0]]})
+tr.predict([[0, 0, 0, 0, 0, 0]])
 ```
 
 For more on running the Truss locally, see [local development](../develop/localhost.md).

docs/deploy/aws.md

@@ -91,5 +91,5 @@ Now that you're task is running, you can make requests to your model! To get the
 If you've been following along with the model above, you can use the snippet below to make a sample request.
 
 ```
-curl -H 'Content-Type: application/json' -d '{"inputs": [[0,0,0,0]]}' -X POST [CONTAINER_LINK]:80/v1/models/model:predict
+curl -H 'Content-Type: application/json' -d '[[0,0,0,0]]' -X POST [CONTAINER_LINK]:80/v1/models/model:predict
 ```

docs/develop/localhost.md

@@ -27,15 +27,15 @@ tr = truss.load("path_to_my_truss")
 From there, you can invoke the Truss to serve the model in your Python environment. Just run:
 
 ```python
-tr.predict({"inputs": [[0, 0, 0, 0]]})
+tr.predict([[0, 0, 0, 0]])
 ```
 
 ### Command line interface
 
 Alternately, you can run your Truss from the command line.
 
 ```
-truss predict path_to_my_truss '{"inputs": [[0, 0, 0, 0]]}'
+truss predict path_to_my_truss '[[0, 0, 0, 0]]'
 ```
 
 You can also specify examples for the model and run them instead. It's much easier to express request data in the example file. Running the example provides for a good dev loop.
@@ -51,7 +51,7 @@ truss run-image path_to_my_truss
 Then, as long as the container is running, you can invoke the model as an API as follows:
 
 ```
-curl -X POST http://127.0.0.1:8080/v1/models/model:predict -d '{"inputs": [[0, 0, 0, 0]]}'
+curl -X POST http://127.0.0.1:8080/v1/models/model:predict -d '{[[0, 0, 0, 0]]'
 ```
 
 ## Setting up local dev
@@ -69,7 +69,7 @@ This is the standard way to run a Truss. It creates a Docker image, runs the Tru
 To run in Docker:
 
 ```python
-tr.docker_predict({"inputs": [[0, 0, 0, 0]]})
+tr.docker_predict([[0, 0, 0, 0]])
 ```
 
 ### Faster dev loop with live reload
@@ -112,11 +112,11 @@ Unlike Docker image, this mechanism requires that you already have the right Pyt
 In the Python environment, get a prediction without Docker by running:
 
 ```python
-tr.predict({"inputs": [[0, 0, 0, 0]]})
+tr.predict([[0, 0, 0, 0]])
 ```
 
 Or in the command line, run:
 
 ```python
-truss predict --run-local path_to_my_truss '{"inputs": [[0, 0, 0, 0]]}'
+truss predict --run-local path_to_my_truss '[[0, 0, 0, 0]]'
 ```

docs/develop/processing.md

@@ -19,9 +19,9 @@ import requests
 import tempfile
 import numpy as np
 
-def preprocess(url):
+def preprocess(self, model_input: Any) -> Any:
     """Preprocess step for ResNet"""
-    request = requests.get(url)
+    request = requests.get(model_input)
     with tempfile.NamedTemporaryFile() as f:
         f.write(request.content)
         f.seek(0)
@@ -41,13 +41,16 @@ Here is a post-processing function from the same [TensorFlow example](../create/
 ```python
 from scipy.special import softmax
 
-def postprocess(predictions, k=5):
+def postprocess(self, model_output: Dict, k=5) -> Dict:
     """Post process step for ResNet"""
-    class_predictions = predictions[0]
+    class_predictions = model_output["predictions"][0]
     LABELS = requests.get(
-        'https://storage.googleapis.com/download.tensorflow.org/data/ImageNetLabels.txt'
-    ).text.split('\n')
+        "https://storage.googleapis.com/download.tensorflow.org/data/ImageNetLabels.txt"
+    ).text.split("\n")
     class_probabilities = softmax(class_predictions)
     top_probability_indices = class_probabilities.argsort()[::-1][:k].tolist()
-    return {LABELS[index]: 100 * class_probabilities[index].round(3) for index in top_probability_indices}
+    return {
+        LABELS[index]: 100 * class_probabilities[index].round(3)
+        for index in top_probability_indices
+    }
 ```

docs/notebooks/huggingface_example.ipynb

@@ -59,7 +59,7 @@
       "outputs": [],
       "source": [
         "# Serve a prediction from the model\n",
-        "tr.predict({\"inputs\": [\"Donatello is a teenage mutant [MASK] turtle\"]})"
+        "tr.predict(\"Donatello is a teenage mutant [MASK] turtle\")"
       ]
     }
   ],
@@ -69,17 +69,25 @@
       "provenance": []
     },
     "kernelspec": {
-      "display_name": "Python 3",
+      "display_name": "truss-qbHCcytf-py3.9",
       "language": "python",
       "name": "python3"
     },
     "language_info": {
+      "codemirror_mode": {
+        "name": "ipython",
+        "version": 3
+      },
+      "file_extension": ".py",
+      "mimetype": "text/x-python",
       "name": "python",
-      "version": "3.8.9 (default, Apr 13 2022, 08:48:06) \n[Clang 13.1.6 (clang-1316.0.21.2.5)]"
+      "nbconvert_exporter": "python",
+      "pygments_lexer": "ipython3",
+      "version": "3.9.16"
     },
     "vscode": {
       "interpreter": {
-        "hash": "31f2aee4e71d21fbe5cf8b01ff0e069b9275f58929596ceb00d14d90e3e16cd6"
+        "hash": "71eaf67bb821e55ec2e9ffe63f35d72d7631d0e358c20043383b5ebba2b8daf8"
       }
     }
   },

docs/notebooks/lightgbm_example.ipynb

@@ -81,7 +81,7 @@
       },
       "outputs": [],
       "source": [
-        "tr.predict({\"inputs\": [[0, 0, 0, 0, 0, 0]]})"
+        "tr.predict([[0, 0, 0, 0, 0, 0]])"
       ]
     }
   ],
@@ -92,11 +92,26 @@
       "provenance": []
     },
     "kernelspec": {
-      "display_name": "Python 3",
+      "display_name": "truss-qbHCcytf-py3.9",
+      "language": "python",
       "name": "python3"
     },
     "language_info": {
-      "name": "python"
+      "codemirror_mode": {
+        "name": "ipython",
+        "version": 3
+      },
+      "file_extension": ".py",
+      "mimetype": "text/x-python",
+      "name": "python",
+      "nbconvert_exporter": "python",
+      "pygments_lexer": "ipython3",
+      "version": "3.9.16"
+    },
+    "vscode": {
+      "interpreter": {
+        "hash": "71eaf67bb821e55ec2e9ffe63f35d72d7631d0e358c20043383b5ebba2b8daf8"
+      }
     }
   },
   "nbformat": 4,

docs/notebooks/mlflow_example.ipynb

@@ -99,8 +99,7 @@
         "# Invoke the MLflow model\n",
         "\n",
         "data = np.array([-4, 1, 0, 10, -2, 1]).reshape(-1, 1)\n",
-        "predictions = tr.predict({\"inputs\": data})\n",
-        "print(predictions)"
+        "tr.predict(data)"
       ]
     }
   ],
@@ -109,17 +108,25 @@
       "provenance": []
     },
     "kernelspec": {
-      "display_name": "Python 3.8.9 64-bit",
+      "display_name": "truss-qbHCcytf-py3.9",
       "language": "python",
       "name": "python3"
     },
     "language_info": {
+      "codemirror_mode": {
+        "name": "ipython",
+        "version": 3
+      },
+      "file_extension": ".py",
+      "mimetype": "text/x-python",
       "name": "python",
-      "version": "3.8.9"
+      "nbconvert_exporter": "python",
+      "pygments_lexer": "ipython3",
+      "version": "3.9.16"
     },
     "vscode": {
       "interpreter": {
-        "hash": "31f2aee4e71d21fbe5cf8b01ff0e069b9275f58929596ceb00d14d90e3e16cd6"
+        "hash": "71eaf67bb821e55ec2e9ffe63f35d72d7631d0e358c20043383b5ebba2b8daf8"
       }
     }
   },