code review fixes

SeldonIO · Jun 11, 2020 · f10d22c · f10d22c
1 parent 6acdcaa
commit f10d22c
Show file tree

Hide file tree

Showing 14 changed files with 230 additions and 185 deletions.
diff --git a/doc/source/reference/apis/metadata.md b/doc/source/reference/apis/metadata.md
@@ -3,6 +3,7 @@
 ![metadata](./metadata.svg)
 
 
+
 ## Incubating feature note
 The model metadata feature has currently "incubating" status.
 This means that we are currently exploring the best possible interface and functionality for this feature.
@@ -15,7 +16,8 @@ Incubating update 1:
 - definition through environmental variables now accepts both `yaml` and `json` input
 
 
-## Examples:
+
+## Examples
 - [Basic Examples for Model with Metadata](../../examples/metadata.html)
 - [SKLearn Server example with MinIO](../../examples/minio-sklearn.html)
 - [Deployment Level Metadata](../../examples/graph-metadata.html).
@@ -24,16 +26,153 @@ Incubating update 1:
 - [Deploying models trained with DVC](../../examples/dvc.html)
 
 
+
 ## Model Metadata (incubating)
 
 With Seldon you can easily add metadata to your models.
+
+### Prepackaged model servers
+
+To add metadata to your prepackaged model servers simply add a file named `metadata.yaml`
+to the S3 bucket with your model:
+```YAML
+name: my-model
+versions: [my-model/v1]
+platform: platform-name
+inputs:
+- datatype: BYTES
+  name: input
+  shape: [ 1, 4 ]
+outputs:
+- datatype: BYTES
+  name: output
+  shape: [ 3 ]
+```
+
+See [SKLearn Server example with MinIO](../../examples/minio-sklearn.html) for more details.
+
+
+### Python Language Wrapper
+
+You can add model metadata you your custom Python model by implementing `init_metadata` method:
+
+```python
+class Model:
+    ...
+    def init_metadata(self):
+        meta = {
+            "name": "my-model",
+            "versions": ["my-model/v1"],
+            "platform": "platform-name",
+            "inputs": [{"name": "input", "datatype": "BYTES", "shape": [1, 4]}],
+            "outputs": [{"name": "output", "datatype": "BYTES", "shape": [1]}],
+        }
+        return meta
+```
+
+See [Python wrapper](../../python/python_component.html#incubating-features) documentation for more details and
+notebook [Basic Examples for Model with Metadata](../../examples/metadata.html).
+
+
+### Overwrite via environmental variable
+
+You can also always specify `MODEL_METADATA` environmental variable which takes ultimate priority.
+
+```YAML
+apiVersion: machinelearning.seldon.io/v1
+kind: SeldonDeployment
+metadata:
+  name: seldon-model
+spec:
+  name: test-deployment
+  predictors:
+  - componentSpecs:
+    - spec:
+        containers:
+        - name: my-model
+          image: ...
+          env:
+          - name: MODEL_METADATA
+            value: |
+              ---
+              name: my-model-name
+              versions: [ my-model-version ]
+              platform: seldon
+              inputs:
+              - datatype: BYTES
+                name: input
+                shape: [ 1, 4 ]
+              outputs:
+              - datatype: BYTES
+                name: output
+                shape: [ 3 ]
+    graph:
+      name: my-model
+      ...
+    name: example
+    replicas: 1
+```
+
+## Deployment Metadata (incubating)
+Model metadata allow you to specify metadata for each of the components (nodes) in your graph.
+New orchestrator engine will probe all nodes for their metadata and derive global `inputs` and `outputs` of your graph.
+It will then expose them together with all nodes' metadata at a single endpoint `/api/v1.0/metadata/` of your deployment.
+
+![graph-metadata](./graph-metadata.svg)
+
+Example response:
+```json
+{
+  "name": "example",
+  "models": {
+    "model-1": {
+      "name": "Model 1",
+      "platform": "platform-name",
+      "versions": ["model-version"],
+      "inputs": [{"datatype": "BYTES", "name": "input", "shape": [1, 5]}],
+      "outputs": [{"datatype": "BYTES", "name": "output", "shape": [1, 3]}]
+    },
+    "model-2": {
+      "name": "Model 2",
+      "platform": "platform-name",
+      "versions": ["model-version"],
+      "inputs": [{"datatype": "BYTES", "name": "input", "shape": [1, 3]}],
+      "outputs": [{"datatype": "BYTES", "name": "output", "shape": [3]}]
+    }
+  },
+  "graphinputs": [{"datatype": "BYTES", "name": "input", "shape": [1, 5]}],
+  "graphoutputs": [{"datatype": "BYTES", "name": "output", "shape": [3]}]
+}
+```
+
+See example [notebook](../../examples/graph-metadata.html) for more details.
+
+
+
+## Metadata endpoint
+
+Model metadata can be obtained through GET request at `/api/v1.0/metadata/{MODEL_NAME}` endpoint of your deployment.
+
+Example response:
+```json
+{
+  "name": "my-model",
+  "versions": ["my-model/v1"],
+  "platform": "platform-name",
+  "inputs": [{"datatype": "BYTES", "name": "input", "shape": [1, 5]}],
+  "outputs": [{"datatype": "BYTES", "name": "output", "shape": [1, 3]}],
+}
+```
+
+
+## V1/V2 format reference
+
 You can add metadata using one of two formats:
-- `v1` format that closely correlate to the current structure of `SeldonMessage`
-- `v2` format that is future-proof and fully compatible with `kfserving` dataplane proposal
+- `v1` format that closely correlates to the current structure of `SeldonMessage`
+- `v2` format that is future-proof and fully compatible with [kfserving dataplane proposal](https://github.com/kubeflow/kfserving/blob/master/docs/predict-api/v2/required_api.md#model-metadata) dataplane proposal
 
 As you will see in following sections difference is minimal and mostly relates to the input/output format.
 
-
 ### V1 DataPlane
 
 In order to use `v1` metadata format you need to specify `apiVersion: v1`.
@@ -43,7 +182,7 @@ In order to use `v1` metadata format you need to specify `apiVersion: v1`.
 apiVersion: v1
 name: my-model-name
 versions: [ my-model-version-01 ]
-platform: seldon-custom
+platform: seldon
 inputs:
   datatype: array
   shape: [ 2, 2 ]
@@ -65,7 +204,7 @@ and
 apiVersion: v1
 name: my-model-name
 versions: [ my-model-version-01 ]
-platform: seldon-custom
+platform: seldon
 inputs:
   datatype: jsonData
   schema:
@@ -94,13 +233,12 @@ The `schema` field is optional and can leaves user total freedom over its struct
 
 Note: as you can see you can mix inputs and outputs of different types!
 
-
 #### strData input/output
 ```YAML
 apiVersion: v1
 name: my-model-name
 versions: [ my-model-version-01 ]
-platform: seldon-custom
+platform: seldon
 inputs:
   datatype: strData
 outputs:
@@ -114,7 +252,6 @@ Example model input:
 ```
 
 
-
 ### V2 DataPlane
 You can easily define metadata for your models that is compatible with [kfserving dataplane proposal](https://github.com/kubeflow/kfserving/blob/master/docs/predict-api/v2/required_api.md#model-metadata) specification.
 ```
@@ -138,95 +275,3 @@ $metadata_tensor =
 ```
 
 Note: this the default format so you do not need to specify `apiVersion`.
-
-### Metadata endpoint
-
-Model metadata can be obtained through GET request at `/api/v1.0/metadata/{MODEL_NAME}` endpoint of your deployment.
-
-Example response:
-```json
-{
-  "name": "my-model",
-  "versions": ["my-model/v1"],
-  "platform": "platform-name",
-  "inputs": [{"datatype": "BYTES", "name": "input", "shape": [1, 5]}],
-  "outputs": [{"datatype": "BYTES", "name": "output", "shape": [1, 3]}],
-}
-```
-
-### Prepackaged model servers
-
-To add metadata to your prepackaged model servers simply add a file named `metadata.yaml`
-to the S3 bucket with your model:
-```YAML
-name: my-model
-versions: [my-model/v1]
-platform: platform-name
-inputs:
-- datatype: BYTES
-  name: input
-  shape: [ 1, 4 ]
-outputs:
-- datatype: BYTES
-  name: output
-  shape: [ 3 ]
-```
-
-See [SKLearn Server example with MinIO](../../examples/minio-sklearn.html) for more details.
-
-
-### Python Language Wrapper
-
-You can add model metadata you your custom Python model by implementing `init_metadata` method:
-
-```python
-class Model:
-    ...
-    def init_metadata(self):
-        meta = {
-            "name": "my-model",
-            "versions": ["my-model/v1"],
-            "platform": "platform-name",
-            "inputs": [{"name": "input", "datatype": "BYTES", "shape": [1, 4]}],
-            "outputs": [{"name": "output", "datatype": "BYTES", "shape": [1]}],
-        }
-        return meta
-```
-
-See [Python wrapper](../../python/python_component.html#incubating-features) documentation for more details and
-notebook [Basic Examples for Model with Metadata](../../examples/metadata.html).
-
-
-## Deployment Metadata (incubating)
-Model metadata allow you to specify metadata for each of the components (nodes) in your graph.
-New orchestrator engine will probe all nodes for their metadata and derive global `inputs` and `outputs` of your graph.
-It will then expose them together with all nodes' metadata at a single endpoint `/api/v1.0/metadata/` of your deployment.
-
-![graph-metadata](./graph-metadata.svg)
-
-Example response:
-```json
-{
-  "name": "example",
-  "models": {
-    "model-1": {
-      "name": "Model 1",
-      "platform": "platform-name",
-      "versions": ["model-version"],
-      "inputs": [{"datatype": "BYTES", "name": "input", "shape": [1, 5]}],
-      "outputs": [{"datatype": "BYTES", "name": "output", "shape": [1, 3]}]
-    },
-    "model-2": {
-      "name": "Model 2",
-      "platform": "platform-name",
-      "versions": ["model-version"],
-      "inputs": [{"datatype": "BYTES", "name": "input", "shape": [1, 3]}],
-      "outputs": [{"datatype": "BYTES", "name": "output", "shape": [3]}]
-    }
-  },
-  "graphinputs": [{"datatype": "BYTES", "name": "input", "shape": [1, 5]}],
-  "graphoutputs": [{"datatype": "BYTES", "name": "output", "shape": [3]}]
-}
-```
-
-See example [notebook](../../examples/graph-metadata.html) for more details.
diff --git a/examples/models/metadata/graph-metadata/combiner.yaml b/examples/models/metadata/graph-metadata/combiner.yaml
@@ -16,7 +16,7 @@ spec:
             value: |
               {"name": "node-combiner",
                "versions": ["generic-node/v0.2"],
-               "platform": "seldon-custom",
+               "platform": "seldon",
                "inputs": [
                    {"name": "input-1", "datatype": "BYTES", "shape": [1, 20]},
                    {"name": "input-2", "datatype": "BYTES", "shape": [1, 30]}
@@ -29,7 +29,7 @@ spec:
             value: |
               {"name": "node-one",
                "versions": ["generic-node/v0.2"],
-               "platform": "seldon-custom",
+               "platform": "seldon",
                "inputs": [{"name": "input", "datatype": "BYTES", "shape": [1, 10]}],
                "outputs": [{"name": "output", "datatype": "BYTES", "shape": [1, 20]}]}            
         - image: seldonio/metadata-generic-node:0.2
@@ -39,7 +39,7 @@ spec:
             value: |
               {"name": "node-two",
                "versions": ["generic-node/v0.2"],
-               "platform": "seldon-custom",
+               "platform": "seldon",
                "inputs": [{"name": "input", "datatype": "BYTES", "shape": [1, 10]}],
                "outputs": [{"name": "output", "datatype": "BYTES", "shape": [1, 30]}]}               
     graph:
@@ -53,4 +53,4 @@ spec:
         type: MODEL
         children: []   
     name: example
-    replicas: 1
+    replicas: 1
diff --git a/examples/models/metadata/graph-metadata/input-transformer.yaml b/examples/models/metadata/graph-metadata/input-transformer.yaml
@@ -16,7 +16,7 @@ spec:
             value: |
               {"name": "node-input-transformer",
                "versions": ["generic-node/v0.2"],
-               "platform": "seldon-custom",
+               "platform": "seldon",
                "inputs": [{"name": "input", "datatype": "BYTES", "shape": [1, 10]}],
                "outputs": [{"name": "output", "datatype": "BYTES", "shape": [1, 20]}]}            
         - image: seldonio/metadata-generic-node:0.2
@@ -26,7 +26,7 @@ spec:
             value: |
               {"name": "node",
                "versions": ["generic-node/v0.2"],
-               "platform": "seldon-custom",
+               "platform": "seldon",
                "inputs": [{"name": "input", "datatype": "BYTES", "shape": [1, 20]}],
                "outputs": [{"name": "output", "datatype": "BYTES", "shape": [1]}]}               
     graph:
@@ -37,4 +37,4 @@ spec:
         type: MODEL
         children: []   
     name: example
-    replicas: 1
+    replicas: 1
diff --git a/examples/models/metadata/graph-metadata/output-transformer.yaml b/examples/models/metadata/graph-metadata/output-transformer.yaml
@@ -16,7 +16,7 @@ spec:
             value: |
               {"name": "node-output-transformer",
                "versions": ["generic-node/v0.2"],
-               "platform": "seldon-custom",
+               "platform": "seldon",
                "inputs": [{"name": "input", "datatype": "BYTES", "shape": [1, 20]}],
                "outputs": [{"name": "output", "datatype": "BYTES", "shape": [1]}]}               
         - image: seldonio/metadata-generic-node:0.2
@@ -26,7 +26,7 @@ spec:
             value: |
               {"name": "node",
                "versions": ["generic-node/v0.2"],
-               "platform": "seldon-custom",
+               "platform": "seldon",
                "inputs": [{"name": "input", "datatype": "BYTES", "shape": [1, 10]}],
                "outputs": [{"name": "output", "datatype": "BYTES", "shape": [1, 20]}]}            
     graph:
@@ -37,4 +37,4 @@ spec:
         type: MODEL
         children: []   
     name: example
-    replicas: 1
+    replicas: 1