Add MLEM model

.github/workflows/deploy-model-with-mlem.yml

@@ -0,0 +1,49 @@
+name: Deploy MLEM model after GTO Stage assignment
+on:
+  push:
+    tags:
+      - "*"
+env:
+  HEROKU_API_KEY: ${{ secrets.HEROKU_API_KEY }}
+  AWS_ACCESS_KEY_ID: ${{ secrets.AWS_ACCESS_KEY_ID }}
+  AWS_SECRET_ACCESS_KEY: ${{ secrets.AWS_SECRET_ACCESS_KEY }}
+
+jobs:
+  parse-git-tag:
+    name: Figure out what was registered/promoted
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@v3
+      with:
+        fetch-depth: 0
+    - name: "GTO: figure out what was registered/promoted and show the Registry state"
+      id: gto
+      uses: iterative/gto-action@v1
+      with:
+        show: true
+        history: true
+    # we define the Job outputs here to let the next Job use them
+    outputs:
+      name: ${{ steps.gto.outputs.name }}
+      stage: ${{ steps.gto.outputs.stage }}
+      event: ${{ steps.gto.outputs.event }}
+      path: ${{ steps.gto.outputs.path }}
+  deploy-model:
+    name: Deploy a MLEM model (act on assigning a new stage)
+    needs: parse-git-tag
+    if: needs.parse-git-tag.outputs.event == 'assignment'
+    runs-on: ubuntu-latest
+    steps:
+    - uses: actions/checkout@v3
+      with:
+        fetch-depth: 0
+    - uses: actions/setup-python@v2
+      with:
+        python-version: '3.10'
+    - name: Install dependencies
+      run: |
+        pip install --upgrade pip setuptools wheel
+        pip install -r requirements.txt
+    - name: Run `mlem deploy``
+      run: |
+        mlem deployment run --load deploy/${{ needs.parse-git-tag.outputs.stage }} --model ${{ needs.parse-git-tag.outputs.path }}

.mlem.yaml

@@ -0,0 +1,3 @@
+core:
+ state:
+   uri: s3://gto-mlem-example/mlem-deployment-state

deploy/dev.mlem

@@ -0,0 +1,3 @@
+object_type: deployment
+type: heroku
+app_name: mlem-dev

deploy/prod.mlem

@@ -0,0 +1,3 @@
+object_type: deployment
+type: heroku
+app_name: mlem-prod

deploy/staging.mlem

@@ -0,0 +1,3 @@
+object_type: deployment
+type: heroku
+app_name: mlem-staging

models/churn.pkl.mlem

@@ -0,0 +1,37 @@
+artifacts:
+  data.pkl:
+    hash: 2979af88eef40c69250d61efef3d94a0
+    size: 319
+    uri: churn.pkl
+model_type:
+  io:
+    type: pickle
+  methods:
+    __call__:
+      args:
+      - name: data
+        type_:
+          ptype: str
+          type: primitive
+      name: model
+      returns:
+        ptype: str
+        type: primitive
+    predict:
+      args:
+      - name: data
+        type_:
+          ptype: str
+          type: primitive
+      name: predict
+      returns:
+        ptype: str
+        type: primitive
+  type: callable
+object_type: model
+requirements:
+- is_package: false
+  module: __main__
+  name: __main__
+  source64zip: eJw1jtEKwyAMRd/9ipCnFoajrwP3I2OEsKadTK2oLezvp5Xdl+SGey6xPm6pQP5mpWzfvRNfzQJEgb0QgTGARJ5tIMKbgqqD3S5gGqg5rcdjekJFnIThfxrhDhOIywIYtk6gOulZFvDbLG6YufDYK5uSlD2Fnu3R9o3maHXmQ4YTugCeM19f75rW8ePwApl9dEKt0GAuyYYVR/UDFWhG0Q==
+  type: custom

requirements.txt

@@ -1 +1,2 @@
-gto
+gto
+mlem[s3,fastapi,heroku]

train.py

@@ -0,0 +1,11 @@
+import sys
+
+import mlem
+
+if __name__ == "__main__":
+    value = sys.argv[1] if len(sys.argv) > 1 else "no value"
+
+    def model(data):
+        return value
+
+    mlem.api.save(model, "models/churn.pkl", sample_data="string")