try out an integration test workflow

.github/workflows/integration_tests.yaml

@@ -0,0 +1,41 @@
+name: integration-tests 
+on:
+  workflow_call:
+    inputs:
+      lorax_tag:
+        required: true
+        type: string
+    secrets:
+      RUNPOD_API_KEY:
+        required: true
+jobs:
+  deploy:
+    runs-on: ubuntu-latest
+    steps:
+      - uses: actions/checkout@v4
+      - name: Install runpod
+        run: |
+          wget -qO- cli.runpod.net | sudo bash
+      - name: Config runpod cli 
+        run: |
+          runpodctl config --apiKey ${{ secrets.RUNPOD_API_KEY }}
+
+      - uses: actions/setup-python@v4
+        with:
+          python-version: 3.x
+
+      # Install the client 
+      - run: pip install -e clients/python
+      - name: create the pod and then delete the pod
+        run: |
+          bash tests/create-pod.sh ${{ inputs.lorax_tag }}
+
+      - name: Run tests 
+        run: |
+          POD_ID=`cat pod_name.txt`
+          python3 tests/test.py $POD_ID
+
+      - name: Delete the pod 
+        run: |
+          POD_ID=`cat pod_name.txt`
+          runpodctl remove pod $POD_ID

tests/create-pod.sh

@@ -0,0 +1,10 @@
+IMAGE_NAME="$1"
+
+runpodctl create pods \
+  --name lorax-tests-new \
+  --gpuType "NVIDIA A40" \
+  --imageName "$IMAGE_NAME" \
+  --containerDiskSize 100 \
+  --volumeSize 100 \
+  --ports "8080/http" \
+  --args "--port 8080 --model-id predibase/Mistral-7B-v0.1-dequantized --adapter-source hub --default-adapter-source pbase --max-batch-prefill-tokens 32768 --max-total-tokens 8192 --max-input-length 8191 --max-concurrent-requests 1024" | awk '{print $2}' > pod_name.txt

tests/test.py

@@ -0,0 +1,9 @@
+from lorax import Client
+import sys
+
+pod_id = sys.argv[1]
+
+client = Client(f"https://{pod_id}-8080.proxy.runpod.net")
+
+response = client.generate("hello!", max_new_tokens=10)
+print(response)