update and add CI/CD for benchmark the model

.github/workflows/test-models.yml

@@ -1,4 +1,4 @@
-name: Test - Models
+name: Test and Benchmark Models
 on:
   workflow_dispatch:
     inputs:
@@ -17,9 +17,14 @@ on:
         required: false
         default: "--mode audio --num_rows 5"
         type: string
+      run_benchmark:
+        description: 'Run benchmark test'
+        required: false
+        default: false
+        type: boolean
 
 jobs:
-  run-test:
+  run-test-and-benchmark:
     runs-on: research
     steps:
       - name: Checkout
@@ -34,4 +39,24 @@ jobs:
       - name: Run tests
         working-directory: ./tests
         run: |
-          python3 test_case.py --model_dir ${{ github.event.inputs.model_id }} --data_dir ${{ github.event.inputs.dataset_id }}  ${{ github.event.inputs.extra_args }}
+          python3 test_case.py --model_dir ${{ github.event.inputs.model_id }} --data_dir ${{ github.event.inputs.dataset_id }} ${{ github.event.inputs.extra_args }}
+
+      - name: Install benchmark dependencies
+        if: ${{ github.event.inputs.run_benchmark == 'true' }}
+        working-directory: ./lm-evaluation-harness
+        run: |
+          pip3 install -e .
+
+      - name: Run benchmark
+        if: ${{ github.event.inputs.run_benchmark == 'true' }}
+        working-directory: ./lm-evaluation-harness
+        run: |
+          chmod +x ./run_benchmark.sh
+          ./run_benchmark.sh ${{ github.event.inputs.model_id }}
+
+      - name: Upload benchmark results
+        if: ${{ github.event.inputs.run_benchmark == 'true' }}
+        uses: actions/upload-artifact@v2
+        with:
+          name: benchmark-results
+          path: ./lm-evaluation-harness/benchmark_results/*.json