test: CI convert llama3.1

.github/workflows/convert-model.yml

@@ -25,16 +25,16 @@ on:
 
 env:
   USER_NAME: cortexso
-  SOURCE_MODEL_ID: google/gemma-2-2b-it #${{ inputs.source_model_id }}
-  SOURCE_MODEL_SIZE: 2b #${{ inputs.source_model_size }}
-  TARGET_MODEL_ID: gemma2 #${{ inputs.target_model_id }}
+  SOURCE_MODEL_ID: meta-llama/Meta-Llama-3.1-8B-Instruct #${{ inputs.source_model_id }}
+  SOURCE_MODEL_SIZE: 8b #${{ inputs.source_model_size }}
+  TARGET_MODEL_ID: llama3.1 #${{ inputs.target_model_id }}
   PRECISION: int4                   # Valid values: int4,fp16,fp3
   EXECUTOR: dml                     # Valid values: cpu,cuda,dml,web
   ONNXRUNTIME_GENAI_VERSION: 0.3.0  # Check version from: https://github.com/microsoft/onnxruntime-genai/releases
 
 jobs:
   converter:
-    runs-on: windows-amd
+    runs-on: windows-cuda-12-0
     steps:
       - name: Checkout
         uses: actions/checkout@v4 # v4.1.7
@@ -131,7 +131,7 @@ jobs:
           huggingface-cli upload ${{ env.USER_NAME }}/${{ env.TARGET_MODEL_ID }} "${{ env.MODEL_NAME }}/onnx" . --revision "${{ env.SOURCE_MODEL_SIZE }}-onnx"
           huggingface-cli logout
 
-      # - name: Cleanup
-      #   if: always()
-      #   run: |
-      #     Remove-Item -Recurse -Force -Path "${{ env.MODEL_NAME }}"
+      - name: Cleanup
+        if: always()
+        run: |
+          Remove-Item -Recurse -Force -Path "${{ env.MODEL_NAME }}"