Fix

Author: winskuo-quic

.ci/scripts/test_llama.sh

@@ -206,11 +206,17 @@ prepare_artifacts_upload() {
   fi
 }
 
-# Download llama artifacts
-download_stories_model_artifacts
-# Create tokenizer.bin.
-echo "Creating tokenizer.bin"
-$PYTHON_EXECUTABLE -m extension.llm.tokenizer.tokenizer -t tokenizer.model -o tokenizer.bin
+# Download and create artifacts.
+PARAMS="params.json"
+CHECKPOINT_FILE_NAME=""
+touch "${PARAMS}"
+if [[ "${MODEL_NAME}" == "llama" ]] || [[ "${MODEL_NAME}" == "stories"* ]] || [[ "${MODEL_NAME}" == "tinyllama" ]]; then
+  CHECKPOINT_FILE_NAME="stories110M.pt"
+  download_stories_model_artifacts
+else
+  echo "Unsupported model name ${MODEL_NAME}"
+  exit 1
+fi
 
 # Check dtype.
 EXPORTED_MODEL_NAME="tinyllama_${MODE}_${DTYPE}"

.ci/scripts/test_qnn_static_llama.sh

@@ -20,13 +20,13 @@ echo "Creating tokenizer.bin"
 $PYTHON_EXECUTABLE -m extension.llm.tokenizer.tokenizer -t tokenizer.model -o tokenizer.bin
 
 export EXECUTORCH_ROOT="$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")/.." && pwd)"
-  export QNN_SDK_ROOT=/tmp/qnn/2.28.0.241029
-  export LD_LIBRARY_PATH="${QNN_SDK_ROOT}/lib/x86_64-linux-clang"
-  export PYTHONPATH=".."
-  cp schema/program.fbs exir/_serialize/program.fbs
-  cp schema/scalar_type.fbs exir/_serialize/scalar_type.fbs
-  cp -f build-x86/backends/qualcomm/PyQnnManagerAdaptor.cpython-310-x86_64-linux-gnu.so backends/qualcomm/python
-  cp -f build-x86/backends/qualcomm/PyQnnWrapperAdaptor.cpython-310-x86_64-linux-gnu.so backends/qualcomm/python
+export QNN_SDK_ROOT=/tmp/qnn/2.28.0.241029
+export LD_LIBRARY_PATH="${QNN_SDK_ROOT}/lib/x86_64-linux-clang"
+export PYTHONPATH=".."
+cp schema/program.fbs exir/_serialize/program.fbs
+cp schema/scalar_type.fbs exir/_serialize/scalar_type.fbs
+cp -f build-x86/backends/qualcomm/PyQnnManagerAdaptor.cpython-310-x86_64-linux-gnu.so backends/qualcomm/python
+cp -f build-x86/backends/qualcomm/PyQnnWrapperAdaptor.cpython-310-x86_64-linux-gnu.so backends/qualcomm/python
 
 
 # Compile only as weight sharing is not applicable on x86