variable update

Author: winskuo-quic

backends/qualcomm/tests/test_qnn_delegate.py

@@ -907,7 +907,6 @@ def test_qnn_backend_example_models(self):
             # Fail during lowering Reopen once resolved
             # MobileBertModelExample(),
             # TorchVisionViTModel(),
-            # Encountered undefined symbol in mainline. Reopen once resolved.
             Wav2LetterModel(),
         ]
         expected_partitions = [

examples/qualcomm/oss_scripts/llama/llama.py

@@ -843,6 +843,7 @@ def post_process():
     )
 
     runner_cmd = ""
+    performance_output_path = "outputs/inference_speed.txt"
     if args.enable_x86_64:
         # x86 emulator is intended for CI and not performance. Check only the first few tokens.
         seq_len = min(seq_len, 16)
@@ -862,6 +863,7 @@ def post_process():
                 f"--model_path {pte_path}",
                 f"--seq_len {seq_len}",
                 f"--output_path {args.artifact}/outputs/outputs.txt",
+                f"--performance_output_path {performance_output_path}",
                 f"--kv_updater ShiftPointer",
                 runner_args,
             ]
@@ -874,7 +876,6 @@ def post_process():
         )
         post_process()
     else:
-        performance_output_path = "outputs/inference_speed.txt"
         runner_cmd = " ".join(
             [
                 f"cd {workspace} &&",