diff --git a/fastdeploy/worker/gcu_model_runner.py b/fastdeploy/worker/gcu_model_runner.py index a3624431be5..045b4833f4d 100644 --- a/fastdeploy/worker/gcu_model_runner.py +++ b/fastdeploy/worker/gcu_model_runner.py @@ -835,7 +835,7 @@ def _dummy_run( ) post_process( - sampler_output=sampler_output, + sampler_or_pooler_output=sampler_output, model_output=model_output_data, share_inputs=self.share_inputs, block_size=self.cache_config.block_size, @@ -1071,7 +1071,7 @@ class at the server level, which is too granular for ModelRunner. else: skip_save_output = False post_process( - sampler_output=sampler_output, + sampler_or_pooler_output=sampler_output, model_output=model_output_data, share_inputs=self.share_inputs, block_size=self.cache_config.block_size,