From 8c167e130cc34b65c3e828e9571027b33b3cabd6 Mon Sep 17 00:00:00 2001 From: EnflameGCU <118410644+EnflameGCU@users.noreply.github.com> Date: Fri, 25 Jul 2025 11:03:03 +0800 Subject: [PATCH] [GCU] Update post_process (#3012) --- fastdeploy/worker/gcu_model_runner.py | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/fastdeploy/worker/gcu_model_runner.py b/fastdeploy/worker/gcu_model_runner.py index e08d6760e..63bf6e7b3 100644 --- a/fastdeploy/worker/gcu_model_runner.py +++ b/fastdeploy/worker/gcu_model_runner.py @@ -772,6 +772,8 @@ class GCUModelRunner(ModelRunnerBase): post_process( sampler_output=sampler_output, model_output=model_output_data, + share_inputs=self.share_inputs, + block_size=self.parallel_config.block_size, speculative_decoding=self.speculative_decoding, skip_save_output=True, ) @@ -989,6 +991,8 @@ class GCUModelRunner(ModelRunnerBase): post_process( sampler_output=sampler_output, model_output=model_output_data, + share_inputs=self.share_inputs, + block_size=self.parallel_config.block_size, save_each_rank=self.parallel_config.use_ep, speculative_decoding=self.speculative_decoding, skip_save_output=skip_save_output,