From cdf0403c64330a626950e9b0319ead6ae99704e5 Mon Sep 17 00:00:00 2001 From: Aurelien Chartier <2567591+achartier@users.noreply.github.com> Date: Fri, 24 Oct 2025 06:44:34 -0700 Subject: [PATCH] [None][feat] Pass KvCacheRetentionConfig to torch LlmRequest (#8634) Signed-off-by: Aurelien Chartier <2567591+achartier@users.noreply.github.com> --- tensorrt_llm/_torch/pyexecutor/llm_request.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/tensorrt_llm/_torch/pyexecutor/llm_request.py b/tensorrt_llm/_torch/pyexecutor/llm_request.py index 8030efc8b5..f5263c991b 100644 --- a/tensorrt_llm/_torch/pyexecutor/llm_request.py +++ b/tensorrt_llm/_torch/pyexecutor/llm_request.py @@ -764,7 +764,8 @@ def executor_request_to_llm_request( cache_salt_id=executor_request.cache_salt_id, arrival_time=getattr(executor_request, "py_arrival_time", None), py_multimodal_data=getattr(executor_request, "py_multimodal_data", - None)) + None), + kv_cache_retention_config=executor_request.kv_cache_retention_config) if child_req_ids: for child_id in child_req_ids: llm_request.create_child_request(child_id)