From fbf0f7c50b4ec5a0cc3852365deacd869f8068af Mon Sep 17 00:00:00 2001 From: Gahow Wang Date: Sun, 19 Jul 2026 11:47:36 +0800 Subject: [PATCH] Set Qwen235 profile KV cache layout --- .../run_qwen235_v020_true_mixed_profiles.sh | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/runs/frontier-fidelity-envelope-v1/run_qwen235_v020_true_mixed_profiles.sh b/runs/frontier-fidelity-envelope-v1/run_qwen235_v020_true_mixed_profiles.sh index 6f7e47c..7afb48f 100755 --- a/runs/frontier-fidelity-envelope-v1/run_qwen235_v020_true_mixed_profiles.sh +++ b/runs/frontier-fidelity-envelope-v1/run_qwen235_v020_true_mixed_profiles.sh @@ -24,7 +24,7 @@ cd "${FRONTIER_SOURCE}" profile() { local gpu="$1" method="$2" tp="$3" root="$4" - env CUDA_VISIBLE_DEVICES="${gpu}" PYTHONPATH="${FRONTIER_SOURCE}" \ + env CUDA_VISIBLE_DEVICES="${gpu}" VLLM_KV_CACHE_LAYOUT=NHD PYTHONPATH="${FRONTIER_SOURCE}" \ "${VENV_ROOT}/bin/python" -m frontier.profiling.attention.main \ --disable_ray --num_gpus 1 --device h20 --output_dir "${root}" --models "${MODEL}" \ --num_tensor_parallel_workers "${tp}" --max_model_len 40960 --max_seq_len 40960 \