From 4db10d6dec8c8293df9cafe9781b990ac38440f6 Mon Sep 17 00:00:00 2001 From: Yueqian Lin <70319226+linyueqian@users.noreply.github.com> Date: Mon, 9 Mar 2026 03:42:15 -0400 Subject: [PATCH] Disable mm processor cache in CI stage configs (#1739) Signed-off-by: linyueqian --- .../e2e/offline_inference/stage_configs/npu/qwen2_5_omni_ci.yaml | 1 + tests/e2e/stage_configs/qwen2_5_omni_thinker_ci.yaml | 1 + tests/e2e/stage_configs/qwen3_omni_ci.yaml | 1 + tests/e2e/stage_configs/rocm/qwen2_5_omni_ci.yaml | 1 + tests/e2e/stage_configs/rocm/qwen3_omni_ci.yaml | 1 + 5 files changed, 5 insertions(+) diff --git a/tests/e2e/offline_inference/stage_configs/npu/qwen2_5_omni_ci.yaml b/tests/e2e/offline_inference/stage_configs/npu/qwen2_5_omni_ci.yaml index d6b04258a..87707e919 100644 --- a/tests/e2e/offline_inference/stage_configs/npu/qwen2_5_omni_ci.yaml +++ b/tests/e2e/offline_inference/stage_configs/npu/qwen2_5_omni_ci.yaml @@ -21,6 +21,7 @@ stage_args: trust_remote_code: true engine_output_type: latent enable_prefix_caching: false + mm_processor_cache_gb: 0 is_comprehension: true final_output: true final_output_type: text diff --git a/tests/e2e/stage_configs/qwen2_5_omni_thinker_ci.yaml b/tests/e2e/stage_configs/qwen2_5_omni_thinker_ci.yaml index 3ebf914e8..23eacafdf 100644 --- a/tests/e2e/stage_configs/qwen2_5_omni_thinker_ci.yaml +++ b/tests/e2e/stage_configs/qwen2_5_omni_thinker_ci.yaml @@ -18,6 +18,7 @@ stage_args: trust_remote_code: true engine_output_type: latent enable_prefix_caching: false + mm_processor_cache_gb: 0 is_comprehension: true final_output: true final_output_type: text diff --git a/tests/e2e/stage_configs/qwen3_omni_ci.yaml b/tests/e2e/stage_configs/qwen3_omni_ci.yaml index 641fc77ae..a557dffd8 100644 --- a/tests/e2e/stage_configs/qwen3_omni_ci.yaml +++ b/tests/e2e/stage_configs/qwen3_omni_ci.yaml @@ -22,6 +22,7 @@ stage_args: max_num_batched_tokens: 32768 max_model_len: 32768 enable_prefix_caching: false + mm_processor_cache_gb: 0 hf_config_name: thinker_config tensor_parallel_size: 1 load_format: dummy diff --git a/tests/e2e/stage_configs/rocm/qwen2_5_omni_ci.yaml b/tests/e2e/stage_configs/rocm/qwen2_5_omni_ci.yaml index ec815ad1f..d51f7a5c8 100644 --- a/tests/e2e/stage_configs/rocm/qwen2_5_omni_ci.yaml +++ b/tests/e2e/stage_configs/rocm/qwen2_5_omni_ci.yaml @@ -22,6 +22,7 @@ stage_args: trust_remote_code: true engine_output_type: latent enable_prefix_caching: false + mm_processor_cache_gb: 0 is_comprehension: true final_output: true final_output_type: text diff --git a/tests/e2e/stage_configs/rocm/qwen3_omni_ci.yaml b/tests/e2e/stage_configs/rocm/qwen3_omni_ci.yaml index 477e6e59f..23e75aae7 100644 --- a/tests/e2e/stage_configs/rocm/qwen3_omni_ci.yaml +++ b/tests/e2e/stage_configs/rocm/qwen3_omni_ci.yaml @@ -20,6 +20,7 @@ stage_args: engine_output_type: latent # Output hidden states for talker distributed_executor_backend: "mp" enable_prefix_caching: false + mm_processor_cache_gb: 0 hf_config_name: thinker_config tensor_parallel_size: 1 load_format: dummy