diff --git a/benchmarks/single_node/agentic/minimaxm3_fp4_b200_trt_mtp.sh b/benchmarks/single_node/agentic/minimaxm3_fp4_b200_trt_mtp.sh index e93fd35fe2..686145460b 100755 --- a/benchmarks/single_node/agentic/minimaxm3_fp4_b200_trt_mtp.sh +++ b/benchmarks/single_node/agentic/minimaxm3_fp4_b200_trt_mtp.sh @@ -134,7 +134,7 @@ enable_chunked_prefill: true enable_autotuner: true trust_remote_code: true reasoning_parser: minimax_m3 -stream_interval: 100 +stream_interval: 20 print_iter_log: true num_postprocess_workers: 8 enable_attention_dp: false diff --git a/perf-changelog.yaml b/perf-changelog.yaml index 806bb4bd15..93b6a7c1b3 100644 --- a/perf-changelog.yaml +++ b/perf-changelog.yaml @@ -6243,3 +6243,11 @@ - "Bump image from lmsysorg/sglang-rocm:v0.5.17-rocm720-mi35x-20260811 to lmsysorg/sglang-rocm:v0.5.17-rocm720-mi35x-20260818" - "Serve the AgentX arm with an fp8_e4m3 KV cache." pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2677 + +- config-keys: + - minimaxm3-fp4-b200-trtllm-agentic-mtp + scenario-type: + - agentic-coding + description: + - "Refresh with lower stream interval to collect correct client metrics" + pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/2686