- Remove USE_FAST_BFP16_MOE=1 which caused torch parameter assignment
crash in fresh containers for w8a8_int8 mode.
- Mount model path at its host absolute path inside the container so the
bench client tokenizer can load it regardless of server --model-path.
- Default WARMUP to 0 and NUM_PROMPTS to 10 for a fast smoke test.
- Default to a single scenario; document how to run a full grid.
- Update experiment README with PLATFORM env var and override examples.
Update BENCHMARK_WORKFLOW.md to require scripts and final outputs to
capture the hardware platform (e.g. NVIDIA H200, Kunlun XPU) and the
inference engine/backend (e.g. vllm-dspark, sglang, vllm-xpu). This
includes directory/file naming conventions, the results.json metadata
schema, and the pre-archive checklist.
