main
yaml 6 lines 267 Bytes
Raw
1 # SPDX-License-Identifier: Apache-2.0
2 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
3 model_name: openai/gpt-oss-20b
4 metric_threshold: 0.568
5 reasoning_effort: low
6 server_args: "--attention-backend ROCM_AITER_UNIFIED_ATTN --tensor-parallel-size 2"