Fix Qwen30 exact trace served-model routing
This commit is contained in:
@@ -108,6 +108,7 @@ fi
|
|||||||
|
|
||||||
"${VENV_ROOT}/bin/python" "${EXACT_TRACE_CLIENT}" \
|
"${VENV_ROOT}/bin/python" "${EXACT_TRACE_CLIENT}" \
|
||||||
--port "${SERVER_PORT}" --requests-file "${REQUESTS_FILE}" \
|
--port "${SERVER_PORT}" --requests-file "${REQUESTS_FILE}" \
|
||||||
|
--served-model "${SERVED_MODEL}" \
|
||||||
--output "${OUTPUT_ROOT}/results/result.json" --tpot-slo-ms 150 \
|
--output "${OUTPUT_ROOT}/results/result.json" --tpot-slo-ms 150 \
|
||||||
--timeout-seconds 1800
|
--timeout-seconds 1800
|
||||||
|
|
||||||
|
|||||||
Reference in New Issue
Block a user