From 7a687e1575e67b6ee8aa1e288e7e3ae8b8c9fd25 Mon Sep 17 00:00:00 2001 From: tgerdes Date: Fri, 22 Mar 2024 09:27:23 -0500 Subject: [PATCH] Fix issue where input data wasn't provided --- .../genai-perf/genai_perf/test_end_to_end.py | 8 ++++---- src/c++/perf_analyzer/genai-perf/genai_perf/wrapper.py | 6 +++--- 2 files changed, 7 insertions(+), 7 deletions(-) diff --git a/src/c++/perf_analyzer/genai-perf/genai_perf/test_end_to_end.py b/src/c++/perf_analyzer/genai-perf/genai_perf/test_end_to_end.py index 76cda3edc..9fec7e1fc 100644 --- a/src/c++/perf_analyzer/genai-perf/genai_perf/test_end_to_end.py +++ b/src/c++/perf_analyzer/genai-perf/genai_perf/test_end_to_end.py @@ -20,10 +20,10 @@ ] base_commands = { - "nim_chat": "genai-pa -s 999 -p 20000 -m llama-2-7b-chat -u http://localhost:9999--output-format openai_chat_completions --service-kind openai --endpoint v1/chat/completions", - "nim_completions": "genai-pa -s 999 -p 20000 -m llama-2-7b -u http://localhost:9999 --output-format openai_completions --service-kind openai --endpoint v1/completions", - "vllm_openai": "genai-pa -s 999 -p 20000 -m mistralai/Mistral-7B-v0.1 --output-format openai_chat_completions --service-kind openai --endpoint v1/chat/completions", - "triton_trtllm": "genai-pa -s 999 -p 20000 -m llama-2-7b -u 0.0.0.0:9999 --service-kind triton --output-format trtllm", + "nim_chat": "genai-perf -s 999 -p 20000 -m llama-2-7b-chat -u http://localhost:9999 --output-format openai_chat_completions --service-kind openai --endpoint v1/chat/completions", + "nim_completions": "genai-perf -s 999 -p 20000 -m llama-2-7b -u http://localhost:9999 --output-format openai_completions --service-kind openai --endpoint v1/completions", + "vllm_openai": "genai-perf -s 999 -p 20000 -m mistralai/Mistral-7B-v0.1 --output-format openai_chat_completions --service-kind openai --endpoint v1/chat/completions", + "triton_trtllm": "genai-perf -s 999 -p 20000 -m llama-2-7b -u 0.0.0.0:9999 --service-kind triton --output-format trtllm", } testname = "" diff --git a/src/c++/perf_analyzer/genai-perf/genai_perf/wrapper.py b/src/c++/perf_analyzer/genai-perf/genai_perf/wrapper.py index 352b64476..27709147a 100644 --- a/src/c++/perf_analyzer/genai-perf/genai_perf/wrapper.py +++ b/src/c++/perf_analyzer/genai-perf/genai_perf/wrapper.py @@ -28,7 +28,7 @@ import subprocess import genai_perf.utils as utils -from genai_perf.constants import DEFAULT_GRPC_URL, LOGGER_NAME +from genai_perf.constants import DEFAULT_GRPC_URL, DEFAULT_INPUT_DATA_JSON, LOGGER_NAME from genai_perf.llm_inputs.llm_inputs import OutputFormat logger = logging.getLogger(LOGGER_NAME) @@ -52,7 +52,7 @@ def add_protocol_args(args): def build_cmd(args, extra_args): skip_args = [ "func", - "dataset", + "input_dataset", "input_type", "input_format", "model", @@ -72,7 +72,7 @@ def build_cmd(args, extra_args): utils.remove_file(args.profile_export_file) - cmd = f"perf_analyzer -m {args.model} --async " + cmd = f"perf_analyzer -m {args.model} --async --input-data {DEFAULT_INPUT_DATA_JSON} " for arg, value in vars(args).items(): if arg in skip_args: pass