diff --git a/examples/serve/genai_perf_client.sh b/examples/serve/genai_perf_client.sh index 91a6308995c..0e7a75ae6c7 100755 --- a/examples/serve/genai_perf_client.sh +++ b/examples/serve/genai_perf_client.sh @@ -1,8 +1,11 @@ #! /usr/bin/env bash +# Set to TinyLlama/TinyLlama-1.1B-Chat-v1.0 to download from Hugging Face. +# Or set to the path of local tokenizer directory. +TOKENIZER_PATH_OR_NAME="/scratch.trt_llm_data/llm-models/llama-models-v2/TinyLlama-1.1B-Chat-v1.0" genai-perf profile \ -m TinyLlama-1.1B-Chat-v1.0 \ - --tokenizer TinyLlama/TinyLlama-1.1B-Chat-v1.0 \ + --tokenizer $TOKENIZER_PATH_OR_NAME \ --endpoint-type chat \ --random-seed 123 \ --synthetic-input-tokens-mean 128 \ diff --git a/examples/serve/genai_perf_client_for_multimodal.sh b/examples/serve/genai_perf_client_for_multimodal.sh index a04f424549f..9611a01fdae 100644 --- a/examples/serve/genai_perf_client_for_multimodal.sh +++ b/examples/serve/genai_perf_client_for_multimodal.sh @@ -1,8 +1,11 @@ #! /usr/bin/env bash +# Set to Qwen/Qwen2.5-VL-3B-Instruct to download from Hugging Face. +# Or set to the path of local tokenizer directory. +TOKENIZER_PATH_OR_NAME="/scratch.trt_llm_data/llm-models/Qwen2.5-VL-3B-Instruct" genai-perf profile \ -m Qwen2.5-VL-3B-Instruct \ - --tokenizer Qwen/Qwen2.5-VL-3B-Instruct \ + --tokenizer $TOKENIZER_PATH_OR_NAME \ --endpoint-type multimodal \ --random-seed 123 \ --image-width-mean 64 \