diff --git a/trt-llm-engine-builder-templates/custom_download_cmd/config.yaml b/trt-llm-engine-builder-templates/custom_download_cmd/config.yaml new file mode 100644 index 000000000..2e1279f23 --- /dev/null +++ b/trt-llm-engine-builder-templates/custom_download_cmd/config.yaml @@ -0,0 +1,21 @@ +model_name: custom-download-cmd +python_version: py311 +requirements: [] +resources: + accelerator: A100 + cpu: "1" + memory: 24Gi + use_gpu: true +system_packages: [] +trt_llm: + build: + base_model: qwen + checkpoint_repository: + download_cmd: "pip install huggingface-cli && huggingface-cli download Qwen/Qwen2.5-0.5B-Instruct --local-dir /tmp/my_model_path" + repo: "/tmp/my_model_path" + source: DOWNLOAD_CMD + max_batch_size: 32 + quantization_type: fp8 + max_beam_width: 1 + max_seq_len: 4096 + num_builder_gpus: 1