python3 convert_checkpoint.py --model_dir /TinyLlama-1.1B-Chat-v0.4 --output_dir llama-2-7b-ckpt
trtllm-build --checkpoint_dir llama-2-7b-ckpt
--gemm_plugin float16
--output_dir ./llama-2-7b-engine
--context_fmha disable
python3 ../run.py --engine_dir /TensorRT-LLM/examples/llama/llama-2-7b-engine --max_output_len 100 --tokenizer_dir TinyLlama/TinyLlama-1.1B-Chat-v0.4 --input_text "How do I count to nine in French?"