It rents GPUs from vast.ai, so make sure you have the vast.ai command line tool installed.
Example:
python3 benchmark.py --model xxx.pth --verbose -n 1 --branch daquexian/test --log-dir log
Please refer to ChatRWKV
backend implementation in benchmark.py. Typically, a benchmark script (like benchmark_chatrwkv.py) is also needed.