sglang/benchmark/json_jump_forward

Run benchmark

Dependencies

llama_cpp_python          0.2.38
guidance                  0.1.10
vllm                      0.2.7
outlines                  0.0.25

Build dataset

When benchmarking long document information retrieval, run the following command to build the dataset:

pip install wikipedia
python3 build_dataset.py

Benchmark sglang

Run Llama-7B

python3 -m sglang.launch_server --model-path meta-llama/Llama-2-7b-chat-hf --port 30000 

Benchmark Character Generation

python3 bench_sglang.py --mode character

Benchmark City Information Retrieval

python3 bench_sglang.py --mode city

Benchmark vllm

Run Llama-7B

python3 -m outlines.serve.serve --tokenizer-mode auto --model meta-llama/Llama-2-7b-chat-hf  --disable-log-requests --port 21000

Benchmark Character Generation

python3 bench_other.py --mode character --backend vllm

Benchmark City Information Retrieval

python3 bench_other.py --mode city --backend vllm

Benchmark guidance

Run Llama-7B and benchmark character generation

python3 bench_other.py --mode character --backend guidance --parallel 1

Run Llama-7B and benchmark city information retrieval

python3 bench_other.py --mode city --backend guidance --parallel 1