Files
sglang/examples/quick_start/srt_example_yi_vl.py
2024-07-17 11:55:39 -07:00

71 lines
1.5 KiB
Python

"""
Usage: python3 srt_example_yi_vl.py
Requirements: transformers==4.38
"""
import sglang as sgl
@sgl.function
def image_qa(s, image_path, question):
s += sgl.user(sgl.image(image_path) + question)
s += sgl.assistant(sgl.gen("answer"))
def single():
state = image_qa.run(
image_path="images/cat.jpeg",
question="What is this?",
max_new_tokens=64,
stop="###",
)
print(state["answer"], "\n")
def stream():
state = image_qa.run(
image_path="images/cat.jpeg",
question="What is this?",
max_new_tokens=64,
stream=True,
stop="###",
)
for out in state.text_iter("answer"):
print(out, end="", flush=True)
print()
def batch():
states = image_qa.run_batch(
[
{"image_path": "images/cat.jpeg", "question": "What is this?"},
{"image_path": "images/dog.jpeg", "question": "What is this?"},
],
max_new_tokens=64,
stop="###",
)
for s in states:
print(s["answer"], "\n")
if __name__ == "__main__":
runtime = sgl.Runtime(model_path="BabyChou/Yi-VL-6B")
# runtime = sgl.Runtime(model_path="BabyChou/Yi-VL-34B")
sgl.set_default_backend(runtime)
# Run a single request
print("\n========== single ==========\n")
single()
# Stream output
print("\n========== stream ==========\n")
stream()
# Run a batch of requests
print("\n========== batch ==========\n")
batch()
runtime.shutdown()