Skip to content

LiquidAI LFM3-VL video understanding through the OpenAI-compatible HTTP API

LiquidAI LFM3-VL video understanding through the OpenAI-compatible HTTP API.

Start the server: mistralrs serve -p 1234 -m LiquidAI/LFM3-VL-2B –isq q4k

Then run python examples/server/lfm3_vl_video.py clip.mp4. The video path must be readable by the server on the same filesystem. Non-GIF video requires FFmpeg on the server PATH; GIFs are decoded natively.

"""
LiquidAI LFM3-VL video understanding through the OpenAI-compatible HTTP API.
Start the server:
mistralrs serve -p 1234 -m LiquidAI/LFM3-VL-2B --isq q4k
Then run `python examples/server/lfm3_vl_video.py clip.mp4`.
The video path must be readable by the server on the same filesystem.
Non-GIF video requires FFmpeg on the server PATH; GIFs are decoded natively.
"""
import argparse
from pathlib import Path
from openai import OpenAI
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument("video", type=Path, help="Path to a video readable by the server")
args = parser.parse_args()
client = OpenAI(api_key="foobar", base_url="http://localhost:1234/v1/")
completion = client.chat.completions.create(
model="default",
messages=[
{
"role": "user",
"content": [
{
"type": "video_url",
"video_url": {"url": args.video.resolve().as_uri()},
},
{
"type": "text",
"text": "Describe what happens in this video, in order.",
},
],
}
],
max_tokens=256,
)
print(completion.choices[0].message.content)

Source: examples/server/lfm3_vl_video.py