PyPI page
Home page
Author:
Yohei Nakajima
License:
Apache-2.0
Summary:
Ask a frozen open vision-language model typed questions about images (yes/no, pick-one, rating) and read probabilities from one forward pass; nothing is generated or trained.
Latest version:
0.3.1
Required dependencies:
accelerate
|
datasets
|
flask
|
litellm
|
matplotlib
|
numpy
|
pillow
|
pydantic
|
python-dotenv
|
pyyaml
|
scikit-learn
|
scipy
|
torch
|
torchvision
|
transformers
Optional dependencies:
num2words
Downloads last day:
0
Downloads last week:
16
Downloads last month:
305
7-day smoothing