Instructions to use microsoft/kosmos-2-patch14-224 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use microsoft/kosmos-2-patch14-224 with Transformers:
# Use a pipeline as a high-level helper # Warning: Pipeline type "image-to-text" is no longer supported in transformers v5. # You must load the model directly (see below) or downgrade to v4.x with: # 'pip install "transformers<5.0.0' from transformers import pipeline pipe = pipeline("image-to-text", model="microsoft/kosmos-2-patch14-224")# Load model directly from transformers import AutoProcessor, AutoModelForImageTextToText processor = AutoProcessor.from_pretrained("microsoft/kosmos-2-patch14-224") model = AutoModelForImageTextToText.from_pretrained("microsoft/kosmos-2-patch14-224") - Notebooks
- Google Colab
- Kaggle
The size of Kosmos-2.pt
#2
by sanshi2023 - opened
Thank you for the great work!
What is the difference between the model and the checkpoint ? The size of Kosmos-2.pt is close to 19.1G, while the size of the Microsoft/Kosmos-2-patch14-224 model is close to 9.5G.
Hi @sanshi2023
I also observed this during the conversion. I forgot about the details now, but I believe it is because Kosmos-2.ptsaved two copies of the weights (or maybe the optimizer states).