Image-to-Text
Transformers
Safetensors
falcon_ocr
text-generation
falcon
ocr
vision-language
document-understanding
v1.5
custom_code
Eval Results
Instructions to use tiiuae/Falcon-OCR with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use tiiuae/Falcon-OCR 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="tiiuae/Falcon-OCR", trust_remote_code=True)# pip install -U transformers accelerate # Load model directly from transformers import AutoModelForCausalLM model = AutoModelForCausalLM.from_pretrained("tiiuae/Falcon-OCR", trust_remote_code=True, device_map="auto") - Notebooks
- Google Colab
- Kaggle
Falcon-OCR-Arabic: weights and licence for on-premises use?
#17
by vladimir1213 - opened
Hi Falcon team,
Thank you for Falcon OCR and the Falcon-OCR-Arabic blog post.
We work with Arabic financial statements from public company filings. We tried
the Falcon-OCR-Arabic playground on one balance sheet page, and all 50 numbers
matched our answer key. We would like to run a larger test, but our deployment
has to run entirely on-premises, so the playground can't be part of it.
I could only find the base weights at tiiuae/Falcon-OCR, so I have a few questions:
- Are the Falcon-OCR-Arabic weights available, or will they be? If so, when?
- What licence would they use, and would it allow commercial on-premises use?
- If there is no public release planned, can we get access under an agreement?
- If the weights won't be available, will you release the fine-tuning method,
so that we could train the base model on Arabic documents ourselves?
Thanks,
Vladimir
Q4