Hello,

I am trying to use your magiv3 model, but I am running into errors and cannot get it to work. I would be very grateful for your help.

I try code

model = AutoModelForCausalLM.from_pretrained("ragavsachdeva/magiv3", torch_dtype=torch.float16, trust_remote_code=True).cuda().eval()
processor = AutoProcessor.from_pretrained("ragavsachdeva/magiv3", trust_remote_code=True)

model.predict_detections_and_associations(images, processor)
model.predict_ocr(images, processor)
model.predict_character_grounding(images, captions, processor)

but i dont understund parametars, dependecies, requiments.tct, nothing :(

When I try to use the model with the standard Florence-2 API, I get a TypeError on the processor. It seems to not accept text, task_prompt, or images as keyword arguments.

Here is the code I am using:

import torch
from PIL import Image
from transformers import AutoModelForCausalLM, AutoProcessor

Load Model and Processor

model = AutoModelForCausalLM.from_pretrained(
"ragavsachdeva/magiv3",
torch_dtype=torch.float16,
trust_remote_code=True
).cuda().eval()

processor = AutoProcessor.from_pretrained(
"ragavsachdeva/magiv3",
trust_remote_code=True
)

Prepare Data

image_paths = ["path/to/page1.png"] # I use my real paths here
images = [Image.open(path).convert("RGB") for path in image_paths]
ocr_prompt = ""

Attempt to run OCR

try:
# This line causes a TypeError
inputs = processor(images=images, text=ocr_prompt, return_tensors="pt").to("cuda", torch.float16)

generated_ids = model.generate(
    input_ids=inputs["input_ids"],
    pixel_values=inputs["pixel_values"],
    max_new_tokens=1024
)
generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
print(generated_text)

except Exception as e:
print(f"An error occurred: {e}")

The error I get is: TypeError: Florence2Processor.__call__() got an unexpected keyword argument 'text' (or 'images', or 'task_prompt').

Could you please provide a simple, working code snippet that shows the correct way to run OCR with magiv3?

Thank you.