metadata
license: mit
github: https://github.com/chongzicbo/MathImg2Latex/tree/main
import torch
from PIL import Image
from transformers import VisionEncoderDecoderModel
from transformers.models.nougat import NougatTokenizerFast
from nougat_latex.util import process_raw_latex_code
from nougat_latex import NougatLaTexProcessor
os.environ["RUN_ON_GPU_IDs"] = "1"
device = torch.device("cpu")
model_path = "chongzicbo/MathImg2Latex"
tokenizer = NougatTokenizerFast.from_pretrained(model_path)
latex_processor = NougatLaTexProcessor.from_pretrained(model_path)
model = VisionEncoderDecoderModel.from_pretrained(model_path)
model.to(device)
img_path = "/data/code/MathOCR/MathImg2Latex/examples/test_data/test86_screenshot_bigger.jpg"
image = Image.open(img_path)
if not image.mode == "RGB":
image = image.convert("RGB")
pixel_values = latex_processor(image, return_tensors="pt").pixel_values
task_prompt = tokenizer.bos_token
decoder_input_ids = tokenizer(
task_prompt, add_special_tokens=False, return_tensors="pt"
).input_ids
with torch.no_grad():
outputs = model.generate(
pixel_values.to(device),
decoder_input_ids=decoder_input_ids.to(device),
max_length=model.decoder.config.max_length,
early_stopping=True,
pad_token_id=tokenizer.pad_token_id,
eos_token_id=tokenizer.eos_token_id,
use_cache=True,
num_beams=1,
bad_words_ids=[[tokenizer.unk_token_id]],
return_dict_in_generate=True,
)
sequence = tokenizer.batch_decode(outputs.sequences)[0]
sequence = (
sequence.replace(tokenizer.eos_token, "")
.replace(tokenizer.pad_token, "")
.replace(tokenizer.bos_token, "")
)
sequence = process_raw_latex_code(sequence)
print(sequence)