Filiberto 124M OCR is a small specialized model for OCR correction of Spanish Golden Age Dramas OCR, based on the Filiberto 124M Spanish Golden Age Drama foundation model.
Filiberto 124M OCR is only 124 million parameters. It can run easily on CPU or provide correction at scale on GPUs (>10k tokens/seconds).
Training
The pre-trained included a collection of individual verses and their correction taken from the TEXORO corpus, via a collaboration with ETSO, totalling ~5 million tokens.
Pre-training ran on 5 epochs with levanter (500 steps total, each processing 1024 sequences of 512 tokens) on a TPUv4-32 for 15 minutes.
Tokenization is currently done with the GPT-2 tokenizer.
Example of OCR correction
Filiberto 124M OCR has been pre-trained on an instruction dataset with a hard-coded structure: ### Text ### for OCRized text submissiong and ### Correction ### for the generated correction.
Filiberto 124M OCR can be imported like any GPT-2 like model:
python
1import torch
2from transformers import GPT2LMHeadModel, GPT2Tokenizer
34# Load pre-trained model and tokenizer5model_name ="bertin-project/filiberto-124M-ocr"6model = GPT2LMHeadModel.from_pretrained(model_name)7tokenizer = GPT2Tokenizer.from_pretrained(model_name)89# Set the device to GPU if available, otherwise use CPU10device = torch.device("cuda"if torch.cuda.is_available()else"cpu")11model.to(device)
And afterwards inference can be run like this:
python
1# Function to generate text2defocr_correction(prompt, max_new_tokens=600):34 prompt =f"""### Text ###\n{prompt}\n\n\n### Correction ###\n"""5 input_ids = tokenizer.encode(prompt, return_tensors="pt").to(device)67# Generate text8 output = model.generate(input_ids,9 max_new_tokens=max_new_tokens,10 pad_token_id=tokenizer.eos_token_id,11 top_k=50)1213# Decode and return the generated text14return tokenizer.decode(output[0], skip_special_tokens=True).split("### Correction ###")[-1].strip()1516ocr_result = ocr_correction(prompt)17print(ocr_result)
An example of an OCRized drama:
Otra vez, Don Iuan, me dad,
y otras mil vezes los braços.
Otra, y otras mil sean lazos
de nuestra antigua amistad.
Como venis?
Yo me siento
tan alegre, tan vfano,
tan venturoso, tan vano,
que no podrà el pensamiento
encareceros jamàs
las venturas que posseo,
porque el pensamiento creo
would yield this result:
Otra vez, Don Iuan, me dad,
y otras mil vezes los braços.
Otra, y otras mil sean lazos
de nuestra antigua amistad.
Como venis?
Yo me siento
tan alegre, tan vfano,
tan venturoso, tan vano,
que no podrà el pensamiento
encareceros jamàs
las venturas que posseo,
porque el pensamiento creo
