coptic-megalaa-finetuned

from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

# Greekification preprocessing (required for megalaa models)
COPTIC_TO_GREEK = {
    "ⲁ": "α", "ⲃ": "β", "ⲅ": "γ", "ⲇ": "δ", "ⲉ": "ε", "ⲋ": "ϛ",
    "ⲍ": "ζ", "ⲏ": "η", "ⲑ": "θ", "ⲓ": "ι", "ⲕ": "κ", "ⲗ": "λ",
    "ⲙ": "μ", "ⲛ": "ν", "ⲝ": "ξ", "ⲟ": "ο", "ⲡ": "π", "ⲣ": "ρ",
    "ⲥ": "σ", "ⲧ": "τ", "ⲩ": "υ", "ⲫ": "φ", "ⲭ": "χ", "ⲯ": "ψ",
    "ⲱ": "ω", "ϣ": "s", "ϥ": "f", "ϧ": "k", "ϩ": "h", "ϫ": "j",
    "ϭ": "c", "ϯ": "t",
}

def greekify(coptic_text):
    """Convert Coptic Unicode to Greek transcription"""
    return "".join(COPTIC_TO_GREEK.get(c.lower(), c.lower()) for c in coptic_text)

# Load model
model_name = "Norelad/coptic-megalaa-finetuned"
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = AutoModelForSeq2SeqLM.from_pretrained(model_name)

# Translate
coptic_text = "ⲁⲛⲟⲕ ⲡⲉ ⲡⲉⲭⲣⲓⲥⲧⲟⲥ"
greekified = greekify(coptic_text)
inputs = tokenizer(greekified, return_tensors="pt", padding=True)
outputs = model.generate(**inputs, max_length=256, num_beams=5)
translation = tokenizer.decode(outputs[0], skip_special_tokens=True)

print(f"Coptic: {coptic_text}")
print(f"English: {translation}")
# Output: "I am Christ."

@misc{coptic-megalaa-finetuned,
  author = {Linden, André},
  title = {Coptic-English Translation Model (Fine-tuned)},
  year = {2025},
  publisher = {HuggingFace},
  howpublished = {\url{https://huggingface.co/Norelad/coptic-megalaa-finetuned}},
  note = {Fine-tuned from megalaa/coptic-english-translator}
}

@misc{megalaa-coptic-translator,
  author = {Megalaa},
  title = {Coptic-English Translator},
  year = {2024},
  publisher = {HuggingFace},
  howpublished = {\url{https://huggingface.co/megalaa/coptic-english-translator}}
}