🔥🔹 Burmese LLM 🔹🔥
Author: Ko Yin Maung
Organization: Myanmar Innovative Group (MIG)
Gradio Demo
The demo includes:
- QA Mode: Ask Buddhist knowledge questions.
- Eng ➤ MM / MM ➤ Eng: Translate between English and Burmese.
Link : Burmese LLM Demo
Sample Queries:
- QA : ဗုဒ္ဓဘာသာရဲ့ အဓိက ယုံကြည်ကိုးကွယ်ရာက ဘာတွေလဲ။
- Eng2Mm : Winners focus on winning, losers focus on winners.
- Mm2Eng : သတ္တဝါ အားလုံး ကျန်းမာကြပါစေ။
Model Summary
This is a Burmese Large Language Model (LLM) fine-tuned for Myanmar ⇄ English translation and Buddhist Question-Answering (QA).
It is designed to:
- Answer basic Buddhist knowledge questions (RAG-based with vector database).
- Translate English ⇄ Myanmar fluently.
- Generate natural Burmese text with context-awareness.
Intended Uses & Limitations
✅ Intended Uses
- Educational purpose (Buddhist knowledge Q&A).
- Myanmar ↔ English translation tasks.
- Research and experimentation with Burmese LLMs.
⚠️ Limitations
- Not guaranteed to be 100% accurate for complex Buddhist scriptures.
- May produce hallucinations if context is missing.
- Translation quality may vary for long or domain-specific texts.
- This model is not intended for production use without human validation.
Example Usage
from transformers import AutoTokenizer, AutoModelForCausalLM
import torch
model_path = "Ko-Yin-Maung/mig-burmese-llm"
tokenizer = AutoTokenizer.from_pretrained(model_path)
model = AutoModelForCausalLM.from_pretrained(
model_path,
device_map="auto",
torch_dtype="auto",
)
query = "Winners focus on winning, losers focus on winners."
prompt = f"<start_of_turn>user\nTranslate to Myanmar: {query}\n<end_of_turn><start_of_turn>model\n"
inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=100)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))
Result
user
Translate to Myanmar: Winners focus on winning, losers focus on winners.
model
အောင်မြင်သူတွေက အောင်မြင်ဖို့ပဲ အာရုံစိုက်တယ်၊ ရှုံးနိမ့်သူတွေကတော့ အောင်မြင်သူတွေကို အာရုံစိုက်ကြတယ် ။
Citation
If you use this model, please cite:
@misc{mig-burmese-llm,
title={MIG Burmese LLM: Translation + Buddhist QA},
author={Ko Yin Maung},
year={2025},
howpublished={Hugging Face Hub},
}
Contact
✉️ Email: koyinmaung007@gmail.com
© Myanmar Innovative Group (MIG)