🇲🇲 Myanmar AI V3

Myanmar AI V3 သည် Qwen2.5-Coder-1.5B base model ပေါ်တွင် LoRA adapter ဖြင့် fine-tune လုပ်ထားသော မြန်မာဘာသာစကားအတွက် Language Model ဖြစ်ပါသည်။

📋 Model Details

Field	Details
Model Name	amkyawdev/myanmar-ai-v3
Model Type	LoRA Adapter (Fine-tuned)
Base Model	Qwen/Qwen2.5-Coder-1.5B-Instruct
Language(s)	မြန်မာ (Burmese), English
License	MIT License
Developer	amkyawdev

🔧 Technical Specifications

Specification	Value
Architecture	Transformer-based Decoder (Causal LM)
Base Model Params	1.5 Billion
LoRA Rank (r)	16
LoRA Alpha	32
LoRA Dropout	0.05
Target Modules	q_proj, k_proj, v_proj, o_proj, gate_proj, up_proj, down_proj
Task Type	Causal Language Modeling

🚀 Quick Start

python

from transformers import AutoModelForCausalLM, AutoTokenizer
from peft import PeftModel

MODEL_NAME = "amkyawdev/myanmar-ai-v3"
BASE_MODEL = "Qwen/Qwen2.5-Coder-1.5B-Instruct"

# Load tokenizer
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)

# Load base model
base_model = AutoModelForCausalLM.from_pretrained(
    BASE_MODEL,
    torch_dtype=torch.float16,
    device_map="auto"
)

# Load LoRA adapter
model = PeftModel.from_pretrained(base_model, MODEL_NAME)

# Generate
text = "မြန်မာနိုင်ငံအကြောင်း ပြောပါ"
inputs = tokenizer(text, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=100)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

📊 Training Data

Dataset: amkyawdev/myanmar-ai-training
Data Format: SFT (Supervised Fine-Tuning) with system/user/assistant messages

⚠️ Limitations

မြန်မာဘာသာစကားဖြင့် conversation ပြုလုပ်နိုင်ပါသည်
English prompts များကိုလည်း နားလည်နိုင်ပါသည်
Programming/code generation tasks များအတွက် အထူးသင့်တော်ပါသည် (Base model: Qwen2.5-Coder)

📞 Contact

Developer: amkyawdev
Demo: Live Demo

Made with ❤️ for the Myanmar AI Community