Upload README.md with huggingface_hub

README.md

@@ -1,12 +1,16 @@
 # 🧠 Myanmar LLM Training
 
-Fine-tune **Llama-3.1-8B-Instruct** with Myanmar language dataset.
+Fine-tune **Qwen2.5-0.5B-Instruct** with Myanmar language dataset.
+
+## ⚡ No License Required!
+
+This model is fully open. No Llama license needed!
 
 ## 📋 Requirements
 
 - Python 3.8+
-- GPU with 16GB+ VRAM (recommended)
-- HuggingFace Account with Llama access
+- GPU with 6GB+ VRAM
+- HuggingFace Account
 
 ## 🚀 Quick Start
 
@@ -18,11 +22,8 @@ pip install -r requirements.txt
 ### 2. Login to HuggingFace
 ```bash
 huggingface-cli login
-# Enter your token
 ```
 
-**Note:** Llama requires accepting the license at https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct
-
 ### 3. Run training
 ```bash
 python train.py
@@ -32,45 +33,47 @@ python train.py
 
 | Parameter | Default | Description |
 |-----------|---------|-------------|
-| MODEL_NAME | meta-llama/Llama-3.1-8B-Instruct | Base model |
+| MODEL_NAME | Qwen/Qwen2.5-0.5B-Instruct | Base model (fully open!) |
 | num_train_epochs | 3 | Training iterations |
-| per_device_train_batch_size | 2 | Batch size (4-bit) |
-| gradient_accumulation_steps | 8 | Effective batch |
-| learning_rate | 1e-5 | Learning rate |
+| per_device_train_batch_size | 4 | Batch size |
+| gradient_accumulation_steps | 4 | Effective batch = 16 |
+| learning_rate | 2e-5 | Learning rate |
 
 ## 📊 Features
 
-- ✅ 4-bit quantization (NF4) - အနည်းဆုံး VRAM နဲ့ run လုပ်နိုင်ပါသည်။
+- ✅ Fully open model - လိုင်စင်မလိုပါသည်။
+- ✅ FP16 precision - ပိုမိုမြန်ပါသည်။
 - ✅ Gradient checkpointing - Memory ချွေတာပါသည်။
 - ✅ Test/Validation evaluation - နှစ်ခုလုံးအတွက် စမ်းသပ်ပါသည်။
-- ✅ BF16 mixed precision - ပိုမိုတိကျတဲ့ training။
 
 ## 📊 Training Data
 
-Dataset: [amkyawdev/myanmar-llm-data](https://huggingface.co/datasets/amkyawdev/myanmar-llm-data)
+Dataset: [amkyawdev/AmkyawDev-Dataset](https://huggingface.co/datasets/amkyawdev/AmkyawDev-Dataset)
 
 | Split | Samples |
 |-------|---------|
-| Train | 1000 |
-| Validation | 1000 |
-| Test | 1000 |
+| Train | ~29,100 |
+| Validation | ~29,100 |
+| Test | ~29,100 |
+
+> **Note:** Each file (train.jsonl, test.jsonl, validation.jsonl) has ~29,100 conversations!
 
 ## 💾 Output
 
-Trained model saved to `./myanmar-llama-output/`
+Trained model saved to `./myanmar-qwen-output/`
 
 ## 📤 Upload to HuggingFace
 
 ```bash
-cd myanmar-llama-output
-huggingface-cli upload amkyawdev/my-myanmar-llama . --repo-type model
+cd myanmar-qwen-output
+huggingface-cli upload amkyawdev/my-myanmar-qwen . --repo-type model
 ```
 
 ## 🖥️ Google Colab
 
 ```python
 # Install
-!pip install transformers datasets torch bitsandbytes accelerate
+!pip install transformers datasets torch accelerate
 
 # Login
 from huggingface_hub import login
@@ -80,10 +83,5 @@ login("YOUR_TOKEN")
 %run train.py
 ```
 
-## ⚠️ Important
-
-1. Llama license လိုပါသည်။ https://huggingface.co/meta-llama/Llama-3.1-8B-Instruct မှာ Accept လုပ်ပါသည်။
-2. Token မှာLlama access ရှိရပါသည်။
-
 ---
 Built by amkyawdev