Upload model_card.yaml with huggingface_hub

model_card.yaml

@@ -0,0 +1,108 @@
+
+---
+language:
+  - my
+  - en
+tags:
+  - myanmar
+  - burmese
+  - llm
+  - code-generation
+  - fine-tuned
+  - lora
+  - phi-2
+  - custom_code
+  - transformers
+  - peft
+  - ai
+  - coding
+  - conversational-ai
+  - nlp
+license: mit
+datasets:
+  - amkyawdev/myanmar-llm-data
+base_model: microsoft/phi-2
+model-index:
+  - name: mm-llm-coder-lite-v1
+    results: []
+---
+
+# Model Card: mm-llm-coder-lite-v1
+
+## Model Details
+
+- **Model Name**: mm-llm-coder-lite-v1
+- **Base Model**: microsoft/phi-2
+- **Model Type**: Large Language Model (LLM)
+- **Fine-tuning Method**: LoRA (Low-Rank Adaptation)
+- **Language**: Burmese (Myanmar)
+- **License**: MIT
+
+## Training Details
+
+- **Training Epochs**: 3
+- **Learning Rate**: 2e-4
+- **LoRA Rank (r)**: 16
+- **LoRA Alpha**: 32
+- **LoRA Dropout**: 0.05
+- **Max Length**: 512
+- **Batch Size**: 4
+- **Gradient Accumulation**: 4
+- **Training Framework**: Hugging Face Transformers + PEFT + TRL
+
+## Dataset
+
+- **Training Data**: amkyawdev/myanmar-llm-data
+- **Train Samples**: ~20,327
+- **Test Samples**: ~17,155
+- **Validation Samples**: ~17,071
+
+### Data Distribution
+
+| Tag | Description | Percentage |
+|-----|-------------|------------|
+| coding | Programming conversations | 90% |
+| translation | English-Myanmar translation | 1% |
+| general | General knowledge Q&A | 1% |
+| greeting | Burmese greetings | 1% |
+
+## Usage
+
+```python
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+model = AutoModelForCausalLM.from_pretrained("amkyawdev/mm-llm-coder-lite-v1")
+tokenizer = AutoTokenizer.from_pretrained("amkyawdev/mm-llm-coder-lite-v1")
+```
+
+## Prompt Format
+
+```
+System: <system_prompt>
+
+User: <user_message>
+
+Assistant: <assistant_response><eos>
+```
+
+## Limitations
+
+- Model is specialized for Myanmar language and code generation
+- Performance may vary for other languages
+- Fine-tuned on limited dataset size
+
+## Ethical Considerations
+
+- Model trained on publicly available Myanmar language data
+- No personal identifiable information included
+- Intended for educational and research purposes
+
+## Acknowledgments
+
+- Microsoft for phi-2 base model
+- Hugging Face for Transformers and PEFT
+- Myanmar NLP community
+
+---
+
+*This model card was auto-generated for mm-llm-coder-lite-v1*