Update README.md

README.md

@@ -1,12 +1,54 @@
----
-title: Phi2 Assistant Demo
-emoji: 🐠
-colorFrom: indigo
-colorTo: red
-sdk: gradio
-sdk_version: 5.20.1
-app_file: app.py
-pinned: false
----
-
-Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
+# Phi-2 Fine-tuned Assistant Demo
+
+This Space demonstrates a fine-tuned version of the Microsoft Phi-2 model, trained on the OpenAssistant dataset using QLoRA (Quantized Low-Rank Adaptation). The model is designed to provide helpful and informative responses to various types of queries and instructions.
+
+## Model Information
+
+- **Base Model:** Microsoft Phi-2
+- **Fine-tuning Dataset:** OpenAssistant
+- **Training Method:** QLoRA with 8-bit quantization
+- **Model Card:** [jatingocodeo/phi2-finetuned-openassistant](https://huggingface.co/jatingocodeo/phi2-finetuned-openassistant)
+
+## How to Use
+
+1. **Enter your query** in the text box
+2. Adjust generation parameters (optional):
+   - **Maximum Length** (50-500): Controls response length
+   - **Temperature** (0.1-1.0): Controls randomness
+   - **Top P** (0.1-1.0): Controls token sampling
+
+## Example Prompts
+
+Try these examples to see what the model can do:
+- "What is machine learning?"
+- "Write a short poem about artificial intelligence"
+- "Explain quantum computing to a 10-year-old"
+- "What are the best practices for writing clean code?"
+
+## Model Capabilities
+
+The model is trained to:
+- Provide informative explanations
+- Answer questions clearly and concisely
+- Generate creative content
+- Give technical explanations
+- Follow instructions and complete tasks
+
+## Limitations
+
+- The model may occasionally generate incorrect information
+- Responses are limited by the training data
+- The model should not be used for critical applications without human oversight
+- Complex or ambiguous queries might receive simplified responses
+
+## Technical Details
+
+The model uses:
+- 8-bit quantization for efficient inference
+- Gradient checkpointing
+- Mixed precision training
+- LoRA fine-tuning techniques
+
+## License
+
+This demo uses a model that inherits the license of the base Phi-2 model and the OpenAssistant dataset.