metadata

title: Phi-2 Fine-tuned Chat Assistant
emoji: 🤖
colorFrom: blue
colorTo: purple
sdk: gradio
sdk_version: 4.44.1
app_file: app.py
pinned: false
license: mit

Phi-2 Fine-tuned Chat Assistant

This Space hosts a fine-tuned version of Microsoft's Phi-2 model using QLoRA (Quantized Low-Rank Adaptation). The model has been trained on the OpenAssistant dataset to improve its conversational abilities.

Model Details

Base Model: Microsoft Phi-2
Training Method: QLoRA (4-bit quantization)
Dataset: OpenAssistant Conversations Dataset
Fine-tuning Parameters:
- LoRA rank: 16
- LoRA alpha: 32
- Dropout: 0.1
- Target modules: q_proj, v_proj

Usage

Simply type your message in the input box and press Enter. The model will generate a response based on your input. You can also try the example prompts provided below the chat interface.

Features

Interactive chat interface
Real-time response generation
Example prompts for quick testing
Configurable generation parameters (temperature, top-p)

Limitations

The model may occasionally generate incorrect or inconsistent responses
Response generation time may vary depending on the input length and server load
The model's knowledge is limited to its training data

License

This Space uses the Microsoft Phi-2 model which is subject to its original license. The fine-tuning additions are provided under the MIT License.