sagar1697/sagar-bitnet-cpu-brain

Sagar CPU Brain (Qwen2.5-3B)

A high-performance CPU inference Space hosting the state-of-the-art Qwen2.5-3B-Instruct model (quantized to Q4_K_M GGUF) using the optimized llama.cpp runtime.

This space provides a highly capable conversational assistant and reasoning engine tailored for support chat dev environments.

Features

• —Qwen2.5-3B-Instruct: Exceptional reasoning, coding, and chat capabilities, outperforming many larger 7B models.
• —Web Chat UI: An elegant, interactive Gradio chat interface accessible directly via the browser at the Space root.
• —OpenAI API Compatible: Exposes a standard /v1/chat/completions API endpoint, allowing drop-in integration with existing OpenAI SDKs.
• —Authenticated Access: Secured via bearer token matching your Hugging Face Space secrets.

API Usage

Endpoint

POST /v1/chat/completions

Headers

Authorization: Bearer <HUGGINGFACE_TOKEN>
Content-Type: application/json

Request Payload

{
  "model": "sagar-cpu-brain-v1",
  "messages": [
    {"role": "system", "content": "You are a helpful support agent."},
    {"role": "user", "content": "How do I reset my password?"}
  ],
  "stream": true,
  "temperature": 0.7
}