iPwnds/qwen2.5-7b-instruct-chat

Qwen2.5-7B-Instruct Chat

A minimal Gradio chat demo for Qwen/Qwen2.5-7B-Instruct, served through Hugging Face's Inference Providers rather than loaded locally — so it runs comfortably on the free CPU Space tier.

How it works

app.py builds a gr.ChatInterface and streams responses from huggingface_hub.InferenceClient.chat_completion. Adjustable controls: system prompt, max new tokens, temperature, and top-p.

Configuration

This Space needs an HF_TOKEN secret (Settings → Variables and secrets) with at least Inference permission, so it can call the model on your behalf.