Spaces:
Running on Zero
Running on Zero
| title: Context Window Extender - Chat Mode | |
| emoji: 🧠 | |
| colorFrom: purple | |
| colorTo: indigo | |
| sdk: gradio | |
| sdk_version: 5.0.0 | |
| app_file: app.py | |
| pinned: false | |
| # Context Window Extender - Chat Mode | |
| Load any causal language model from Hugging Face Hub and extend its context window dynamically. | |
| ## Features | |
| - **Recent Models**: Default model is `Qwen/Qwen3-30B-A3B-Thinking-2507` (256K context, extendable to 1M) | |
| - **Conversational UI**: Chat-style interface instead of form-based | |
| - **Dynamic Context Multiplier**: Expand context by 2x, 5x, 10x, 20x, 50x, or 100x | |
| - **Streaming Responses**: Real-time streaming of model outputs | |
| - **RoPE Extension**: Support for linear, dynamic, and YaRN RoPE scaling | |
| ## Available Models | |
| - Qwen/Qwen3-30B-A3B-Thinking-2507 (default) | |
| - Qwen/Qwen2.5-1.5B-Instruct | |
| - Qwen/Qwen2.5-3B-Instruct | |
| - microsoft/phi-4-mini-instruct | |
| - deepseek-ai/DeepSeek-R1-Distill-Qwen-7B | |