Spaces:
Running
Running
| title: ASI V2.5 ZeroGPU Demo | |
| emoji: π | |
| colorFrom: blue | |
| colorTo: red | |
| sdk: gradio | |
| sdk_version: "4.44.0" | |
| app_file: app.py | |
| pinned: false | |
| license: mit | |
| # π ASI V2.5: ZeroGPU H200 Performance Demo | |
| **REAL GPU ASI V2.5 Testing on NVIDIA H200!** | |
| This Space demonstrates the **Adaptive Structured Intelligence V2.5** attention mechanism running on **ZeroGPU H200** with **70GB VRAM** - finally showing the real speedups ASI was designed for! | |
| ## β¨ Features | |
| - π **ZeroGPU H200**: NVIDIA H200 with 70GB VRAM (FREE with Pro!) | |
| - π₯ **Real GPU Performance**: Actual CUDA optimizations and tensor cores | |
| - π **Long Sequences**: Test up to 8192 tokens (impossible on CPU) | |
| - β‘ **Mixed Precision**: FP16 optimization for H200 | |
| - π― **Adaptive Attention**: Linear O(L) for long sequences, Exact O(LΒ²) for short | |
| ## π― Performance Expectations | |
| | Sequence Length | Expected Speedup | Attention Type | | |
| |-----------------|------------------|----------------| | |
| | 1024 tokens | 1.2-1.5x | Linear | | |
| | 2048 tokens | 1.5-2.0x | Linear | | |
| | 4096 tokens | 2.0-2.5x | Linear | | |
| | 8192 tokens | 2.5x+ | Linear | | |
| ## π§ How to Use | |
| 1. Configure ASI parameters (threshold, feature_dim, etc.) | |
| 2. Set sequence lengths to test (try long ones!) | |
| 3. Click "Run ZeroGPU ASI V2.5 Test" | |
| 4. See REAL GPU speedups! | |
| ## π‘ Why ZeroGPU? | |
| - β **FREE** with HuggingFace Pro ($9/mois) | |
| - β **70GB VRAM** - handles very long sequences | |
| - β **NVIDIA H200** - latest architecture | |
| - β **8x quota** with Pro priority | |
| ## π Technical Details | |
| - **ASI V2.5**: Adaptive attention switching | |
| - **CUDA Optimized**: Tensor cores + mixed precision | |
| - **Memory Efficient**: 70GB VRAM utilization | |
| - **Production Ready**: Real deployment example | |
| ## π Links | |
| - π€ **Model**: [khopilot/asi-v25-longformer-core](https://huggingface.co/khopilot/asi-v25-longformer-core) | |
| - π **GitHub**: [khopilot/asi-v25-longformer-core](https://github.com/khopilot/asi-v25-longformer-core) | |
| - π¦ **PyPI**: [asi-v25-longformer](https://pypi.org/project/asi-v25-longformer/) | |
| --- | |
| **Finally - ASI V2.5 running on real GPU with the speedups it was designed for!** π | |