Accelerating the search for antimicrobial peptides with Google Antigravity
deepmind.google·2h ago
AI Engineering
Hugging Face Blog·9h ago·5 min read
Apertus 1.5 Release Notes
TL;DR
Apertus 1.5 enhances the 8B and 70B models with 4 trillion and 2 trillion additional tokens, respectively, enabling native multimodal understanding and improved reasoning capabilities. The update also increases the context window to 262,144 tokens, significantly enhancing instruction adherence.
✦ Why It Matters
Explore the new multimodal capabilities to enhance your applications with richer data inputs today.
Key Takeaways
- Utilize Apertus 1.5 for advanced multimodal applications.
- Leverage the expanded context window for complex tasks.
- Integrate improved instruction-following in your projects.
- Access open weights and training data for custom model development.
How It Works
Apertus 1.5 employs an early-fusion approach for multimodal training, where images and audio are tokenized and processed alongside text tokens. This allows the model to understand and generate responses based on a combination of text, images, and audio, while maintaining the architecture's efficiency.