nova-trained-v1
Nova Core model - A post-transformer LLM without attention, tokens, or layers.
Architecture
- Field Dynamics (O(n)) instead of Attention (O(n²))
- Continuous Pulses instead of discrete tokens
- Adaptive Depth Cores instead of fixed layers
Training
Trained on 100 examples (sentiment, math, Q&A) for 10 epochs. Achieves 100% accuracy on training data using learned response associations.
Usage
Download and use with Nova Core CLI.
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support