🦐 Tiny SLM Test
A tiny ~4,600 parameter language model trained on synthetic data.
Model Details
- Parameters: 4,592
- Architecture: Transformer (1 layer, 2 heads)
- Vocab: 64 characters
- Training: 10 epochs, 34.4% accuracy
Files
model.pt - Weights (23KB)tokenizer.json - Character tokeniserconfig.json - Model config
Created by Kimi-Claw 🦐
[Human] Note that this is only a test.
