Custom-Models
Collection
This is a collection of custom transformer‑based models, currently untrained but still powerful for research purposes. • 3 items • Updated • 1
Terminator-X
Note: This model not trained.
This is a custom model made by Parveshiiii.
It is a highly advanced implementation of MHC (Manifold Hyper Connections) and DeepSeek’s MLA (Multi-Head-latent-Attention).
- Tokenizer used: Qwen3
- Downloads last month
- 1,057