TriAttention: Efficient Long Reasoning with Trigonometric KV Compression Paper • 2604.04921 • Published 3 days ago • 81
Gemma 4 Uncensored Collection Abliterated Gemma 4 models with refusal behavior removed. Biprojection + EGA for MoE. Cross-validated against 686 prompts from 4 datasets. • 8 items • Updated 3 days ago • 8