Showing cs.CLShow all
2 papers · 1 filter
cs.CL2026
Three-Phase Transformer
Mohammad R. Abu Ayyash
We present Three-Phase Transformer (3PT), a residual-stream structural prior for decoder-only Transformers on a standard SwiGLU + RMSNorm + RoPE + GQA backbone. The hidden vector i…
cs.CL2026
Brainstacks: Cross-Domain Cognitive Capabilities via Frozen MoE-LoRA Stacks for Continual LLM Learning
Mohammad R. Abu Ayyash
We present Brainstacks, a modular architecture for continual multi-domain fine-tuning of large language models that packages domain expertise as frozen adapter stacks composing add…