view article Article Emergent Semantics Beyond Token Embeddings: A GPT-like Transformer Learns with Frozen 16‑D Binary Token-ID Embeddings (n_embed=16) 2 days ago
Growing Transformers:Layer-wise Expansion Comparative Study Collection Paper: 2507.07129 'Growing Transformers: Modular Composition and Layer-wise Expansion on a Frozen Substrate' (4.2.2, 5.2. Results) • 8 items • Updated 4 days ago • 1