Princeton’s new Recurrent Looped Transformer (RLT) lets a decoder‑only model pull 96 blocks per token with unlimited depth—thanks to a clever attention cache and sliding‑window tricks. The future of token inference just got a serious boost. #RLT #AttentionCache #SlidingWindowAttention
🔗
🔗
September 13, 2026 at 5:22 PM
Princeton’s new Recurrent Looped Transformer (RLT) lets a decoder‑only model pull 96 blocks per token with unlimited depth—thanks to a clever attention cache and sliding‑window tricks. The future of token inference just got a serious boost. #RLT #AttentionCache #SlidingWindowAttention
🔗
🔗
Imagine Transformers that remember the long haul without slowing down. Test‑Time Training adds a dual‑memory system with sliding‑window attention, keeping inference cheap while boosting context. Dive into the details! #TestTimeTraining #DualMemory #SlidingWindowAttention
🔗
🔗
January 6, 2026 at 11:42 PM
Imagine Transformers that remember the long haul without slowing down. Test‑Time Training adds a dual‑memory system with sliding‑window attention, keeping inference cheap while boosting context. Dive into the details! #TestTimeTraining #DualMemory #SlidingWindowAttention
🔗
🔗