Nemotron 3: How a Mamba-Transformer Hybrid Runs 1M Tokens
Nemotron 3 swaps most attention layers for Mamba-2 and MoE to reach 1M-token context at 3.3x throughput. Here is how the hybrid architecture actually works.
Long-Context
Sparse Attention Explained: How LLMs Handle Million-Token Contexts Without Melting Your GPU
How sparse attention cuts LLM inference cost by 10x on long contexts. Covers DeepSeek NSA, MInference, H2O, and The Sparse Frontier's findings.
AsyncTLS: 4.7x Faster Long-Context LLM Inference With Two-Level Sparse Attention
AsyncTLS sparse attention fuses block filtering, token selection, and async KV cache offloading for 1.3-4.7x throughput gains at 48k-96k token contexts.
Recursive Language Models: How RLMs Beat Long Context
Recursive language models treat a huge prompt as a Python variable the model can grep and recurse over. MIT's paper shows it beats GPT-5 on long context.
TriAttention Compresses KV Cache 10.7x — How Trigonometry Fixed Long-Context Reasoning
TriAttention uses pre-RoPE vector concentration and trigonometric scoring to compress KV cache 10.7x while matching full attention accuracy on reasoning tasks.