Speculative decoding

Quasar

Quasar: Quantized Self-Speculative Acceleration for Rapid Inference via Memory-Efficient Verification

On the frontierfirst seen Mar 2, 2026

Current frontier — recent, not yet superseded in the knowledge base

0 papers critique it · 0 beat it on benchmarks

What to use instead

Recent methods in the same sub-problem, not yet superseded in the knowledge base — arXiv benchmark leaders, not vetted production recommendations.