Reaction
VSA is claimed to improve reasoning at 100K tokens by up to 9.5 points
A user says the proprietary long-context solution can be swapped into DS-lineage models, and possibly others.
TLDR
A user describes VSA as a proprietary long-context solution that can be swapped into DS-lineage models and speculates it could work with others. They highlight a claim that VSA improves reasoning at 100K tokens by up to 9.5 points.
Combined views
7.5K
2 Sources, first seen ago