Poster · Sydney
OCTOPUS: Optimized KV Cache for Transformers via Octahedral Parametrization Under optimal Squared error quantization
- When
- Thu 10 Dec 2026, 17:00–20:00 AEDT
- Where
- Hall 1-4, ICC Sydney
Exhibition Centre, Level 1 - Session
- Sydney Poster Session 6
- Authors
- Mark Boss, Vikram Voleti, Simon Donné, Shimon Vainer
- Organisations
- Unity Technologies, Stability AI
- QRS classification
- Attention and transformers, Efficient and distributed ML systems