Evict3R
Training-Free Token Eviction for Memory-Bounded Streaming Visual Geometry Transformers
bounds the KV cache of streaming 3D transformers without retraining; peak memory 18.63 GB → 9.39 GB on 7-Scenes
projects and paper notes on world models and representation learning, most with something to poke at
Training-Free Token Eviction for Memory-Bounded Streaming Visual Geometry Transformers
bounds the KV cache of streaming 3D transformers without retraining; peak memory 18.63 GB → 9.39 GB on 7-Scenes
No notes published yet. These are next: