Maximum activations in modern open LLMs span nearly four orders of magnitude across families, with MoE models exhibiting 14-23x lower peaks than dense counterparts and residual streams carrying the global max in most cases.
Vision transformers need registers
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
citation-role summary
background 2
citation-polarity summary
years
2026 4verdicts
UNVERDICTED 4roles
background 2representative citing papers
Sinks are equivalent to hard attention switches that zero out outputs and are cheaper than diagonal patterns when self-communication is allowed, closing the gap between oversmoothing prevention needs and what sinks provide.
Decouples semantic and spatial tokens in NVS transformers to resolve representation ambiguity, yielding consistent gains with near-zero added latency.
A 12-step single-block recurrent ViT-B reaches accuracy comparable to a standard ViT-B on ImageNet-1K while using an order of magnitude fewer parameters.
citing papers explorer
-
Measuring Maximum Activations in Open Large Language Models
Maximum activations in modern open LLMs span nearly four orders of magnitude across families, with MoE models exhibiting 14-23x lower peaks than dense counterparts and residual streams carrying the global max in most cases.
-
Sink vs. diagonal patterns as mechanisms for attention switch and oversmoothing prevention
Sinks are equivalent to hard attention switches that zero out outputs and are cheaper than diagonal patterns when self-communication is allowed, closing the gap between oversmoothing prevention needs and what sinks provide.
-
Resolving Representation Ambiguity in Feedforward Novel View Synthesis Transformer via Semantic-Spatial Decoupling
Decouples semantic and spatial tokens in NVS transformers to resolve representation ambiguity, yielding consistent gains with near-zero added latency.
-
bViT: Investigating Single-Block Recurrence in Vision Transformers for Image Recognition
A 12-step single-block recurrent ViT-B reaches accuracy comparable to a standard ViT-B on ImageNet-1K while using an order of magnitude fewer parameters.