LIVE PULSE
3.9 Anthropic CEO Amodei calls for slower AI development and shared safety rules11 src2.1 Agility Robotics unveils Digit 5 humanoid for warehouses and factories2 src1.9 Apple ships rebuilt Siri with Google Gemini, but not in the EU2 src1.7 Siri AI in macOS 27 Golden Gate: FAQ, Germany availability, privacy questions2 src1.3 Sam Altman says OpenAI will not go public in 2026, citing AI safety concerns5 src1.1 OpenAI contractors review real ChatGPT conversations to rate responses, report says2 src1.0 Anthropic data retention policy prompts firms to limit Claude use for sensitive work1 src1.0 arXiv paper proposes emotion regulation framework for empathetic speech dialogue in audio-language models1 src1.0 Paper Studies Graph Matching Relaxations for Supervised Graph Prediction1 src1.0 arXiv Paper Proposes Framework for Cognitive Attribution in Acquired Representations1 src3.9 Anthropic CEO Amodei calls for slower AI development and shared safety rules11 src2.1 Agility Robotics unveils Digit 5 humanoid for warehouses and factories2 src1.9 Apple ships rebuilt Siri with Google Gemini, but not in the EU2 src1.7 Siri AI in macOS 27 Golden Gate: FAQ, Germany availability, privacy questions2 src1.3 Sam Altman says OpenAI will not go public in 2026, citing AI safety concerns5 src1.1 OpenAI contractors review real ChatGPT conversations to rate responses, report says2 src1.0 Anthropic data retention policy prompts firms to limit Claude use for sensitive work1 src1.0 arXiv paper proposes emotion regulation framework for empathetic speech dialogue in audio-language models1 src1.0 Paper Studies Graph Matching Relaxations for Supervised Graph Prediction1 src1.0 arXiv Paper Proposes Framework for Cognitive Attribution in Acquired Representations1 src
HEATPULSEAI MAGAZINES
FLIP · FOLLOW · SAVE

Gated DeltaNet

model1 events
papersTODAY 04:00 UTC

SpectralShift Method Extends Context Window of Gated DeltaNet Models

A new arXiv paper proposes SpectralShift, a technique that reparameterizes the spectral properties of gated DeltaNet layers to stretch their usable context window. The authors note that linear attention layers are increasingly used in place of softmax attention for long-context work, but existing extension methods typically rely on continued pretraining without altering the layer internals. Their approach instead modifies the layers themselves, aiming to make context extension more effective.