zunhaisu

zunhaisu on Hugging Face Daily Papers: 6 papers, 1 in the top 3 of their day, 271 upvotes.

1. Massive Activations in Hybrid Linear Attention Large Language Models: Pre-Attention Spikes and Inter-Spike Plateaus 30 upvotes, #11 of 2026-08-14
2. UniMoMo: Expert Merging-Based MoE Acceleration for Large Recommendation Models 14 upvotes, #10 of 2026-08-12
3. OScaR: The Occam's Razor for Extreme KV Cache Quantization in LLMs and Beyond 39 upvotes, #8 of 2026-05-21
4. Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation 75 upvotes, #4 of 2026-04-14
5. Locate, Steer, and Improve: A Practical Survey of Actionable Mechanistic Interpretability in Large Language Models 46 upvotes, #4 of 2026-01-21
6. DoPE: Denoising Rotary Position Embedding 89 upvotes, #1 of 2025-11-17

Data: hysts-bot-data/daily-papers-stats and the Daily Papers API. Open data: tardellirs/paper-pulse-data. Sister project: Model Pulse, the download history of every model on the Hub.