Attention Sink in Transformers: A Survey on Utilization, Interpretation, and Mitigation
Paper • 2604.10098 • Published • 68
Prince Canuma PRO
prince-canuma
AI & ML interests
None yet
Recent Activity
upvoted a paper 1 day ago
The Past Is Not Past: Memory-Enhanced Dynamic Reward Shaping published a model 8 days ago
mlx-community/GLM-5.1-nvfp4Organizations
![Unofficial Mistral Community [deprecated]'s profile picture](https://cdn-avatars.huggingface.co/v1/production/uploads/6141a88b3a0ec78603c9e784/HAdAzcs1CDB9OxgxT2W9K.png)
upvoted 2
papers 1 day ago
published 8
models 8 days ago
upvoted 5
papers 9 days ago
Plug-and-Play 1.x-Bit KV Cache Quantization for Video Large Language Models
Paper • 2503.16257 • Published • 28
KIVI: A Tuning-Free Asymmetric 2bit Quantization for KV Cache
Paper • 2402.02750 • Published • 5
Token Warping Helps MLLMs Look from Nearby Viewpoints
Paper • 2604.02870 • Published • 33
Self-Distilled RLVR
Paper • 2604.03128 • Published • 160
Turbo Sparse: Achieving LLM SOTA Performance with Minimal Activated Parameters
Paper • 2406.05955 • Published • 28
updated 5
models 12 days ago
mlx-community/gemma-4-e4b-mxfp8
Any-to-Any • 3B • Updated • 4.09k • 1
mlx-community/gemma-4-e4b-it-mxfp8
Any-to-Any • 3B • Updated • 13.5k • 8
mlx-community/gemma-4-e4b-it-bf16
Any-to-Any • 8B • Updated • 3.01k • 3
mlx-community/gemma-4-e4b-it-8bit
Any-to-Any • 3B • Updated • 39.7k • 14
mlx-community/gemma-4-e4b-it-6bit
Any-to-Any • 2B • Updated • 2.2k • 1