Random Attention: Rethinking KV Cache Eviction for Efficient Reasoning
Paper • 2609.03430 • Published • 158 •
2
Librarian Bot (Bot)
librarian-bot
AI & ML interests
I am a friendly librarian bot working to help improve metadata on the 🤗 hub. Run by Hugging Face'ss Machine Learning Librarian (HF username: davanstrien)
Recent Activity
new activity about 4 hours ago
ruotian/ClickContrast:Librarian Bot: Add language metadata for dataset updated a dataset about 8 hours ago
librarian-bots/paper-recommendations-v2 commentedon a paper about 8 hours ago
Random Attention: Rethinking KV Cache Eviction for Efficient ReasoningOrganizations
Librarian Bot: Add language metadata for dataset
#2 opened about 4 hours ago
by librarian-bot
commented 19
papers about 8 hours ago
Editable Visual Design
Paper • 2609.04034 • Published • 39 •
2
Locked at the Entrance, Open Inside: Where RLVR Narrows the Solution Space
Paper • 2608.29188 • Published • 8 •
2
Terminal-Universe: Turning Agent Trajectories into Scalable Terminal Environments
Paper • 2609.04148 • Published • 223 •
1
PACE: Towards Surfacing Hidden Conflicts in User Requests
Paper • 2609.03293 • Published • 22 •
2
LatentPress: Context Compression Beyond Text and Vision
Paper • 2609.01507 • Published • 107 •
2
Compile by Training: Turning Natural-Language Specifications into Local Neural Functions
Paper • 2609.04199 • Published • 271 •
3
Let Confidence Change, Not the Prediction: Prediction-Preserving Repair for Post-hoc Calibration
Paper • 2609.01072 • Published • 13 •
2
The Missing Temporal Link: Temporal Context Routing for Script-Driven Audio-Video Generation
Paper • 2609.02367 • Published • 33 •
2
QCell: Recombining and Aligning Cell Queries for Overlapping Instance Segmentation
Paper • 2608.29253 • Published • 13 •
5
CORE: Improving Compositional Reasoning in MLLM Embedding via Reranker Distillation
Paper • 2609.04083 • Published • 24 •
2
DRACO: Fine-Grained Credit Assignment with Dynamic Rubrics for Long-Horizon Agent Training
Paper • 2609.04094 • Published • 23 •
2
LLaDA-Image: Building Strong Image Generators with Fully Open Training Recipes
Paper • 2609.03796 • Published • 204 •
3
Knowing When Not to Reuse: Conditional Experience Transfer in Autonomous LLM Post-Training
Paper • 2608.26730 • Published • 149 •
2
Beyond Retrieval: Progressive Latent Memory Evolution for Streaming Video Understanding
Paper • 2609.04131 • Published • 29 •
2
Rethinking On-Policy Distillation of Large Language Models II: One Training Example
Paper • 2609.04172 • Published • 73 •
2
A Common Measure of Communication for Speech Brain-Computer Interfaces
Paper • 2609.02887 • Published • 8 •
2
Puffin-World: Scaling a Unified Multimodal Model with Native 3D World States
Paper • 2609.04196 • Published • 65 •
3
Select, Compress, Reinvest: A Controlled Study of Visual-Token Allocation in Long-Video MLLMs
Paper • 2609.03820 • Published • 15 •
3