zunhaisu

zunhai's activity

• All
• Models
• Datasets
• Spaces
• Buckets
• Papers
• Collections
• Community
• Posts
• Upvotes
• Likes
• Articles

authored a paper 16 days ago

DecisionPivot: A Diagnostic Benchmark for Decision Models in LLM Systems

Paper • 2610.28522 • Published 17 days ago • 25

upvoted a paper 16 days ago

DecisionPivot: A Diagnostic Benchmark for Decision Models in LLM Systems

Paper • 2610.28522 • Published 17 days ago • 25

upvoted a collection 18 days ago

StartLux-Decision-v1.1 AutoThink

Collection

v1.1, October 2026. Read-or-Think: one-pass answer, a short thought only when unsure. 4B, 9B, 27B. • 13 items • Updated 23 days ago • 9

submitted a paper to Daily Papers 18 days ago

ReaDy: Recurrent States as Dynamic Context for Speculative Drafting in Hybrid Linear Attention Large Language Models

Paper • 2610.20503 • Published 23 days ago • 9

authored a paper 19 days ago

ReaDy: Recurrent States as Dynamic Context for Speculative Drafting in Hybrid Linear Attention Large Language Models

Paper • 2610.20503 • Published 23 days ago • 9

upvoted a paper 20 days ago

ReaDy: Recurrent States as Dynamic Context for Speculative Drafting in Hybrid Linear Attention Large Language Models

Paper • 2610.20503 • Published 23 days ago • 9

authored a paper 20 days ago

StartLux-Decision: AutoThink Decision Models with Reinforcement Learning for Calibrated Deliberative Decisions

Paper • 2610.17309 • Published 25 days ago • 27

upvoted a paper 21 days ago

StartLux-Decision: AutoThink Decision Models with Reinforcement Learning for Calibrated Deliberative Decisions

Paper • 2610.17309 • Published 25 days ago • 27

upvoted a collection 23 days ago

StartLux-Decision-v1.1 Instant

Collection

v1.1, October 2026. Typed decisions in one forward pass: a probability for every option. 0.8B to 27B. • 21 items • Updated 23 days ago • 6

authored a paper 25 days ago

QuantMLA: Function-Aligned Dual-Path Quantization for Low-Bit MLA KV Caching

Paper • 2609.36760 • Published Sep 30 • 10

upvoted 6 papers 25 days ago

SnapMLA: Efficient Long-Context MLA Decoding via Hardware-Aware FP8 Quantized Pipelining

Paper • 2602.10718 • Published Apr 28 • 1

XStreamVGGT: Extremely Memory-Efficient Streaming Vision Geometry Grounded Transformer with KV Cache Compression

Paper • 2601.01204 • Published Jan 3 • 1

RotateKV: Accurate and Robust 2-Bit KV Cache Quantization for LLMs via Outlier-Aware Adaptive Rotations

Paper • 2501.16383 • Published Jan 25, 2025 • 1

AKVQ-VL: Attention-Aware KV Cache Adaptive 2-Bit Quantization for Vision-Language Models

Paper • 2501.15021 • Published Jan 25, 2025 • 1

LongCat-Flash Technical Report

Paper • 2509.01322 • Published Sep 1, 2025 • 12

QuantMLA: Function-Aligned Dual-Path Quantization for Low-Bit MLA KV Caching

Paper • 2609.36760 • Published Sep 30 • 10

submitted a paper to Daily Papers 25 days ago

QuantMLA: Function-Aligned Dual-Path Quantization for Low-Bit MLA KV Caching

Paper • 2609.36760 • Published Sep 30 • 10

upvoted a paper 29 days ago

TRIAGE: Direction-Aware Mismatch Stabilization of Native NVFP4 Reinforcement Learning

Paper • 2610.07043 • Published Oct 5 • 35

upvoted a collection about 1 month ago

StartLux-Decision-v1

Collection

Typed decision models from 0.8B to 27B. Every question comes back with a probability for each option. • 21 items • Updated 27 days ago • 13

upvoted a collection 2 months ago

Massive Activations in Hybrid Linear Attention Models

Collection

Official PAS, ISP, gating-ablation, and scale-study checkpoints for Massive Activations in Hybrid Linear Attention Models. • 10 items • Updated Aug 13 • 2