Negligible in Size, Significant in Effect: On Scale Vectors in Large Language Models Paper • 2605.26895 • Published 2 days ago • 13 • 3
WTF GENIUS PAPERS Collection Papers that made me appreciate my major and my life a little more. obs=Observation, innov=Innovation. Most papers are abt improving tiny models. • 154 items • Updated 1 day ago • 30
Negligible in Size, Significant in Effect: On Scale Vectors in Large Language Models Paper • 2605.26895 • Published 2 days ago • 13
Beyond Final Answers: Auditing Trajectory-Level Hallucinations in Multi-Agent Industrial Workflows Paper • 2605.24219 • Published 2 days ago • 7
The MiniMax-M2 Series: Mini Activations Unleashing Max Real-World Intelligence Paper • 2605.26494 • Published 2 days ago • 30
WTF GENIUS PAPERS Collection Papers that made me appreciate my major and my life a little more. obs=Observation, innov=Innovation. Most papers are abt improving tiny models. • 154 items • Updated 1 day ago • 30
Efficient-DLM: From Autoregressive to Diffusion Language Models, and Beyond in Speed Paper • 2512.14067 • Published Dec 16, 2025 • 18
view article Article Towards Speed-of-Light Text Generation with Nemotron-Labs Diffusion Language Models nvidia • 5 days ago • 25
HRM-Text: Efficient Pretraining Beyond Scaling Paper • 2605.20613 • Published 8 days ago • 88 • 2
WTF GENIUS PAPERS Collection Papers that made me appreciate my major and my life a little more. obs=Observation, innov=Innovation. Most papers are abt improving tiny models. • 154 items • Updated 1 day ago • 30
ThriftAttention: Selective Mixed Precision for Long-Context FP4 Attention Paper • 2605.23081 • Published 7 days ago • 31
DVAO: Dynamic Variance-adaptive Advantage Optimization for Multi-reward Reinforcement Learning Paper • 2605.25604 • Published 3 days ago • 126
WTF GENIUS PAPERS Collection Papers that made me appreciate my major and my life a little more. obs=Observation, innov=Innovation. Most papers are abt improving tiny models. • 154 items • Updated 1 day ago • 30
KVServe: Service-Aware KV Cache Compression for Communication-Efficient Disaggregated LLM Serving Paper • 2605.13734 • Published 15 days ago • 12
WTF GENIUS PAPERS Collection Papers that made me appreciate my major and my life a little more. obs=Observation, innov=Innovation. Most papers are abt improving tiny models. • 154 items • Updated 1 day ago • 30
Forecasting Downstream Performance of LLMs With Proxy Metrics Paper • 2605.18607 • Published 10 days ago • 14