ComfyAI·Apr 25Bridging the Data Gap: How H-RDT Leverages Human Video to Master Robot ManipulationThe biggest bottleneck in the Robot Learning (VLA) field right now is painfully obvious: teleoperation data is too expensive and too…
ComfyAI·Apr 24Scaling Down the “Wait…”: How TIV Fixes the Overthinking Problem in Large Reasoning ModelsThe rise of Large Reasoning Models (LRMs) like DeepSeek-R1 has been a double-edged sword. While their reasoning capabilities are…
ComfyAI·Apr 23Beyond Just Differentiable: Why Fudan’s VeteranAD is a Game-Changer for End-to-End Autonomous…The current landscape of end-to-end autonomous driving is dominated by a “Perception $\rightarrow$ Planning” serial pipeline. While heavy…
ComfyAI·Apr 22Can AI Actually “Fold” a Piece of Paper? Why Origami is the New Stress Test for Multimodal ModelsFor a long time, paper-folding puzzles have been a staple for testing spatial intelligence in everyone from elementary school students to…
ComfyAI·Apr 15Breaking the “Illusion” of LLM Unlearning: Why ALMPU is a Game-Changer for Model SafetyA major headache in current AI safety is that Large Language Model (LLM) unlearning often feels like a game of Whac-A-Mole. You think…
