FuseReg: Regularizing Layer Fusion Mitigates the Reconstruction-Generation Gap in Representation Autoencoders
Paper • 2609.31620 • Published • 19
YehuiTang
WizardTY
·
AI & ML interests
None yet
Recent Activity
posted an update 8 days ago
If you want a hareness tuned for your scenario, try Gear. We just open-sourced a RSI infra for adapting agents to real-world tasks. On Automation bench, Gear-tuned DSH + Luna beats Astra + Codex on objective completion rate (84% vs 84%), with 95% lower API cost.
try Gear: https://github.com/rsi-gear/gear authored a paper over 1 year ago
DualTHOR: A Dual-Arm Humanoid Simulation Platform for Contingency-Aware
PlanningOrganizations
None yet
upvoted a paper 41 minutes ago
posted an update 8 days ago
Post
46
If you want a hareness tuned for your scenario, try Gear. We just open-sourced a RSI infra for adapting agents to real-world tasks. On Automation bench, Gear-tuned DSH + Luna beats Astra + Codex on objective completion rate (84% vs 84%), with 95% lower API cost.
try Gear: https://github.com/rsi-gear/gear
try Gear: https://github.com/rsi-gear/gear
authored a
paper over 1 year ago
upvoted a paper over 1 year ago
upvoted a paper almost 2 years ago
liked a
model about 2 years ago
liked a
dataset over 2 years ago