Junjie Wang

Verified execution experience can be transformed into a reusable resource, boosting model performance on complex workflows by up to 15.5 percentage points.

AutoResearch achieves a 1.85-point improvement in mean Recall while reducing audit-confirmed issues, showcasing a new standard for reliability in autonomous research systems.

Public skill version histories can dramatically enhance task-specific skill evolution, leading to significant performance gains in AI agents.

KinRT achieves over 23% improvement in expert routing accuracy by leveraging kinematic archetypes, transforming how MoE systems can operate without kinematic signals at inference.

Babies leverage touch to enhance their visual learning, revealing a crucial link between tactile experiences and cognitive development.

Mainstream models falter in multi-reference image generation, but DyRef's innovative training framework boosts their performance significantly.

Quadrotors can autonomously recover from falls in arbitrary orientations using only lightweight sensors, achieving robust performance even in challenging conditions.

Environment engineering, not just agent workflows, is the key to unlocking the full potential of autonomous scientific discovery, as demonstrated by EurekAgent's record-breaking results.

Smaller models can significantly enhance the training of larger models by providing structured exploration signals that improve performance without the noise of traditional methods.

Ditch the documentation: compact "Gene" representations of experience outperform documentation-heavy "Skill" packages for test-time control and iterative evolution in scientific code-solving.