Robot One-Shot Imitation: Bigger Models, Little Real Progress

Researchers contrast one-shot imitation learning in 2020 versus today: specialized architectures gave way to generalist foundation models like GEN-1.5 and Skild that learn from video prompts. Yet Animesh Garg notes one-shot imitation performance has not meaningfully improved with scale.

2026-08-30 ~ 2026-08-30 · 2 related posts

• Robot One-Shot Imitation: From Specialized Architectures to Generalist Models — animesh_garg · 2026-08-30
• One-shot imitation performance remains flat despite model scale — animesh_garg · 2026-08-30