paper

One-Shot Imitation Learning

paperactiveone-shot-imitation-learning-87057c12·1 events·first seen 28d ago

Aliases: One-Shot Imitation Learning

Co-occurring entities

More like this (12)

Imitation Learning few-shot learning zero-shot learning Language Models are Few-Shot Learners zero-shot evaluation Difference-Aware Retrieval Policies for Imitation Learning Reasoning Imitation Q-learning Soft Q-Learning self-play reinforcement learning Matryoshka Representation Learning on-policy self-distillation

Recent events (1)

4Openai Blog·28d ago·source ↗

One-shot imitation learning

OpenAI published research on one-shot imitation learning, a technique enabling agents to learn new tasks from a single demonstration. The approach allows a policy network to observe a demonstration and immediately generalize to new instances of the same task without additional training. This was an early contribution to the field of meta-learning and few-shot generalization in robotics and sequential decision-making.

Agent and Tool Ecosystem Alignment and RLHF One-Shot Imitation Learning meta-learning Imitation Learning +1 more