OIR relabels failed trajectories via an LLM into open-ended instructions and trains a unified instruction-following policy, outperforming PQN and ELLM on Craftax.
Grounding language for transfer in deep reinforcement learning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Learning Instruction-Following Policies through Open-Ended Instruction Relabeling with Large Language Models
OIR relabels failed trajectories via an LLM into open-ended instructions and trains a unified instruction-following policy, outperforming PQN and ELLM on Craftax.