SGFT uses a simulation-trained value function to guide real-world exploration via potential-based reward shaping and short-horizon objectives, substantially improving fine-tuning sample efficiency.
∞X t=0 γt¯r(st) # = V π real(s) − Vs(s0) ≥ Eρπ real(s)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.RO 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Rapidly Adapting Policies to the Real World via Simulation-Guided Fine-Tuning
SGFT uses a simulation-trained value function to guide real-world exploration via potential-based reward shaping and short-horizon objectives, substantially improving fine-tuning sample efficiency.