REVIEW 4 cited by
True Knowledge Comes from Practice: Aligning LLMs with Embodied Environments via Reinforcement Learning
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Despite the impressive performance across numerous tasks, large language models (LLMs) often fail in solving simple decision-making tasks due to the misalignment of the knowledge in LLMs with environments. On the contrary, reinforcement learning (RL) agents learn policies from scratch, which makes them always align with environments but difficult to incorporate prior knowledge for efficient explorations. To narrow the gap, we propose TWOSOME, a novel general online framework that deploys LLMs as decision-making agents to efficiently interact and align with embodied environments via RL without requiring any prepared datasets or prior knowledge of the environments. Firstly, we query the joint probabilities of each valid action with LLMs to form behavior policies. Then, to enhance the stability and robustness of the policies, we propose two normalization methods and summarize four prompt design principles. Finally, we design a novel parameter-efficient training architecture where the actor and critic share one frozen LLM equipped with low-rank adapters (LoRA) updated by PPO. We conduct extensive experiments to evaluate TWOSOME. i) TWOSOME exhibits significantly better sample efficiency and performance compared to the conventional RL method, PPO, and prompt tuning method, SayCan, in both classical decision-making environment, Overcooked, and simulated household environment, VirtualHome. ii) Benefiting from LLMs' open-vocabulary feature, TWOSOME shows superior generalization ability to unseen tasks. iii) Under our framework, there is no significant loss of the LLMs' original ability during online PPO finetuning.
Forward citations
Cited by 4 Pith papers
-
CLEANER: Self-Purified Trajectories Boost Agentic Reinforcement Learning
Replacing execution-error segments in RL trajectories with the model's own successful self-corrections improves accuracy on math and code benchmarks while cutting training steps by roughly two-thirds.
-
BIMgent: Towards Autonomous Building Modeling via Computer-use Agents
BIMgent, a GUI-controlling LLM agent, completes 32% of BIM building modeling tasks end-to-end, outperforming baseline computer-use agents that complete none.
-
IDEA: Augmenting Design Intelligence through Design Space Exploration
IDEA combines LLM-generated constraints with Monte Carlo Tree Search over a formal design space to automate design decision-making in data storytelling and pictorial visualization.
-
Shapley-Coop: Credit Assignment for Emergent Cooperation in Self-Interested LLM Agents
Shapley-Coop asks LLM agents to negotiate prices for contributions based on Shapley value reasoning, improving cooperation and reward fairness in three multi-agent tasks.
Discussion (0). Continue with ORCID to comment.