REVIEW 18 cited by
ArCHer: Training Language Model Agents via Hierarchical Multi-Turn RL
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
ArCHer: Training Language Model Agents via Hierarchical Multi-Turn RL
read the original abstract
A broad use case of large language models (LLMs) is in goal-directed decision-making tasks (or "agent" tasks), where an LLM needs to not just generate completions for a given prompt, but rather make intelligent decisions over a multi-turn interaction to accomplish a task (e.g., when interacting with the web, using tools, or providing customer support). Reinforcement learning (RL) provides a general paradigm to address such agent tasks, but current RL methods for LLMs largely focus on optimizing single-turn rewards. By construction, most single-turn RL methods cannot endow LLMs with the ability to intelligently seek information over multiple turns, perform credit assignment, or reason about their past actions -- all of which are critical in agent tasks. This raises the question: how can we design effective and efficient multi-turn RL algorithms for LLMs? In this paper, we develop a framework for building multi-turn RL algorithms for fine-tuning LLMs, that preserves the flexibility of existing single-turn RL methods for LLMs (e.g., proximal policy optimization), while accommodating multiple turns, long horizons, and delayed rewards effectively. To do this, our framework adopts a hierarchical RL approach and runs two RL algorithms in parallel: a high-level off-policy value-based RL algorithm to aggregate reward over utterances, and a low-level RL algorithm that utilizes this high-level value function to train a token policy within each utterance or turn. Our hierarchical framework, Actor-Critic Framework with a Hierarchical Structure (ArCHer), can also give rise to other RL methods. Empirically, we find that ArCHer significantly improves efficiency and performance on agent tasks, attaining a sample efficiency of about 100x over existing methods, while also improving with larger model capacity (upto the 7 billion scale that we tested on).
Forward citations
Cited by 18 Pith papers
-
The Dark Room in the Reward Channel: Dense Prediction Rewards Collapse GRPO-Trained LLM Agents -- and What Actually Works
A potential-based prediction reward collapses GRPO-trained LLM agents into a predictable 'dark room' state, and the collapse is caused by GRPO's std normalization rather than by the reward's magnitude.
-
Unlocking Proactivity in Task-Oriented Dialogue
Introduces a Cognitive User Simulator modeling stratified personas with hidden concerns and Simulator-Induced Asymmetric-View Policy Optimization to unlock proactive behavior in task-oriented dialogue agents.
-
Teaching LLMs to Self-Evolve: Cultivating Core Meta-Skills with Reinforcement Learning
MetaEvolve trains LLMs with reinforcement learning on synthesized code-refinement trajectories, reporting large gains on coding and numerical-optimization benchmarks.
-
Diagnosing Task Insensitivity in Language Agents
The paper diagnoses task insensitivity in LLM agents as a cause of weak OOD generalization, links it to attention drift, and proposes Task-Perturbed NLL Optimization as a contrastive regularizer to improve task dependence.
-
When Denser Credit Is Not Enough: Evidence-Calibrated Policy Optimization for Long-Horizon LLM Agent Training
ECPO improves GiGPO by shrinking low-count action advantages and suppressing noisy anchor states, yielding +5.2/+7.3 success gains on ALFWorld/WebShop with Qwen2.5-1.5B models at negligible extra cost.
-
ECHO: Terminal Agents Learn World Models for Free
ECHO is a hybrid RL objective that trains agents to predict environment observation tokens from their actions, doubling GRPO pass@1 on TerminalBench-2.0 while improving dynamics prediction on held-out trajectories.
-
Rewarding Beliefs, Not Actions: Consistency-Guided Credit Assignment for Long-Horizon Agents
ReBel uses belief-consistency supervision and belief-aware grouping to improve credit assignment in long-horizon RL for LLM agents, achieving up to 20.4 percentage points higher success and 2.1x better sample efficien...
-
SOD: Step-wise On-policy Distillation for Small Language Model Agents
SOD reweights on-policy distillation strength step-by-step using divergence to stabilize tool use in small language model agents, yielding up to 20.86% gains and 26.13% on AIME 2025 for a 0.6B model.
-
From History to State: Constant-Context Skill Learning for LLM Agents
Constant-context skill learning trains reusable task-family modules for LLM agents using a deterministic state block for progress tracking and subgoal rewards, achieving 89.6% unseen success on ALFWorld, 76.8% on WebS...
-
From Refusal to Recovery: A Control-Theoretic Approach to Generative AI Guardrails
Control-theoretic guardrails enable proactive correction of risky LLM agent actions in latent space, preventing catastrophes like collisions or bankruptcy while preserving task performance in simulated environments.
-
Training Language Models to Self-Correct via Reinforcement Learning
SCoRe uses multi-turn online RL with regularization on self-generated traces to improve LLM self-correction, achieving 15.6% and 9.1% gains on MATH and HumanEval for Gemini models.
-
ABot-AgentOS: A General Robotic Agent OS with Lifelong Multi-modal Memory
A hierarchical robotic Agent OS with source-grounded multi-modal graph memory and split-gated self-evolution improves long-horizon embodied execution and memory QA over single-controller and prior memory baselines.
-
Modularized Reinforcement Learning on LLMs: From MDP Creation to Exploration and Learning
Survey mapping RL techniques onto LLM training and highlighting gaps in value-based, off-policy, and bootstrapping methods.
-
SIRI: Self-Internalizing Reinforcement Learning with Intrinsic Skills for LLM Agent Training
SIRI trains LLM agents to discover, validate, and internalize reusable skills from their own rollouts without external generators or inference-time skill banks, yielding gains on ALFWorld and WebShop.
-
Trust Region On-Policy Distillation
TrOPD stabilizes on-policy distillation for LLMs with trust-region learning, outlier estimation, and off-policy guidance, outperforming prior OPD methods on reasoning and code benchmarks.
-
Unlocking Proactivity in Task-Oriented Dialogue
Introduces a user concern simulator and asymmetric policy optimization to enable proactive behavior in task-oriented dialogues by using latent concerns as a training signal.
-
StraTA: Incentivizing Agentic Reinforcement Learning with Strategic Trajectory Abstraction
StraTA improves LLM agent success rates to 93.1% on ALFWorld and 84.2% on WebShop by sampling a compact initial strategy and training it jointly with action execution via hierarchical GRPO-style rollouts.
-
Towards Large Reasoning Models: A Survey of Reinforced Reasoning with Large Language Models
The paper surveys reinforced reasoning techniques for LLMs, covering automated data construction, learning-to-reason methods, and test-time scaling as steps toward Large Reasoning Models.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.