Fine-Grained Credit Assignment

Follow this topic Ask about this topic

Fine-grained credit assignment asks which decisions, actions, or reasoning steps contributed to an outcome, especially when many steps receive only a single success or failure signal. In reinforcement learning for language-model agents and multi-agent systems, it can make training more efficient and clarify responsibility; related methods also analyze contributions in recommendation, ranking, advertising, and other settings. Approaches include comparing counterfactual trajectories, using process or verifier feedback, tracing dependencies between actions and outcomes, and allocating marginal contributions with methods such as Shapley values. The central tension is that detailed signals can improve learning but are hard to identify reliably from sparse outcomes, noisy evaluators, and interacting actions, while obtaining them may add cost or misdirect updates.

Latest papers

43 papers9 papers in the last four weeks, up 125% on the four weeks before. 0.1% of all new papers.

SpatialPC: Deep Computation with Sublinear Local Credit Propagation
Oct 26, 2026Benjamin Prada, Shion Matsumoto, Ankur Arjun MaliLocal Learning Fine-Grained Credit Assignment
Trustworthy Knowledge-Driven AI Lab Bellini College of Artificial Intelligence, Cybersecurity and Computing University of South Florida Tampa, FL 33620
When Fine-Grained Credit Assignment Helps and Hurts: Multimodal Agentic Reinforcement Learning through Video Editing
Oct 26, 2026Lei Zhao, Zongheng Cao, Xinyu HuReinforcement Learning Fine-Grained Credit Assignment
University of Pennsylvania · Philo Labs
P2TPO: Structure-Aligned Page-to-Token Credit Assignment for Full-Page Document Parsing
Oct 15, 2026Yong Zu, Ninghua Yang, Zezhong Tan +3Document Parsing Token-Level Credit Assignment
Qwen Business Unit, Alibaba · Xidian University
CASI: Credit Assignment via Set Intervention for Agentic Retrieval
Oct 14, 2026Xinyi Cheng, Tianfu Zhang, Jiwen Li +5Credit Assignment Fine-Grained Credit Assignment
Xiaohongshu Inc.
Of Errors and Echoes: Responsibility-Aware Credit Assignment for Tool-Use Agents
Oct 12, 2026Xingjian Diao, Tianyu Yang, Wenjun Huang +8Credit Assignment Fine-Grained Credit Assignment
Dartmouth College · University of Notre Dame · University of California, Irvine +3
Learning after Execution with Action-Specific Credit
Oct 10, 2026Yufeng Tian, Jiahao Tang, Minghao WuReinforcement Learning Credit Assignment
The Chinese University of Hong Kong, Shenzhen, China · Central South University, Changsha, China
Credit Attribution in Timestamped Soccer Networks Using Multivariate Hawkes Processes
Oct 8, 2026Phillip W. Kornberg, Kevin S. XuSoccer Analytics Fine-Grained Credit Assignment
Case Western Reserve University 10900 Euclid Avenue, Cleveland, OH 44106-7071, USA
Credit Where It Matters: Dependency-Aware Policy Optimization for Terminal Agents
Oct 2, 2026Yu Li, Guangfeng Cai, Long-Fei Li +5Credit Assignment Step-Level Credit Assignment in RL
School of Computer Science and Engineering, Southeast University, Nanjing, China · Huawei Noah’s Ark Lab, China
My FAULT: Self-Diagnosis as Credit Assignment in Self-Evolving Agentic Reinforcement Learning
Oct 1, 2026Yihua Zhu, Qianying Liu, Weixu Qiao +10Agentic RL Credit Assignment in RL
Alibaba Group · Kyoto University · NII LLMC +6
SCA: Spatial Credit Assignment for Reinforcement Learning of GUI Agents
Sep 29, 2026Shengtian Yang, Ziyu Xiong, Kaibing Yang +6Reinforcement Learning RL for GUI Agents
Southeast University, Nanjing, China · Kuaishou Technology, Beijing, China · Unaffiliated
Targeting Pivotal Decisions for Credit Assignment in Agentic Reinforcement Learning
Sep 28, 2026Dongwon Jung, Hemanth Neelgund Ramesh, Yifan Wang +7Agentic RL Group Relative Policy Optimization
University of California, Davis · Microsoft · University of Washington
From Interests to Semantic IDs: Retrieval-Grounded Credit Assignment for Generative Recommendation
Sep 24, 2026Mengdan Zhu, Yufan Zhao, Yao Zhao +5Semantic ID Recommendation Credit Assignment
Emory University Atlanta, GA, USA · Microsoft Redmond, WA, USA · Cornell University Ithaca, NY, USA
CausalLoss-Fin: Attributing Financial-Agent Loss to Decisions and Infrastructure Faults
Sep 22, 2026Abhishek SharmaAgent Failure Analysis Financial Services
From Citations to Contributions: LLM-Assisted Credit Scoring of Research Articles
Sep 7, 2026Sana Ebrahimi, Suraj Shetiya, Abolfazl AsudehImportance Weighting Bibliometric Analysis
University of Illinois Chicago · IIT Bombay
DRACO: Fine-Grained Credit Assignment with Dynamic Rubrics for Long-Horizon Agent Training
Sep 3, 2026Shubham Gandhi, Saurabh Goyal, Kiran Kate +1Agentic RL Step-Level Credit Assignment in RL
1Carnegie Mellon University · 2IBM Research
VICT: Verifier-Instrumented Credit Tracing for Long-Horizon LLM Agent Reinforcement Learning
Aug 28, 2026Pengcheng Li, Zhengyang Zhang, Dongxu Zhang +2Fine-Grained Credit Assignment RL for LLM Agents
Tsinghua University · Xi’an Jiaotong University · Jiaxing Nanhu University
Coverage, Not Credit: Failure-Credit Routing of Zeroth-Order Perturbation Budgets Does Not Improve On-Pool Sample Efficiency for LLM Agents
Aug 28, 2026Yuxu GeLLM Agent Training Fine-Grained Credit Assignment
University of York
How Much, Then Where: Credit-Conserving Action-to-Token Allocation for Multi-Turn Agent Reinforcement Learning
Aug 7, 2026Lichao Ma, Yang Sun, Shuaitao Zhao +9Token-Level Credit Assignment Credit Assignment in RL
1Peking University · 2Meituan LongCat Interaction Team · 3Fudan University +4
Shared Prefixes, Better Credit: Adaptive Routing for Multi-Agent Reasoning
Aug 3, 2026Yiqing Liu, Zihao Wang, Hantao Yao +2Adaptive Model Routing Multi-Agent Collaboration
University of Science and Technology of China
Element-Aware Group Learning for E-Commerce Image Generation
Aug 1, 2026Jingtong Chen, Jiahui Wang, Xue Zhao +2Group Relative Policy Optimization Image Generation
MARS-RA: Rank Aggregation for Credit Assignment via Multimodal Comparisons in Embodied Multi-Agent Cooperation
Jul 30, 2026Dawei Wang, Di Zhao, Xinyuan Liu +5Credit Assignment in RL Cooperative MARL
Newcastle University, United Kingdom · University of Auckland, New Zealand
Attribution Markets: A Fisher-Market Formulation for Fractional Credit Assignment Between Planned Tasks and Performed Actions
Jul 22, 2026Salavat IshbulatovMechanism Design Fine-Grained Credit Assignment
Independent researcher
MJ: Multi-turn LLM Jailbreaking via Decomposed Credit Assignment
Jul 13, 2026Junyoung Park, Namgyu Park, Sechan Lee +3RL for Language Models LLM Jailbreak Attacks
POSTECH GSAI · Samsung SDS · POSTECH GSAI & CSE
TRIAGE: Role-Typed Credit Assignment for Agentic Reinforcement Learning
Jun 30, 2026Yuanda Xu, Zhengze Zhou, Hejian Sang +6Agentic RL Step-Level Credit Assignment in RL
1LinkedIn Corporation · 2Harvard University · 3Johns Hopkins University +1
Towards Value-Constrained Credit Assignment in Fully Delegated AI Cooperatives
Jun 26, 2026Young Yoon, Jimin Kim, Soyeon ParkGradient-Based Attribution Value Alignment
Hongik University, Seoul, Republic of Korea
The Weakest Link Tells It All: Outcome-Supervised Process Reward Modeling via Learnable Credit Assignment
Jun 26, 2026Tianyu Jia, Yue Fang, Hongxin Ding +6Process Reward Models Fine-Grained Credit Assignment
National Engineering Research Center of Software Engineering, Peking University, Beijing, China · School of Computer Science, Peking University, Beijing, China · Key Laboratory of High Confidence Software Technologies, Ministry of Education, Beijing, China +3
Attributed, But Not Incremental: Cannibalization-Corrected Attribution for Large-Scale Advertising
Jun 25, 2026Donghui Li, Bowen Yuan, Zili Yang +2Causal Effect Estimation Online Advertising
TikTok · San Jose, CA, USA · Shanghai, China +1
Blame is easier than praise: Measuring off-ball defensive performance in football
Jun 18, 2026Jonas Bischofberger, Runqing Ma, Pascal Bauer +2Automated Evaluation Fine-Grained Credit Assignment
Centre for Sport Science and University Sports, University of Vienna, Vienna, Austria · Vienna Doctoral School of Pharmaceutical, Nutritional and Sport Sciences (VDS-PhaNuSpo), University of Vienna, Vienna, Austria · VfB Stuttgart 1893 AG, Stuttgart, Germany +2
PBSD: Privileged Bayesian Self-Distillation for Long-Horizon Credit Assignment
Jun 8, 2026Yang Tian, Rui Wang, Xumeng Wen +5Step-Level Credit Assignment in RL Reinforcement Learning with Verifiable Rewards
School of AI, Shanghai Jiao Tong University · 2XYZ AI Lab
ARCA: Adapter-Residual Credit Assignment When Token Signals Degenerate
May 29, 2026Rodney Lafuente-MercadoToken-Level Credit Assignment RL for Language Model Reasoning
Scale AI

Oct 26, 2026J/K move · Enter open · S save

SpatialPC: Deep Computation with Sublinear Local Credit Propagation

Benjamin Prada, Shion Matsumoto, Ankur Arjun Mali

Trustworthy Knowledge-Driven AI Lab Bellini College of Artificial Intelligence, Cybersecurity and Computing University of South Florida Tampa, FL 33620

Local Learning Fine-Grained Credit Assignment Communication Complexity Predictive Coding

Open paper AskPDF arXiv

On a chain of

N

states, any local learning rule with bounded communication range needs

Θ(N)

rounds before the loss can influence the earliest parameters, and this latency is usually attributed to depth. We show that it is a property of the communication geometry rather than of depth. For

N=m^d

states placed at constant density in

d

dimensions, a light-cone argument gives an

Ω(N^{1/d})

lower bound on the number of bounded-range rounds that any local algorithm needs before its gradient estimates can be sign-correct at every parameter. We construct a unit-range, degree-

2d

directed acyclic graph that contains a Hamiltonian computation of length

N-1

and in which every state lies within directed distance

d(m-1)=Θ(N^{1/d})

of the output, matching the lower bound. In the realizable-circuit model, depth and credit depth are therefore independent parameters: represented depth

Θ(N)

and causal credit latency

Θ(N^{1/d})

coexist, with a separation of order

N^{1-1/d}

. SpatialPC realizes this separation within predictive coding. Credit is propagated by adjoint prediction dynamics (APD), a local update whose

K

-round state on a settled acyclic graph is the truncated reverse-path sum

a^K=\sum_{k<K}(J^\top)^k b

, with error exactly

(J^\top)^K a^\star

. Graph geometry fixes when credit can first arrive at a state; the Jacobian-weighted path tail fixes how many further rounds accurate credit requires. A sparse two-dimensional Moore compiler retains the

N-1

backbone with

\tfrac32N-2

unit-range edges and has measured worst-node reachability

\sqrt N+2

up to

N=65{,}536

, where a 65,535-edge backbone is reached in 258 local hops. With

T=2\sqrt N

rounds, SpatialPC keeps finite-round task gradients aligned with settled gradients (cosine

0.93

on MNIST and

0.97

on Fashion-MNIST at

N=256

) and reaches

97.6\%

and

88.1\%

test accuracy after 20 epochs.