Datasets:

Purdy0228
/

rl-method-tracker

arxiv stringlengths 10 10	title stringlengths 9 175	month stringdate 2025-01-01 00:00:00 2026-06-01 00:00:00	label stringclasses 7 values	scope stringclasses 2 values	component stringclasses 12 values	label_confidence stringclasses 3 values	is_model_report bool 2 classes	general_method bool 2 classes	included bool 2 classes	methods listlengths 0 5	n_frameworks_ever int64 0 7	n_frameworks_at_snapshot int64 0 7
2501.00039	Speech Recognition With LLMs Adapted to Disordered Speech Using Reinforcement Learning	2025-01	rl_application	null	null	high	false	false	false	[]	0	0
2501.00226	Generative Emergent Communication: Large Language Model is a Collective World Model	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.00562	An Overview and Discussion on Using Large Language Models for Implementation Generation of Solutions to Open-Ended Problems	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.00581	Are the Values of LLMs Structurally Aligned with Humans? A Causal Perspective	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.00656	2 OLMo 2 Furious	2025-01	rl_application	null	null	high	true	false	false	[]	0	0
2501.01332	Decoding Knowledge in Large Language Models: A Framework for Categorization and Comprehension	2025-01	analysis	null	null	medium	false	false	false	[]	0	0
2501.01830	Auto-RT: Automatic Jailbreak Strategy Exploration for Red-Teaming Large Language Models	2025-01	rl_method	embedded	sampling	low	false	false	false	[]	0	0
2501.02189	A Survey of State of the Art Large Vision Language Models: Alignment, Benchmark, Evaluations and Challenges	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.02790	Segmenting Text and Learning Their Rewards for Improved RLHF in Language Model	2025-01	rl_method	general	credit_assignment	high	false	true	false	[]	0	0
2501.02997	CALM: Curiosity-Driven Auditing for Large Language Models	2025-01	rl_application	null	null	high	false	false	false	[]	0	0
2501.03262	REINFORCE++: Stabilizing Critic-Free Policy Optimization with Global Advantage Normalization	2025-01	rl_method	general	advantage	high	false	true	true	[ "reinforce_pp" ]	5	5
2501.03486	Align-Pro: A Principled Approach to Prompt Optimization for LLM Alignment	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.03884	AlphaPO: Reward Shape Matters for LLM Alignment	2025-01	preference_offline	null	null	high	false	false	false	[]	0	0
2501.04070	More is not always better? Enhancing Many-Shot In-Context Learning with Differentiated and Reweighting Objectives	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.04682	Towards System 2 Reasoning in LLMs: Learning How to Think With Meta Chain-of-Thought	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.04686	Unlocking Multimodal Mathematical Reasoning via Process Reward Model	2025-01	rl_method	embedded	credit_assignment	low	true	false	false	[]	0	0
2501.05057	LearningFlow: Automated Policy Learning Workflow for Urban Driving with Large Language Models	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.05366	Search-o1: Agentic Search-Enhanced Large Reasoning Models	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.05790	Understanding Impact of Human Feedback via Influence Functions	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.06248	Utility-inspired Reward Transformations Improve Reinforcement Learning Training of Language Models	2025-01	rl_method	general	multi_reward	high	false	true	false	[]	0	0
2501.06252	Transformer-Squared: Self-adaptive LLMs	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.06423	AlgoPilot: Fully Autonomous Program Synthesis Without Human-Written Programs	2025-01	other	null	null	low	false	false	false	[]	0	0
2501.06655	Personalized Preference Fine-tuning of Diffusion Models	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.06695	DVM: Towards Controllable LLM Agents in Social Deduction Games	2025-01	rl_application	null	null	low	false	false	false	[]	0	0
2501.06842	SPAM: Spike-Aware Adam with Momentum Reset for Stable LLM Training	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.06911	Risk-Averse Finetuning of Large Language Models	2025-01	rl_method	embedded	other	high	false	false	false	[]	0	0
2501.06980	Combining LLM decision and RL action selection to improve RL policy for adaptive interventions	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.07324	Foundation Models at Work: Fine-Tuning for Fairness in Algorithmic Hiring	2025-01	rl_application	null	null	high	false	false	false	[]	0	0
2501.07886	Iterative Label Refinement Matters More than Preference Optimization under Weak Supervision	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.08071	CuAsmRL: Optimizing GPU SASS Schedules via Deep Reinforcement Learning	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.08156	Are DeepSeek R1 And Other Reasoning Models More Faithful?	2025-01	analysis	null	null	high	false	false	false	[]	0	0
2501.08598	LlamaRestTest: Effective REST API Testing with Small Language Models	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.08600	AutoRestTest: A Tool for Automated REST API Testing Using LLMs and MARL	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.08925	Disentangling Exploration of Large Language Models by Optimal Exploitation	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.09254	Clone-Robust AI Alignment	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.09620	Beyond Reward Hacking: Causal Rewards for Large Language Model Alignment	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.09685	Inference-Time Alignment in Diffusion Models with Reward-Guided Generation: Tutorial and Review	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.09686	Towards Large Reasoning Models: A Survey of Reinforced Reasoning with Large Language Models	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.09766	iTool: Reinforced Fine-Tuning with Dynamic Deficiency Calibration for Advanced Tool Use	2025-01	preference_offline	null	null	high	false	false	false	[]	0	0
2501.10120	PaSa: An LLM Agent for Comprehensive Academic Paper Search	2025-01	rl_application	null	null	high	true	false	false	[]	0	0
2501.11006	GREEN-CODE: Learning to Optimize Energy Efficiency in LLM-based Code Generation	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.11110	Chain-of-Reasoning: Towards Unified Mathematical Reasoning in Large Language Models via a Multi-Paradigm Perspective	2025-01	other	null	null	high	true	false	false	[]	0	0
2501.11223	Reasoning Language Models: A Blueprint	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.11284	RedStar: Does Scaling Long-CoT Data Unlock Better Slow-Reasoning Systems?	2025-01	other	null	null	high	true	false	false	[]	0	0
2501.11417	Neural Contextual Reinforcement Framework for Logical Structure Language Generation	2025-01	rl_application	null	null	low	false	false	false	[]	0	0
2501.11463	Curiosity-Driven Reinforcement Learning from Human Feedback	2025-01	rl_method	general	other	high	false	true	false	[]	0	0
2501.11621	Trojan Detection Through Pattern Recognition for Large Language Models	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.11651	T1: Advancing Language Model Reasoning through Reinforcement Learning and Inference Scaling	2025-01	rl_method	embedded	sampling	high	true	false	false	[]	0	0
2501.12051	MedS^3: Towards Medical Slow Thinking with Self-Evolved Soft Dual-sided Process Supervision	2025-01	rl_method	embedded	credit_assignment	high	false	false	false	[]	0	0
2501.12368	InternLM-XComposer2.5-Reward: A Simple Yet Effective Multi-Modal Reward Model	2025-01	other	null	null	medium	true	false	false	[]	0	0
2501.12539	Compositional Instruction Following with Language Models and Reinforcement Learning	2025-01	other	null	null	low	false	false	false	[]	0	0
2501.12570	O1-Pruner: Length-Harmonizing Fine-Tuning for O1-Like Reasoning Pruning	2025-01	rl_method	general	length	high	false	true	false	[]	0	0
2501.12599	Kimi k1.5: Scaling Reinforcement Learning with LLMs	2025-01	rl_method	embedded	other	low	true	false	false	[]	0	0
2501.12698	Training Dialogue Systems by AI Feedback for Improving Overall Dialogue Impression	2025-01	rl_application	null	null	high	false	false	false	[]	0	0
2501.12735	Online Preference Alignment for Language Models via Count-based Exploration	2025-01	rl_method	general	other	high	false	true	false	[]	0	0
2501.12948	DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning	2025-01	rl_application	null	null	high	true	false	false	[]	0	0
2501.13011	MONA: Myopic Optimization with Non-myopic Approval Can Mitigate Multi-step Reward Hacking	2025-01	rl_method	general	other	high	false	true	false	[]	0	0
2501.13264	OpenGenAlign: A Preference Dataset and Benchmark for Trustworthy Reward Modeling in Open-Ended, Long-Context Generation	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.13333	AgentRec: Agent Recommendation Using Sentence Embeddings Aligned to Human Feedback	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.13896	GUI-Bee: Align GUI Action Grounding to Novel Environments via Autonomous Exploration	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.13927	CRPO: Confidence-Reward Driven Preference Optimization for Machine Translation	2025-01	preference_offline	null	null	high	false	false	false	[]	0	0
2501.14225	Multi-agent KTO: Reinforcing Strategic Interactions of Large Language Model in Language Game	2025-01	preference_offline	null	null	high	false	false	false	[]	0	0
2501.15228	Improving Retrieval-Augmented Generation through Multi-Agent Reinforcement Learning	2025-01	rl_application	null	null	high	false	false	false	[]	0	0
2501.15304	Music Generation using Human-In-The-Loop Reinforcement Learning	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.15453	Data-adaptive Safety Rules for Training Reward Models	2025-01	other	null	null	high	true	false	false	[]	0	0
2501.15942	TimeHF: Billion-Scale Time Series Models Guided by Human Feedback	2025-01	other	null	null	high	true	false	false	[]	0	0
2501.15971	REINFORCE-ING Chemical Language Models for Drug Discovery	2025-01	other	null	null	low	false	false	false	[]	0	0
2501.16215	Smarter Together: Combining Large Language Models and Small Models for Physiological Signals Visual Inspection	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.16664	Improving Vision-Language-Action Model with Online Reinforcement Learning	2025-01	rl_method	general	other	low	false	true	false	[]	0	0
2501.16727	xJailbreak: Representation Space Guided Reinforcement Learning for Interpretable LLM Jailbreaking	2025-01	rl_application	null	null	high	false	false	false	[]	0	0
2501.17030	Challenges in Ensuring AI Safety in DeepSeek-R1 Models: The Shortcomings of Reinforcement Learning Strategies	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.17104	COS(M+O)S: Curiosity and RL-Enhanced MCTS for Exploring Story Space via Language Models	2025-01	preference_offline	null	null	high	false	false	false	[]	0	0
2501.17112	Decoding Human Preferences in Alignment: An Improved Approach to Inverse Constitutional AI	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.17148	AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoders	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.17206	Integrating Reinforcement Learning and AI Agents for Adaptive Robotic Interaction and Assistance in Dementia Care	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.17974	Think Smarter not Harder: Adaptive Reasoning with Inference Aware Optimization	2025-01	rl_method	general	length	high	false	true	false	[]	0	0
2501.18101	Diverse Preference Optimization	2025-01	preference_offline	null	null	high	false	false	false	[]	0	0
2501.18663	Joint Optimization of Prompt Security and System Performance in Edge-Cloud LLM Systems	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.18668	Simulation Streams: A Programming Paradigm for Controlling Large Language Models and Building Complex Systems with Generative AI	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.18858	BRiTE: Bootstrapping Reinforced Thinking Process to Enhance Language Model Reasoning	2025-01	rl_method	general	other	low	false	true	false	[]	0	0
2501.18880	RLS3: RL-Based Synthetic Sample Selection to Enhance Spatial Reasoning in Vision-Language Models for Indoor Autonomous Perception	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.19153	Test-Time Training Scaling Laws for Chemical Exploration in Drug Design	2025-01	other	null	null	high	false	false	false	[]	0	0
2501.19266	Jackpot! Alignment as a Maximal Lottery	2025-01	rl_method	general	other	medium	false	true	false	[]	0	0
2501.19358	The Energy Loss Phenomenon in RLHF: A New Perspective on Mitigating Reward Hacking	2025-01	rl_method	general	entropy	high	false	true	false	[]	0	0
2501.19400	Vintix: Action Model via In-Context Reinforcement Learning	2025-01	other	null	null	high	true	false	false	[]	0	0
2502.00212	STP: Self-play LLM Theorem Provers with Iterative Conjecturing and Proving	2025-02	other	null	null	high	false	false	false	[]	0	0
2502.00409	Doing More with Less: A Survey on Routing Strategies for Resource Optimisation in Large Language Model-Based Systems	2025-02	other	null	null	high	false	false	false	[]	0	0
2502.00640	CollabLLM: From Passive Responders to Active Collaborators	2025-02	rl_method	general	credit_assignment	high	false	true	true	[ "collabllm" ]	1	1
2502.00657	LLM Safety Alignment is Divergence Estimation in Disguise	2025-02	preference_offline	null	null	low	false	false	false	[]	0	0
2502.00666	Avoiding exp(R_{max)} scaling in RLHF through Preference-based Exploration	2025-02	rl_method	general	other	low	false	true	false	[]	0	0
2502.00691	To Code or not to Code? Adaptive Tool Integration for Math Language Models via Expectation-Maximization	2025-02	rl_method	general	offpolicy	high	false	true	false	[]	0	0
2502.00698	MM-IQ: Benchmarking Human-Like Abstraction and Reasoning in Multimodal Models	2025-02	rl_application	null	null	high	true	false	false	[]	0	0
2502.00792	RTBAgent: A LLM-based Agent System for Real-Time Bidding	2025-02	other	null	null	high	false	false	false	[]	0	0
2502.00814	Disentangling Length Bias In Preference Learning Via Response-Conditioned Modeling	2025-02	preference_offline	null	null	high	false	false	false	[]	0	0
2502.01203	KL-Regularized RLHF with Multiple Reference Models: Exact Solutions and Sample Complexity	2025-02	analysis	null	null	high	false	false	false	[]	0	0
2502.01208	On Almost Surely Safe Alignment of Large Language Models at Inference-Time	2025-02	other	null	null	high	false	false	false	[]	0	0
2502.01236	Eliciting Language Model Behaviors with Investigator Agents	2025-02	preference_offline	null	null	high	false	false	false	[]	0	0
2502.01237	The Differences Between Direct Alignment Algorithms are a Blur	2025-02	preference_offline	null	null	high	false	false	false	[]	0	0
2502.01387	TeLL-Drive: Enhancing Autonomous Driving with Teacher LLM-Guided Deep Reinforcement Learning	2025-02	other	null	null	high	false	false	false	[]	0	0
2502.01456	Process Reinforcement through Implicit Rewards	2025-02	rl_method	general	credit_assignment	high	false	true	true	[ "prime" ]	1	1

End of preview. Expand in Data Studio

RL Method Tracker

Which RL post-training methods for language models reached seven open-source training frameworks (verl, TRL, OpenRLHF, NeMo-RL, ROLL, slime, AReaL), when they landed, and how each framework offers them. Of 1,598 arXiv papers from January 2025 to June 2026 that propose a general RL training method, 37 reached the main branch of at least one framework.

Paper: From RL Papers to Framework Code: Inclusion and Review in Seven Open-Source Training Frameworks
Project page with search: https://pth2002.github.io/rl-method-tracker/
Code and the full data release: https://github.com/pth2002/rl-method-tracker
Archived release: doi:10.5281/zenodo.23495744

This is the frozen v1.0 data behind the report, with framework histories up to the 2026-10-18 snapshot. The GitHub repository refreshes the framework histories every week.

Load

from datasets import load_dataset

papers = load_dataset("Purdy0228/rl-method-tracker", "papers", split="train")
methods = load_dataset("Purdy0228/rl-method-tracker", "methods", split="train")

Subsets

Subset	Rows	One row per
`papers`	6,939	screened arXiv paper on RL for language models
`methods`	65	method in the catalog, with its first landing in each framework
`landings`	101	first commit that put a method on a framework's main branch
`proposals`	48	public proposal for a general method that no framework included

papers

arxiv, title, month: arXiv id, title and month of the first version.
label: what the paper contributes. rl_method, rl_application, analysis, systems, distillation, preference_offline or other.
scope: for RL methods, general when the method is the paper's main contribution and embedded when it sits inside a paper about a task, domain, agent or application.
component: the part of training the method changes, such as advantage, clipping, sampling or entropy.
label_confidence: high, medium or low.
is_model_report: the paper is the technical report of a model.
general_method: the paper counts toward the 1,598 general methods.
included: the paper's method reached the main branch of at least one framework.
methods: catalog ids of the included methods the paper introduces.
n_frameworks_ever, n_frameworks_at_snapshot: frameworks that ever included the method, and those that still include it at the snapshot.

methods

id, name, family, arxiv, source_date: the method and its first public source.
n_fw_ever, n_fw_head, n_default: frameworks that ever included it, that include it at the snapshot, and that turn it on by default.
first_adoption, first_lag_days, first_lag_is_upper_bound: date of the first landing in any framework and the days from the source to it. The lag is an upper bound when that first landing arrived with a repository's initial commit or a bulk release.
n_self_contributed: frameworks whose landing commit came from an author of the paper.
<framework>_first: date of the first landing in that framework.
<framework>_head: tier at the snapshot. default is on in a run that keeps every setting. conditional_default is on once a related feature is enabled. core is a selectable algorithm in the main code path. option is a switch that modifies the main algorithm. experimental and recipe sit outside the main path.

landings

method, framework, date, pr, tier: the landing and the pull request that made it.
first_in_any: this framework was the first to include the method.
form: how the method changes training, as coefficient, term or pipeline.
commit_author_relation: whether the landing commit came from an author of the paper.
prior_request: an issue or pull request asked for the method before it landed.
pr_author, merged_by, author_prior_merger, self_merged, human_reviewers: who proposed and merged the pull request. A prior merger had merged a pull request in the framework before.
days_open, core_lines, files, lines, tests, docs, deps, new_files: size and contents of the change.

proposals

arxiv, method, framework, type, number: the proposal, as a pull request (pr) or a request in an issue or discussion (request).
outcome: open_no_reply, open_engaged, declined, redirected, stale_cleanup, closed_inactive, closed_no_comment or withdrawn.
reasons: concerns the maintainers stated, such as demand, surface, placement or evidence.
quote: one short quote from the thread as evidence for the code.

Thread texts stay on GitHub. The labelling rules and coding guides are in the repository.

Citation

@misc{pan2026rlframeworks,
  title         = {From {RL} Papers to Framework Code: Inclusion and Review in Seven Open-Source Training Frameworks},
  author        = {Pan, Taiheng},
  year          = {2026},
  eprint        = {2610.28027},
  archivePrefix = {arXiv},
  primaryClass  = {cs.LG},
  doi           = {10.48550/arXiv.2610.28027},
  url           = {https://arxiv.org/abs/2610.28027}
}

License

CC BY 4.0. arXiv titles come from arXiv metadata.

Downloads last month: 29

Paper for Purdy0228/rl-method-tracker

From RL Papers to Framework Code: Inclusion and Review in Seven Open-Source Training Frameworks

Paper • 2610.28027 • Published 4 days ago • 12