AstraFlow decouples RL components into autonomous dataflow services to natively support multi-policy agentic LLM training, elastic scaling, and cross-region execution with 2.7x speedup on math, code, search, and AgentBench workloads.
hub Canonical reference
SWE-RL: Advancing LLM Reasoning via Reinforcement Learning on Open Software Evolution
Canonical reference. 90% of citing Pith papers cite this work as background.
30
Pith papers citing it
Background
90% of classified citations
abstract
The recent DeepSeek-R1 release has demonstrated the immense potential of reinforcement learning (RL) in enhancing the general reasoning capabilities of large language models (LLMs). While DeepSeek-R1 and other follow-up work primarily focus on applying RL to competitive coding and math problems, this paper introduces SWE-RL, the first approach to scale RL-based LLM reasoning for real-world software engineering. Leveraging a lightweight rule-based reward (e.g., the similarity score between ground-truth and LLM-generated solutions), SWE-RL enables LLMs to autonomously recover a developer's reasoning processes and solutions by learning from extensive open-source software evolution data -- the record of a software's entire lifecycle, including its code snapshots, code changes, and events such as issues and pull requests. Trained on top of Llama 3, our resulting reasoning model, Llama3-SWE-RL-70B, achieves a 41.0% solve rate on SWE-bench Verified -- a human-verified collection of real-world GitHub issues. To our knowledge, this is the best performance reported for medium-sized (<100B) LLMs to date, even comparable to leading proprietary LLMs like GPT-4o. Surprisingly, despite performing RL solely on software evolution data, Llama3-SWE-RL has even emerged with generalized reasoning skills. For example, it shows improved results on five out-of-domain tasks, namely, function coding, library use, code reasoning, mathematics, and general language understanding, whereas a supervised-finetuning baseline even leads to performance degradation on average. Overall, SWE-RL opens up a new direction to improve the reasoning capabilities of LLMs through reinforcement learning on massive software engineering data.
hub tools
citation-role summary
background 9
dataset 1
citation-polarity summary
representative citing papers
10.7% of passing SWE-agent trajectories are Lucky Passes with chaotic behaviors, and a quality score based on process references changes model rankings across eight backends.
StepCodeReasoner aligns code reasoning with verifiable stepwise execution traces via print anchors and bi-level GRPO reinforcement learning, reaching SOTA results on CRUXEval (91.1%) and LiveCodeBench (86.5%) for a 7B model.
RewardHarness self-evolves a tool-and-skill library from 100 preference examples to reach 47.4% accuracy on image-edit evaluation, beating GPT-5, and yields stronger RL-tuned models.
An agentic LLM system discovers the XC functional SAFS26-a that improves on the ωB97M-V baseline by roughly 9% on a held-out thermochemistry dataset while warning that such systems can exploit unphysical shortcuts.
Faithful-Agent raises Trap SR in GUI agents from 13.88% to 80.21% via faithfulness-oriented SFT and GuAE-enhanced RFT with consistency rewards while retaining general performance.
ReCodeAgent uses a multi-agent system to translate and validate large code repositories across multiple programming languages, achieving 60.8% higher test pass rates than prior neuro-symbolic and agentic methods on 118 real-world projects.
SWE-EVO shows GPT-5.4 with OpenHands reaching only 25% success on complex multi-file evolution tasks versus 72.8% on SWE-Bench Verified, and introduces Fix Rate as a partial-progress metric.
TokenBuncher constrains response entropy via entropy-as-reward RL and a Token Noiser to stop harmful RL fine-tuning while keeping benign performance intact.
GiGPO adds a hierarchical grouping mechanism to group-based RL so that LLM agents receive both global trajectory and local step-level credit signals, yielding >12% gains on ALFWorld and >9% on WebShop over GRPO while keeping the same rollout and memory footprint.
Spreadsheet-RL applies RL fine-tuning and a custom Gym environment to raise LLM agent Pass@1 scores on spreadsheet benchmarks from roughly 8-12% to 17-23%.
DAgger-style training with turn-level policy interpolation raises 4B and 8B LLM agents to 27.3% and 29.8% on SWE-bench Verified, beating several larger published systems.
BoostAPR boosts automated program repair by training a sequence-level assessor and line-level credit allocator from execution outcomes, then applying them in PPO to reach 40.7% on SWE-bench Verified.
REC RL improves LLM code generation by automatically assessing and optimizing requirement difficulty with adaptive curriculum sampling, yielding 1.23-5.62% Pass@1 gains over baselines.
CodePivot uses Python as a pivot language plus an Aggressive-Partial-Functional RL reward to train a 7B model that outperforms much larger LLMs on multilingual code transpilation without parallel corpora.
AutoOR uses synthetic data generation and RL post-training with solver feedback to enable 8B LLMs to autoformalize linear, mixed-integer, and non-linear OR problems, matching larger models on benchmarks.
TensorHub uses Reference-Oriented Storage to enable scalable weight transfer in LLM RL training by referencing replicated GPU weights, achieving up to 19x reduction in cross-datacenter stall time.
ARuleCon uses AI agents plus execution-based checks to convert SIEM rules across vendors with 15% higher fidelity than standard LLM translation.
Systematic false positives in verifiers can cause RLVR training to reach suboptimal plateaus or collapse, with outcomes driven by error patterns rather than overall error rate.
Evo-Memory is a new streaming benchmark and evaluation framework for self-evolving memory in LLM agents, unifying over ten memory modules and introducing the ReMem pipeline for continual improvement on multi-turn and reasoning datasets.
TestPrune minimizes regression test suites to improve bug reproduction and patch validation in LLM-based agentic repair pipelines, delivering 6-13% relative gains on SWE-Bench benchmarks at low API cost.
BPO framework achieves state-of-the-art performance with improved token efficiency on ALFWorld, ScienceWorld, and WebShop by bootstrapping efficient reasoning, extrapolating via curriculum, and refining on reward-selected experiences.
HydroAgent fine-tunes Qwen3-4B on 2,576 expert calibration trajectories and applies Group-Relative Policy Optimization with NSE reward from live CREST simulations to improve hydrologic model calibration over frontier LLMs.
SpecRL uses the fraction of negative tests rejected by candidate specifications as a reward signal in RL training to produce stronger and more verifiable formal specifications than prior methods.
citing papers explorer
-
AstraFlow: Dataflow-Oriented Reinforcement Learning for Agentic LLMs
AstraFlow decouples RL components into autonomous dataflow services to natively support multi-policy agentic LLM training, elastic scaling, and cross-region execution with 2.7x speedup on math, code, search, and AgentBench workloads.
-
AgentLens: Revealing The Lucky Pass Problem in SWE-Agent Evaluation
10.7% of passing SWE-agent trajectories are Lucky Passes with chaotic behaviors, and a quality score based on process references changes model rankings across eight backends.
-
StepCodeReasoner: Aligning Code Reasoning with Stepwise Execution Traces via Reinforcement Learning
StepCodeReasoner aligns code reasoning with verifiable stepwise execution traces via print anchors and bi-level GRPO reinforcement learning, reaching SOTA results on CRUXEval (91.1%) and LiveCodeBench (86.5%) for a 7B model.
-
RewardHarness: Self-Evolving Agentic Post-Training
RewardHarness self-evolves a tool-and-skill library from 100 preference examples to reach 47.4% accuracy on image-edit evaluation, beating GPT-5, and yields stronger RL-tuned models.
-
Agentic Discovery of Exchange-Correlation Density Functionals
An agentic LLM system discovers the XC functional SAFS26-a that improves on the ωB97M-V baseline by roughly 9% on a held-out thermochemistry dataset while warning that such systems can exploit unphysical shortcuts.
-
Faithful Mobile GUI Agents with Guided Advantage Estimator
Faithful-Agent raises Trap SR in GUI agents from 13.88% to 80.21% via faithfulness-oriented SFT and GuAE-enhanced RFT with consistency rewards while retaining general performance.
-
ReCodeAgent: A Multi-Agent Workflow for Language-agnostic Translation and Validation of Large-scale Repositories
ReCodeAgent uses a multi-agent system to translate and validate large code repositories across multiple programming languages, achieving 60.8% higher test pass rates than prior neuro-symbolic and agentic methods on 118 real-world projects.
-
SWE-EVO: Benchmarking Coding Agents in Long-Horizon Software Evolution Scenarios
SWE-EVO shows GPT-5.4 with OpenHands reaching only 25% success on complex multi-file evolution tasks versus 72.8% on SWE-Bench Verified, and introduces Fix Rate as a partial-progress metric.
-
Token Buncher: Shielding LLMs from Harmful Reinforcement Learning Fine-Tuning
TokenBuncher constrains response entropy via entropy-as-reward RL and a Token Noiser to stop harmful RL fine-tuning while keeping benign performance intact.
-
Group-in-Group Policy Optimization for LLM Agent Training
GiGPO adds a hierarchical grouping mechanism to group-based RL so that LLM agents receive both global trajectory and local step-level credit signals, yielding >12% gains on ALFWorld and >9% on WebShop over GRPO while keeping the same rollout and memory footprint.
-
Spreadsheet-RL: Advancing Large Language Model Agents on Realistic Spreadsheet Tasks via Reinforcement Learning
Spreadsheet-RL applies RL fine-tuning and a custom Gym environment to raise LLM agent Pass@1 scores on spreadsheet benchmarks from roughly 8-12% to 17-23%.
-
Revisiting DAgger in the Era of LLM-Agents
DAgger-style training with turn-level policy interpolation raises 4B and 8B LLM agents to 27.3% and 29.8% on SWE-bench Verified, beating several larger published systems.
-
BoostAPR: Boosting Automated Program Repair via Execution-Grounded Reinforcement Learning with Dual Reward Models
BoostAPR boosts automated program repair by training a sequence-level assessor and line-level credit allocator from execution outcomes, then applying them in PPO to reach 40.7% on SWE-bench Verified.
-
Improving LLM Code Generation via Requirement-Aware Curriculum Reinforcement Learning
REC RL improves LLM code generation by automatically assessing and optimizing requirement difficulty with adaptive curriculum sampling, yielding 1.23-5.62% Pass@1 gains over baselines.
-
CodePivot: Bootstrapping Multilingual Transpilation in LLMs via Reinforcement Learning without Parallel Corpora
CodePivot uses Python as a pivot language plus an Aggressive-Partial-Functional RL reward to train a 7B model that outperforms much larger LLMs on multilingual code transpilation without parallel corpora.
-
AutoOR: Scalably Post-training LLMs to Autoformalize Operations Research Problems
AutoOR uses synthetic data generation and RL post-training with solver feedback to enable 8B LLMs to autoformalize linear, mixed-integer, and non-linear OR problems, matching larger models on benchmarks.
-
TensorHub: Scalable and Elastic Weight Transfer for LLM RL Training
TensorHub uses Reference-Oriented Storage to enable scalable weight transfer in LLM RL training by referencing replicated GPU weights, achieving up to 19x reduction in cross-datacenter stall time.
-
ARuleCon: Agentic Security Rule Conversion
ARuleCon uses AI agents plus execution-based checks to convert SIEM rules across vendors with 15% higher fidelity than standard LLM translation.
-
Delay, Plateau, or Collapse: Evaluating the Impact of Systematic Verification Error on RLVR
Systematic false positives in verifiers can cause RLVR training to reach suboptimal plateaus or collapse, with outcomes driven by error patterns rather than overall error rate.
-
Evo-Memory: Benchmarking LLM Agent Test-time Learning with Self-Evolving Memory
Evo-Memory is a new streaming benchmark and evaluation framework for self-evolving memory in LLM agents, unifying over ten memory modules and introducing the ReMem pipeline for continual improvement on multi-turn and reasoning datasets.
-
Can Old Tests Do New Tricks for Resolving SWE Issues?
TestPrune minimizes regression test suites to improve bug reproduction and patch validation in LLM-based agentic repair pipelines, delivering 6-13% relative gains on SWE-Bench benchmarks at low API cost.
-
Beyond Policy Optimization: A Data Curation Flywheel for Sparse-Reward Long-Horizon Planning
BPO framework achieves state-of-the-art performance with improved token efficiency on ALFWorld, ScienceWorld, and WebShop by bootstrapping efficient reasoning, extrapolating via curriculum, and refining on reward-selected experiences.
-
HydroAgent: Closing the Gap Between Frontier LLMs and Human Experts in Hydrologic Model Calibration via Simulator-Grounded RL
HydroAgent fine-tunes Qwen3-4B on 2,576 expert calibration trajectories and applies Group-Relative Policy Optimization with NSE reward from live CREST simulations to improve hydrologic model calibration over frontier LLMs.
-
Reinforcement Learning with Negative Tests as Completeness Signal for Formal Specification Synthesis
SpecRL uses the fraction of negative tests rejected by candidate specifications as a reward signal in RL training to produce stronger and more verifiable formal specifications than prior methods.
-
AceGRPO: Adaptive Curriculum Enhanced Group Relative Policy Optimization for Autonomous Machine Learning Engineering
AceGRPO trains 30B-parameter LLM agents to achieve 100% valid submissions and competitive performance on MLE-Bench-Lite through evolving data buffers and adaptive task sampling.
-
Webscale-RL: Automated Data Pipeline for Scaling RL Data to Pretraining Levels
Webscale-RL generates 1.2M verifiable QA pairs from pretraining corpora, enabling RL training that matches continual pretraining performance with up to 100x fewer tokens.
-
Towards Enabling An Artificial Self-Construction Software Life-cycle via Autopoietic Architectures
Proposes autopoietic architectures for self-constructing software as a fundamental shift in the SDLC, leveraging foundation models for autonomous evolution and maintenance.
-
Curr-RLCER:Curriculum Reinforcement Learning For Coherence Explainable Recommendation
Curr-RLCER applies curriculum reinforcement learning with coherence-driven rewards to align generated explanations with predicted ratings in explainable recommendation systems.
-
Agentic Reasoning for Large Language Models
The survey structures agentic reasoning for LLMs into foundational, self-evolving, and collective multi-agent layers while distinguishing in-context orchestration from post-training optimization and reviewing applications across domains.
-
A Survey of Scaling in Large Language Model Reasoning
A survey categorizing scaling in LLM reasoning across input size, steps, rounds, training, and future directions, noting that scaling can negatively affect performance.