ReLibra uses pre-known token-to-expert routing from RL rollouts to perform inter-batch expert reordering and intra-batch replication, delivering up to 1.6x higher throughput than Megatron-LM and 1.2x over oracle-equipped EPLB while staying within 6-10% of an ideal balanced baseline.
hub
Kimi k1.5: Scaling Reinforcement Learning with LLMs
97 Pith papers cite this work. Polarity classification is still indexing.
97
Pith papers citing it
abstract
Language model pretraining with next token prediction has proved effective for scaling compute but is limited to the amount of available training data. Scaling reinforcement learning (RL) unlocks a new axis for the continued improvement of artificial intelligence, with the promise that large language models (LLMs) can scale their training data by learning to explore with rewards. However, prior published work has not produced competitive results. In light of this, we report on the training practice of Kimi k1.5, our latest multi-modal LLM trained with RL, including its RL training techniques, multi-modal data recipes, and infrastructure optimization. Long context scaling and improved policy optimization methods are key ingredients of our approach, which establishes a simplistic, effective RL framework without relying on more complex techniques such as Monte Carlo tree search, value functions, and process reward models. Notably, our system achieves state-of-the-art reasoning performance across multiple benchmarks and modalities -- e.g., 77.5 on AIME, 96.2 on MATH 500, 94-th percentile on Codeforces, 74.9 on MathVista -- matching OpenAI's o1. Moreover, we present effective long2short methods that use long-CoT techniques to improve short-CoT models, yielding state-of-the-art short-CoT reasoning results -- e.g., 60.8 on AIME, 94.6 on MATH500, 47.3 on LiveCodeBench -- outperforming existing short-CoT models such as GPT-4o and Claude Sonnet 3.5 by a large margin (up to +550%).
hub tools
citation-role summary
background 1
citation-polarity summary
claims ledger
- abstract Language model pretraining with next token prediction has proved effective for scaling compute but is limited to the amount of available training data. Scaling reinforcement learning (RL) unlocks a new axis for the continued improvement of artificial intelligence, with the promise that large language models (LLMs) can scale their training data by learning to explore with rewards. However, prior published work has not produced competitive results. In light of this, we report on the training practice of Kimi k1.5, our latest multi-modal LLM trained with RL, including its RL training techniques,
co-cited works
roles
background 1polarities
background 1representative citing papers
RDPO applies magnitude-aware quantile normalization and Mahalanobis whitening to decorrelate heterogeneous rewards in multi-objective RL, improving instruction following and writing quality on LongCat-Flash post-training while staying competitive on reasoning and coding.
FinVQA is a new multilingual benchmark for Indic financial VQA with three difficulty levels and four formats, paired with the FIND framework for faithful numerical reasoning via fine-tuning and constrained decoding.
AutoLLMResearch trains agents via a multi-fidelity environment and MDP pipeline to extrapolate configuration principles from inexpensive to costly LLM experiments.
Unsupervised PRMs derived from LLM probabilities achieve up to 15% better error detection than LLM judges and match supervised PRMs in verification and RL tasks.
LEAD uses online adaptive mechanisms including Potential-Scaled Instability and symmetric efficiency rewards based on correct rollouts to achieve higher accuracy-efficiency scores with substantially shorter reasoning outputs than base models on math benchmarks.
RaPO reduces catastrophic forgetting in visual continual learning by shaping rewards around policy drift and stabilizing advantages with cross-task exponential moving averages during reinforcement fine-tuning of multimodal models.
BubbleSpec exploits long-tail bubbles in synchronous RL by using faster ranks' idle time to pre-generate rollout drafts for speculative decoding, reducing steps by 50% and raising throughput up to 1.8x while preserving exact synchrony.
vOPD stabilizes on-policy distillation gradients by subtracting a closed-form per-token negative reverse KL baseline as a detached control variate, preserving unbiasedness while lowering variance and matching expensive full-vocabulary methods.
The cumulative token IS ratio gives unbiased prefix correction and lower variance than full-sequence ratios for token-level gradients in LLM policy optimization, enabling CTPO to outperform GRPO and GSPO baselines on mathematical reasoning tasks.
Lighthouse Attention enables faster long-context pre-training via gradient-free symmetrical hierarchical compression of QKV while preserving causality, followed by a short full-attention recovery that yields lower loss than standard full-attention training.
Post-Reasoning boosts LLM accuracy by reversing the usual answer-after-reasoning order, delivering mean relative gains of 17.37% across 117 model-benchmark pairs with zero extra cost.
Reference-sampled weighted SFT with prompt-normalized Boltzmann weights induces the same policy as fixed-reference KL-regularized RLVR, with BOLT as the estimator and a finite one-shot error decomposition separating coverage, variance, and other terms.
UCPO modifies GRPO with a uniformity penalty over correct solutions to prevent diversity collapse in RLVR, yielding up to 10% higher Pass@64 on AIME24 and 45% more equation-level diversity.
Unstructured pruning augments test-time scaling reasoning performance in LLMs and can outperform the unpruned model on benchmarks, contrary to expectations from structured pruning studies.
SAS stabilizes efficient LLM reasoning by step-level advantage masking, improving Pass@1 accuracy by 0.86 points and cutting reasoning length by 16.3% versus length-aware baselines.
CoT-PoT ensembling achieves self-consistency accuracy in LLMs with only two samples for 78.6% of tasks, reducing computation by 9.3x compared to standard methods.
Language models achieve a perfect LSAT score, with experiments showing that internal thinking phases and a fine-tuned process reward model are key to high performance on logical reasoning questions.
SMTPO uses multi-task SFT to improve simulator feedback quality and RL with fine-grained rewards to optimize multi-turn preference reasoning in LLM-based conversational recommendation.
DeepEyes uses reinforcement learning to teach vision-language models active perception and image-based thinking, yielding gains on perception, reasoning, grounding, and hallucination benchmarks.
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.
Video-R1 uses temporal-aware RL and mixed datasets to boost video reasoning in MLLMs, with a 7B model reaching 37.1% on VSI-Bench and surpassing GPT-4o.
o1-like models overthink easy tasks; self-training reduces compute use without accuracy loss on GSM8K, MATH500, GPQA, and AIME.
TGPO improves on-policy LLM distillation by using teacher predictions conditioned on student rollouts to supply informative guidance when the two distributions diverge.
citing papers explorer
-
ReLibra: Routing-Replay-Guided Load Balancing for MoE Training in Reinforcement Learning
ReLibra uses pre-known token-to-expert routing from RL rollouts to perform inter-batch expert reordering and intra-batch replication, delivering up to 1.6x higher throughput than Megatron-LM and 1.2x over oracle-equipped EPLB while staying within 6-10% of an ideal balanced baseline.
-
Multi-Objective and Mixed-Reward Reinforcement Learning via Reward-Decorrelated Policy Optimization
RDPO applies magnitude-aware quantile normalization and Mahalanobis whitening to decorrelate heterogeneous rewards in multi-objective RL, improving instruction following and writing quality on LongCat-Flash post-training while staying competitive on reasoning and coding.
-
FIND: Toward Multimodal Financial Reasoning and Question Answering for Indic Languages
FinVQA is a new multilingual benchmark for Indic financial VQA with three difficulty levels and four formats, paired with the FIND framework for faithful numerical reasoning via fine-tuning and constrained decoding.
-
AutoLLMResearch: Training Research Agents for Automating LLM Experiment Configuration -- Learning from Cheap, Optimizing Expensive
AutoLLMResearch trains agents via a multi-fidelity environment and MDP pipeline to extrapolate configuration principles from inexpensive to costly LLM experiments.
-
Unsupervised Process Reward Models
Unsupervised PRMs derived from LLM probabilities achieve up to 15% better error detection than LLM judges and match supervised PRMs in verification and RL tasks.
-
LEAD: Length-Efficient Adaptive and Dynamic Reasoning for Large Language Models
LEAD uses online adaptive mechanisms including Potential-Scaled Instability and symmetric efficiency rewards based on correct rollouts to achieve higher accuracy-efficiency scores with substantially shorter reasoning outputs than base models on math benchmarks.
-
Overcoming Catastrophic Forgetting in Visual Continual Learning with Reinforcement Fine-Tuning
RaPO reduces catastrophic forgetting in visual continual learning by shaping rewards around policy drift and stabilizing advantages with cross-task exponential moving averages during reinforcement fine-tuning of multimodal models.
-
BubbleSpec: Turning Long-Tail Bubbles into Speculative Rollout Drafts for Synchronous Reinforcement Learning
BubbleSpec exploits long-tail bubbles in synchronous RL by using faster ranks' idle time to pre-generate rollout drafts for speculative decoding, reducing steps by 50% and raising throughput up to 1.8x while preserving exact synchrony.
-
KL for a KL: On-Policy Distillation with Control Variate Baseline
vOPD stabilizes on-policy distillation gradients by subtracting a closed-form per-token negative reverse KL baseline as a detached control variate, preserving unbiasedness while lowering variance and matching expensive full-vocabulary methods.
-
Rethinking Importance Sampling in LLM Policy Optimization: A Cumulative Token Perspective
The cumulative token IS ratio gives unbiased prefix correction and lower variance than full-sequence ratios for token-level gradients in LLM policy optimization, enabling CTPO to outperform GRPO and GSPO baselines on mathematical reasoning tasks.
-
Long Context Pre-Training with Lighthouse Attention
Lighthouse Attention enables faster long-context pre-training via gradient-free symmetrical hierarchical compression of QKV while preserving causality, followed by a short full-attention recovery that yields lower loss than standard full-attention training.
-
Post Reasoning: Improving the Performance of Non-Thinking Models at No Cost
Post-Reasoning boosts LLM accuracy by reversing the usual answer-after-reasoning order, delivering mean relative gains of 17.37% across 117 model-benchmark pairs with zero extra cost.
-
Reference-Sampled Boltzmann Projection for KL-Regularized RLVR: Target-Matched Weighted SFT, Finite One-Shot Gaps, and Policy Mirror Descent
Reference-sampled weighted SFT with prompt-normalized Boltzmann weights induces the same policy as fixed-reference KL-regularized RLVR, with BOLT as the estimator and a finite one-shot error decomposition separating coverage, variance, and other terms.
-
Uniform-Correct Policy Optimization: Breaking RLVR's Indifference to Diversity
UCPO modifies GRPO with a uniformity penalty over correct solutions to prevent diversity collapse in RLVR, yielding up to 10% higher Pass@64 on AIME24 and 45% more equation-level diversity.
-
Doing More With Less: Revisiting the Effectiveness of LLM Pruning for Test-Time Scaling
Unstructured pruning augments test-time scaling reasoning performance in LLMs and can outperform the unpruned model on benchmarks, contrary to expectations from structured pruning studies.
-
Stabilizing Efficient Reasoning with Step-Level Advantage Selection
SAS stabilizes efficient LLM reasoning by step-level advantage masking, improving Pass@1 accuracy by 0.86 points and cutting reasoning length by 16.3% versus length-aware baselines.
-
Self-Consistency from Only Two Samples: CoT-PoT Ensembling for Efficient LLM Reasoning
CoT-PoT ensembling achieves self-consistency accuracy in LLMs with only two samples for 78.6% of tasks, reducing computation by 9.3x compared to standard methods.
-
AI Achieves a Perfect LSAT Score
Language models achieve a perfect LSAT score, with experiments showing that internal thinking phases and a fine-tuned process reward model are key to high performance on logical reasoning questions.
-
User Simulator-Guided Multi-Turn Preference Optimization for Reasoning LLM-based Conversational Recommendation
SMTPO uses multi-task SFT to improve simulator feedback quality and RL with fine-grained rewards to optimize multi-turn preference reasoning in LLM-based conversational recommendation.
-
DeepEyes: Incentivizing "Thinking with Images" via Reinforcement Learning
DeepEyes uses reinforcement learning to teach vision-language models active perception and image-based thinking, yielding gains on perception, reasoning, grounding, and hallucination benchmarks.
-
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.
-
Video-R1: Reinforcing Video Reasoning in MLLMs
Video-R1 uses temporal-aware RL and mixed datasets to boost video reasoning in MLLMs, with a 7B model reaching 37.1% on VSI-Bench and surpassing GPT-4o.
-
Do NOT Think That Much for 2+3=? On the Overthinking of o1-Like LLMs
o1-like models overthink easy tasks; self-training reduces compute use without accuracy loss on GSM8K, MATH500, GPQA, and AIME.
-
Teacher-Guided Policy Optimization for LLM Distillation
TGPO improves on-policy LLM distillation by using teacher predictions conditioned on student rollouts to supply informative guidance when the two distributions diverge.
-
STOP: Structured On-Policy Pruning of Long-Form Reasoning in Low-Data Regimes
STOP uses structured on-policy analysis to prune long reasoning traces to their earliest correct node, cutting token usage 19-42% with little accuracy loss on math benchmarks.
-
Learning to See What You Need: Gaze Attention for Multimodal Large Language Models
Gaze Attention groups visual embeddings into selectable regions and dynamically restricts attention to task-relevant ones, matching dense baselines with up to 90% fewer visual KV entries via added context tokens.
-
Nice Fold or Hero Call: Learning Budget-Efficient Thinking for Adaptive Reasoning
BET reduces reasoning tokens by about 55% on average while improving performance across benchmarks by learning to short-solve easy queries, fold early on unsolvable ones, and preserve budget for hard solvable queries.
-
Anti-Self-Distillation for Reasoning RL via Pointwise Mutual Information
Anti-Self-Distillation reverses self-distillation signals via PMI to fix overconfidence on structural tokens, matching GRPO baseline accuracy 2-10x faster with up to 11.5 point gains across 4B-30B models.
-
Understanding and Preventing Entropy Collapse in RLVR with On-Policy Entropy Flow Optimization
OPEFO prevents entropy collapse in RLVR by rescaling token updates according to their entropy change contributions, yielding more stable optimization and better results on math benchmarks.
-
Drop the Act: Probe-Filtered RL for Faithful Chain-of-Thought Reasoning
ProFIL trains an activation probe on a frozen base model to zero advantages on theatrical post-commitment rollouts in GRPO, cutting theater 11-100%, raising faithful fractions, and shortening chains 4-19% without accuracy loss.
-
PruneTIR: Inference-Time Tool Call Pruning for Effective yet Efficient Tool-Integrated Reasoning
PruneTIR prunes erroneous tool-call trajectories during LLM inference via three trigger-based components to raise Pass@1 accuracy and efficiency while shortening context.
-
Beyond Thinking: Imagining in 360$^\circ$ for Humanoid Visual Search
Imagining in 360° decouples visual search into a single-step probabilistic semantic layout predictor and an actor, removing the need for multi-turn CoT reasoning and trajectory annotations while improving efficiency in 360° environments.
-
Forge: Quality-Aware Reinforcement Learning for NP-Hard Optimization in LLMs
OPT-BENCH trains LLMs on NP-hard optimization via quality-aware RLVR, achieving 93.1% success rate and 46.6% quality ratio on Qwen2.5-7B while outperforming GPT-4o and transferring gains to other domains.
-
MLS-Bench: A Holistic and Rigorous Assessment of AI Systems on Building Better AI
MLS-Bench shows that current AI agents fall short of reliably inventing generalizable ML methods, with engineering tuning easier than genuine invention.
-
Hint Tuning: Less Data Makes Better Reasoners
Hint Tuning uses an instruct model as a difficulty probe to create 1K multi-level hint examples that train reasoning models to calibrate chain-of-thought length, cutting tokens by 31.5% on average across 4B-32B models without accuracy loss.
-
AIPO: : Learning to Reason from Active Interaction
AIPO trains LLMs to expand their reasoning capability boundary via active multi-agent interaction with Verify, Knowledge, and Reasoning agents during RLVR, using importance sampling and clipping to handle feedback, then drops the agents at inference.
-
HTPO: Towards Exploration-Exploitation Balanced Policy Optimization via Hierarchical Token-level Objective Control
HTPO introduces hierarchical token-level objective control in RLVR to balance exploration and exploitation by grouping tokens according to difficulty, correctness, and entropy, yielding up to 8.6% gains on AIME benchmarks over DAPO.
-
Implicit Compression Regularization: Concise Reasoning via Internal Shorter Distributions in RL Post-Training
ICR creates a virtual shorter distribution from shortest correct on-policy responses to regularize RL post-training toward concise yet accurate reasoning, improving the accuracy-length Pareto frontier on math and knowledge benchmarks.
-
Schedule-and-Calibrate: Utility-Guided Multi-Task Reinforcement Learning for Code LLMs
ASTOR improves a single code LLM across four tasks by 9.0-9.5% over the best specialist and 7.5-12.8% over prior multi-task RL baselines via utility-driven data scheduling and adaptive KL regularization.
-
Piper: Efficient Large-Scale MoE Training via Resource Modeling and Pipelined Hybrid Parallelism
Piper introduces resource modeling and pipelined hybrid parallelism for MoE training, delivering 2-3.5X higher MFU than prior frameworks and 1.2-9X better all-to-all bandwidth.
-
T$^2$PO: Uncertainty-Guided Exploration Control for Stable Multi-Turn Agentic Reinforcement Learning
T²PO improves stability and performance in multi-turn agentic RL by using uncertainty dynamics at token and turn levels to guide exploration and avoid wasted rollouts.
-
MiniCPM-o 4.5: Towards Real-Time Full-Duplex Omni-Modal Interaction
MiniCPM-o 4.5 uses the Omni-Flow streaming framework to deliver real-time full-duplex omni-modal interaction with proactive behavior in a 9B model that approaches Gemini 2.5 Flash performance.
-
Length Value Model: Scalable Value Pretraining for Token-Level Length Modeling
LenVM models token-level remaining generation length as a bounded discounted value function derived from constant negative per-token rewards, providing a scalable proxy for generation horizon.
-
DORA: A Scalable Asynchronous Reinforcement Learning System for Language Model Training
DORA's multi-version streaming rollout enables 2-3x higher throughput in asynchronous RL for LLMs while preserving convergence by maintaining policy consistency, data integrity, and bounded staleness.
-
Why Does Reinforcement Learning Generalize? A Feature-Level Mechanistic Study of Post-Training in Large Language Models
RL generalizes better than SFT by preserving and slowly evolving a compact set of task-agnostic features from the base model rather than introducing many specialized ones.
-
ViPO: Visual Preference Optimization at Scale
Poly-DPO improves robustness to noisy preference data in visual models, and the new ViPO dataset enables superior performance, with the method reducing to standard DPO on high-quality data.
-
See Further, Think Deeper: Advancing VLM's Reasoning Ability with Low-level Visual Cues and Reflection
ForeSight lets VLMs use low-level visual cues and mask-based visual feedback within an RL loop to reason more accurately, with the 7B model beating same-scale peers and some closed-source SOTA on a new benchmark.
-
SFT-then-RL Outperforms Mixed-Policy Methods for LLM Reasoning
Correcting DeepSpeed optimizer and OpenRLHF loss bugs reveals SFT-then-RL outperforms mixed-policy methods by 3.8-22.2 points on math benchmarks.
-
SSL-R1: Self-Supervised Visual Reinforcement Post-Training for Multimodal Large Language Models
SSL-R1 reformulates visual SSL tasks into verifiable puzzles to supply rewards for RL post-training of MLLMs, yielding gains on multimodal benchmarks without external supervision.
-
WebGen-R1: Incentivizing Large Language Models to Generate Functional and Aesthetic Websites with Reinforcement Learning
WebGen-R1 uses end-to-end RL with scaffold-driven generation and cascaded rewards for structure, function, and aesthetics to transform a 7B model into a generator of deployable multi-page websites that rivals much larger models.