Contrastive RL

RL: Reinforcement Learning

Follow this topic Ask about this topic

Momentum

1 paper in the last four weeks, against 2 the four weeks before. 0.0% of all new papers.

Jul 13Week of Sep 28

Latest papers 18

Oct 4, 2026cs.LG

Direction-Conditioned Policies for Online Goal-Conditioned Reinforcement Learning

Contrastive Reinforcement Learning (CRL) learns representations that estimate goal reachability, yet its policy remains conditioned on raw goals and therefore does not directly exploit the geometry encoded by its critic. We introduce Direction-Conditioned Policies (DCP), a method built around a small modification to CRL: DCP selects previously visited states as waypoints during online training and conditions the policy on their direction and distance in representation space. At deployment, DCP applies the same interface directly to the final goal, requiring neither waypoint selection nor planning. Across nine navigation and manipulation tasks, DCP attains higher final success rates than CRL on seven tasks and spends more time near the goal on seven. Controlled maze experiments further show that DCP captures shortest-path geometry more accurately and that the supplied direction causally influences the actor's behavior. We identify waypoint coverage and ranking as limits to exploration, and show that learned candidate generation improves goal reaching in two controlled mazes.

S K Swaminathan, Damiya Gondha, Theyanesh Eswaramoorthy Rajahkrishnan +1

Department of Computer Science and Engineering Indian Institute of Technology Kharagpur, India · Department of Mechanical Engineering Indian Institute of Technology Kharagpur, India

Reinforcement Learning Contrastive RL Goal-Conditioned RL

Sep 2, 2026cs.CV

Learning to Zoom Efficiently with a Contrastive Curriculum

Using a zoom-in tool is an important foundational part of modern visual agents, because it allows to efficiently handle tasks involving high-resolution images. Most previous methods need an extensive warm-start supervised fine-tuning phase for teaching models zoom-in. We show that this is not necessary by proposing a new intrinsic reward for learning tool use in MLLMs without the need for additional labels or warm-start SFT. Our InfoNCE-style reward uses a curriculum of increasingly hard negative tool calls as a contrastive training signal. Empirical experiments on

V^*

, HRBench and MME-RealWorld show that our approach is competitive while being more efficient. When used as a drop-in replacement for SFT, we even outperform all baselines. To directly measure the zoom-in ability of models, we further introduce the scalable synthetic Muffin&Chihuahua (M&C) dataset. Each image consists of a grid with every cell either showing a muffin or chihuahua. Leveraging the M&C dataset's unique region of interest labels, we find that recall is the metric that most strongly correlates the zoom-in region with final task performance. Our model and code for reproduction is publicly available under https://github.com/UKPLab/emnlp2026-zoom-in

Falko Helm, Iryna Gurevych

Ubiquitous Knowledge Processing Lab (UKP Lab) Department of Computer Science and Hessian Center for AI (hessian.AI) Technical University of Darmstadt

Tool Use in VLMs Contrastive RL RL for Tool Use

Aug 31, 2026cs.LG

Three Steps at a Time: Learning Representations from Action Sequences in Contrastive RL

While self-supervised approaches to reinforcement learning have achieved strong results by learning representations of states and actions, a key open question is the time scale over which actions should be modeled. Departing from the standard formulation relying on single-step actions, we extend contrastive reinforcement learning (CRL), a prototypical self-supervised method, to operate over action chunks, and find that this results in large, pervasive gains across established offline and online benchmarks: +31.7% and +93.1% across 18 and 11 environments respectively. While action-chunking-driven gains are generally explained through the ability to model non-Markovian, temporally extended policies, and to propagate unbiased multi-step returns, interestingly, we find that these arguments only partially apply to CRL. Our empirical studies suggest that, in the context of CRL, an action chunk carries more information about the goal than a single action, measurably improving the critic's representations, and rendering the algorithm significantly more effective.

Michal Korniak, Kamil Dybek, Benjamin Eysenbach +2

ETH Zurich · University of Warsaw · Princeton University +3

Reinforcement Learning Offline RL Representation Learning for RL Contrastive RL

Jul 30, 2026cs.LG

Contrastive Reinforced Policy Optimization via Privileged Self-Distillation

Recent advances in post-training Large Language Models (LLMs) increasingly rely on Reinforcement Learning with Verifiable Rewards (RLVR) or On-Policy Self-Distillation (OPSD). While OPSD provides dense, logit-level supervision, it inherently suffers from exposure bias due to the privileged information of the self-teacher. In multi-turn agentic settings, this leads to reasoning route convergence and the loss of clear optimization directions. To tackle these challenges, we introduce Contrastive Reinforced Policy Optimization (CRPO), which reformulates agentic OPSD from a contrastive learning perspective. By leveraging predictive entropy to distinguish between positive positions (reflective exploration) and negative positions (exposure bias), CRPO conducts group-wise contrast to preserve reliable, fine-grained optimization signals. Extensive evaluations across 13 challenging reasoning and deep-search benchmarks demonstrate that CRPO consistently outperforms existing reinforcement learning and self-distillation baselines, significantly enhancing training stability and generalization in long-horizon interactions.

Xingjian Wu, Junlin Liu, Xingchen Liu +6

Meituan

Agentic RL RL for Language Model Reasoning On-Policy Distillation RL Post-Training Contrastive RL

Jul 29, 2026cs.LG

Good Rankers, Bad Objectives: Bilinear Contrastive Critics under Expressive Policy Search

Good action rankings do not make a contrastive critic safe to maximize. These critics increasingly act as value-like objectives for best-of-

K

selection, planning, and critic-guided generation. Unbounded bilinear scores can let large embedding norms inflate off-support values, but cosine bounding does not remove the failure. A controlled support decomposition attributes most raw bilinear regret to norm drift. Cosine and hybrid critics nevertheless select off-support actions from most pools and incur comparable regret. Contrastive scores are weakly calibrated or inverted in the top score decile across four OGBench navigation tasks, and they fail to order fixed-query actions by value. Bellman-trained TD-Q succeeds, including in a parameter-matched function-class control. Realized costs depend on the task: simulator rollouts reveal single-step selection costs on PointMaze and the exact-

Q^*

toy but well-powered nulls on AntMaze and HumanoidMaze, where the controller can self-correct. A training/readout decomposition traces the lost ordering to the cosine training objective; raw-trained embeddings retain weak ordering after inference-time normalization. Candidate maximization can therefore exploit false positives caused by norm drift, score saturation, or in-support misranking. Contrastive critics remain useful compatibility rankers on navigation and manipulation tasks, but action selection requires a value-calibrated scalar.

Ayushman Singh, Siddharth Aphale

Stanford University

Learning to Rank Contrastive RL

Jul 20, 2026cs.LG

RRPO: Reference-Relative Policy Optimization with Stratified Conditional Rollouts

Group Relative Policy Optimization (GRPO) has shown strong effectiveness in reinforcement learning from verifiable feedback, where sampled rollouts can be compared within a group using task-provided correctness signals. However, extending group-relative optimization beyond verifiable settings is challenging because success in many tasks is not captured by a single correctness criterion. We propose \textbf{Reference-Relative Policy Optimization (RRPO)}, which generalizes GRPO by replacing direct correctness-based advantage construction with reference-relative contrastive comparisons. RRPO first uses \emph{stratified conditional rollouts} to construct positive and negative anchor sets, and then trains a metric projection head with a set-contrastive objective to compare candidate rollouts against these anchors. The resulting alignment scores directly define contrastive advantages: during policy optimization, the projection head is frozen, and the scores are centered within each rollout group in a standard group-relative objective. We evaluate RRPO using anchor-based contrastive advantages throughout policy optimization, without relying on task ground-truth verifiers. Across verifiable reasoning, open-ended generation, and post-SFT settings, RRPO remains competitive with verifier-based optimization, improves over weakly supervised baselines, and provides additional gains after supervised fine-tuning.

Yuxin Xiong, Xunyi Jiang, Rohan Surana +8

University of California San Diego · 2Adobe Research

Reinforcement Learning RL for Language Model Reasoning Group Relative Policy Optimization RL Post-Training Contrastive RL

Jun 23, 2026cs.LG

GCT-MARL: Graph-Based Contrastive Transfer for Sample-Efficient Cooperative Multi-Agent Reinforcement Learning

In cooperative multi-agent reinforcement learning (MARL), from a deployment perspective, it is challenging and expensive to train agents from scratch for each new environment or task. In this work, we propose GCT-MARL, a transfer learning framework that builds on the multi-view graph contrastive backbone of MAIL and augments it with a per-view, adaptively weighted alignment loss and a two-phase training protocol specifically designed for transfer across populations of varying sizes and compositions. We empirically demonstrate that the proposed framework markedly accelerates convergence on the target task relative to from-scratch training, in both homogeneous (within-faction, varying N) and heterogeneous (cross-faction and mixed unit-type) transfer scenarios. Furthermore, we show that the framework naturally supports continual learning by sequentially chaining the two-phase transfer protocol across a series of related tasks. Overall, this work provides a unified approach to mitigating key limitations in current MARL transfer methods with new insights at both methodological and empirical levels.

Animesh Animesh, Satheesh K Perepu, Kaushik Dey

Department of Artificial Intelligence, Indian Institute of Technology Kharagpur, India · Work done during research internship at Ericsson Research, Bangalore, India · Ericsson Research, Bangalore, India

Graph Contrastive Learning Contrastive RL Cooperative MARL Multi-Agent Reinforcement Learning

Jun 15, 2026cs.CL

Context-Aware RL for Agentic and Multimodal LLMs

Large language models (LLMs) often fail when answering requires identifying a small but decisive piece of evidence within a long or complex context, such as a single line in a tool trace or a subtle detail in an image. We propose ContextRL, a context-aware reinforcement learning (RL) method that improves long-horizon reasoning and multimodal performance through an \emph{indirect} auxiliary objective. Instead of supervising only the final answer, ContextRL presents the model with a query, an answer, and two highly similar contexts, and rewards it for selecting the context that supports the query--answer pair, thereby encouraging fine-grained grounding. We construct contrastive context data in two domains: for coding agents, trajectories serve as contexts, yielding 1k pairs built via condition filtering; for multimodal reasoning, images serve as contexts, yielding 7K pairs built via generative editing and similarity search. ContextRL achieves average gains of +2.2% over standard GRPO on 5 long-horizon benchmarks, and +1.8% across 12 diverse visual question answering benchmarks. To disentangle the effect of the proposed objective from that of additional data, we compare against data-augmentation baselines that repurpose the same contrastive contexts as standard query--context--answer examples. These baselines provide little to no improvement, showing that the gains arise from the proposed context-selection objective rather than from the contrastive data alone.

Peiyang Xu, Bangzheng Li, Sijia Liu +4

1Princeton University · 2UC Davis

Multimodal Grounding Contrastive RL RL for VLMs RL for LLM Agents

Jun 15, 2026cs.LG

Direction-Conditioned Policies via Compositional Subgoal Scoring for Online Goal-Conditioned Reinforcement Learning

Hamilton-Jacobi-Bellman theory implies that the optimal goal-conditioned action depends on the goal only through the gradient of the goal-reaching distance at the current state, yet standard online GCRL still conditions the actor on the raw goal -- a signal that is geometrically uninformative when the goal is far from the data distribution. We propose Direction-Conditioned Policies (DCP), a fully online method that decomposes goal-reaching into two components sharing one InfoNCE representation

ψ

: a subgoal-scoring step that selects a visited state

z_t

aligned with the final goal

g

in

ψ_g

, and a direction-conditioned actor that consumes the unit direction

d_t

and magnitude

r_t

from

ψ(s_t)

to

ψ(z_t)

. The two components train jointly, factor cleanly at deployment (subgoal scoring is removed, while direction conditioning remains with

g

in place of

z_t

), and admit independent modification at the same

(d_t,r_t)

interface. We prove three results. First, direction sufficiency under HJB: the optimal action under control-affine dynamics depends on the goal only through the value gradient. Second, a quantitative bound showing that, under mild conditions on the learned representation and assuming the scoring rule returns an on-path

z_t

, the actor's conditioning input at training and at deployment coincide up to representation error and geodesic slack. Third, a controllable-subspace characterization of when directional conditioning fails. Across nine environments, DCP improves over Contrastive RL on most final metrics, with the largest gains on manipulation and obstacle-interaction tasks; a qualitative analysis of the learned

ψ

-distance landscape shows the contrastive representation behaves as an online quasimetric encoding environment topology, and the single failure case (AntSoccer) localizes to a learned-gradient pathology that the theory anticipates.

Swaminathan S K, Damiya Gondha, Theyanesh Eswaramoorthy Rajahkrishnan +1

Department of Computer Science and Engineering, Indian Institute of Technology Kharagpur, Kharagpur, India · Department of Mechanical Engineering, Indian Institute of Technology Kharagpur, Kharagpur, India.

Contrastive RL Policy Learning Goal-Conditioned RL

Jun 10, 2026cs.RO

Learning Object Manipulation from Scratch via Contrastive Interaction

Contrastive Reinforcement Learning (CRL) has seen recent success in a wide variety of goal-conditioned robotics tasks by learning structured representations of the dynamics. However, despite its success in locomotion and simpler control domains, CRL often struggles in interaction-rich manipulation. We argue that a key source of this difficulty is object-centric interaction, such as contact or grasping, that induces distinct changes in the underlying dynamic modes. In this work, we formulate manipulation dynamics as a piecewise-smooth Markov process and show that interaction-induced mode changes create piecewise nonlinear reachability structures that are difficult for standard CRL energy functions to represent and plan over. Based on this analysis, we introduce Interaction-weighted Resampling (IWR). IWR performs interaction-aware resampling around phases before, during, and after interactions, encouraging the learned representation to preserve the mode boundaries that determine future reachability to capture multi-modal and piecewise nonlinear reachability. Across interaction-centric environments, including 2D dynamic control, robotic manipulation, and robot air hockey, IWR improves both sample efficiency and overall performance over prior CRL methods, with 19.8% average improvement in simulation. Finally, using a sim-to-real pipeline with policies trained by IWR, we demonstrate the first real-world goal-conditioned robot air hockey agent capable of hitting goals, improving success from 25% to 60%. Project Page: IWR-arxiv.github.io.

Tongle Shen, Caleb Chuck, Fan Feng +1

1UC San Diego · 2UT Austin

Dynamical Systems Reinforcement Learning Robot Skill Learning Contact-Rich Robotic Manipulation Contrastive RL Goal-Conditioned RL

May 25, 2026cs.CL

Harmony in Diversity: Multi-domain Contrastive Policy Optimization for Large Reasoning Models

Post-training via Reinforcement Learning (RL) has enabled Large Reasoning Models (LRMs) to achieve strong performance in individual domain. However, real-world applications increasingly require general-purpose reasoners rendering strong performance across diverse domains. Mixed-domain post-training aims to achieve this goal by jointly training with mixed domain data, but this often induces capability compromise and degradation among different domains. Existing methods attribute this performance degradation to harmful cross-domain interactions and propose various strategies to mitigate them, but these strategies may also impede beneficial knowledge sharing across domains and in turn fail to match or surpass single-domain performance. To address this problem, we propose \textbf{M}ulti-domain \textbf{C}ontrastive \textbf{P}olicy \textbf{O}ptimization (MCPO), which uses contrastive learning to utilize both positive and negative cross-domain interactions for knowledge sharing and competition. Specifically, we partition each rollout generated by LRMs according to its underlying reasoning structures and use reasoning segments to capture these structures. We thus formulate positive and negative pairs of reasoning segments as mutually augmented examples, which provide supportive and competing signals for knowledge sharing. Subsequently, we design complementary contrastive objectives for cross-domain knowledge sharing and intra-domain knowledge consolidation, targeting compatibility across domains and discriminability within each domain to form a harmonious reasoning space. Experimental results across a broad range of domains show that MCPO alleviates performance degradation caused by mixed-domain training and outperforms single-domain training in most cases.

Zongji Yu, Wenshui Luo, Yiliu Sun +5

Shanghai Jiao Tong University · Shanghai AI Laboratory · Shandong University

Reinforcement Learning RL for Language Model Reasoning Cross-Domain Transfer Learning Multi-Task RL Contrastive RL Large Reasoning Models

May 13, 2026cs.LG

Self-Supervised On-Policy Reinforcement Learning via Contrastive Proximal Policy Optimisation

Contrastive reinforcement learning (CRL) learns goal-conditioned Q-values through a contrastive objective over state-action and goal representations, removing the need for hand-crafted reward functions. Despite impressive success in achieving viable self-supervised learning in RL, all existing CRL algorithms rely on off-policy optimisation and are mostly constrained to continuous action spaces, with little research invested in discrete environments. This leaves CRL disconnected from widely used and effective, modern on-policy training pipelines adopted across both single-agent and multi-agent RL in continuous and discrete environments. To establish a first connection, we introduce Contrastive Proximal Policy Optimisation (CPPO). CPPO is an on-policy contrastive RL algorithm that derives policy advantages directly from contrastive Q-values and optimises them via the standard PPO objective, without requiring a reward function or a replay buffer. We evaluate CPPO across continuous and discrete, single-agent and cooperative multi-agent tasks. Whilst the existence of an on-policy approach is inherently useful, we observe that \textbf{CPPO not only significantly outperforms the previous CRL baselines in 14 out of 18 tasks, but also matches or exceeds PPO's performance, which uses hand-crafted dense rewards, in 12 out of the 18 tasks tested.}

Asim Osman, Sasha Abramowitz, Mark Bergh +13

InstaDeep, AIMS · InstaDeep, University of Stellenbosch

Reinforcement Learning Contrastive RL Goal-Conditioned RL

May 13, 2026cs.LG

Revisiting Reinforcement Learning with Verifiable Rewards from a Contrastive Perspective

Group Relative Policy Optimization (GRPO) is one of the most widely adopted RLVR algorithms for post-training large language models on reasoning tasks. We first show that GRPO admits an equivalent discriminative reformulation, in which policy optimization maximizes the expected score gap between verified positive and negative rollouts. This reformulation reveals two objective-level limitations: likelihood-misaligned surrogate scores, in which clipped ratio-based scores are optimized rather than the sequence likelihoods that govern generation, and score-insensitive credit assignment, in which rollout-level credit does not reflect the current score gaps between positive and negative rollouts. To address these limitations, we propose ConSPO, a Contrastive Sequence-level Policy Optimization method that uses length-normalized sequence log-probabilities as rollout scores and contrasts verified positive rollouts against negative distractors within the same group. ConSPO optimizes a group-wise InfoNCE-style objective to adaptively strengthen updates for poorly separated positives and high-scoring negatives, together with a curriculum-scheduled margin that preserves separation pressure as training progresses. Experiments across diverse settings show that ConSPO outperforms strong baselines on challenging reasoning benchmarks. Code will be released upon paper acceptance.

Feng Zhang, Xinhong Ma, Ziqiang Dong +5

Beijing Institute of Technology · Qwen Business Unit of Alibaba · The Chinese University of Hong Kong, Shenzhen +1

RL for Language Model Reasoning Group Relative Policy Optimization Policy Optimization Reinforcement Learning with Verifiable Rewards RL Post-Training Contrastive RL

May 12, 2026cs.LG

From Generic Correlation to Input-Specific Credit in On-Policy Self Distillation

On-policy self-distillation has emerged as a promising paradigm for post-training language models, in which the model conditions on environment feedback to serve as its own teacher, providing dense token-level rewards without external teacher models or step-level annotations. Despite its empirical success, what this reward actually measures and what kind of credit it assigns remain unclear. Under a posterior-compatibility interpretation of feedback conditioning, standard in the implicit-reward literature, we show that the self-distillation token reward is a Bayesian filtering increment whose trajectory sum is exactly the pointwise mutual information between the response and the feedback given the input. This pMI can be raised by input-specific reasoning or by input-generic shortcuts, so we further decompose the teacher log-probability along the input axis. Based on this analysis, we propose CREDIT (Contrastive REward from DIsTillation), which isolates the input-specific component with a batch-contrastive baseline. At the sequence level, CREDIT is a teacher-side surrogate for a contrastive pMI objective that also penalizes responses remaining likely under unrelated inputs. Across coding, scientific reasoning, and tool-use benchmarks on two model families, CREDIT delivers the strongest aggregate performance at negligible additional compute.

Guobin Shen, Lei Huang, Xiang Cheng +4

Xiaohongshu Inc. · Institute of Automation, Chinese Academy of Sciences

Reinforcement Learning On-Policy Self-Distillation Token-Level Credit Assignment Language Model Post-Training Contrastive RL Mutual Information

Apr 30, 2026cs.AI

PRTS: A Primitive Reasoning and Tasking System via Contrastive Representations

Vision-Language-Action (VLA) models advance robotic control via strong visual-linguistic priors. However, existing VLAs predominantly frame pretraining as supervised behavior cloning, overlooking the fundamental nature of robot learning as a goal-reaching process that requires understanding temporal task progress. We present \textbf{PRTS} (\textbf{P}rimitive \textbf{R}easoning and \textbf{T}asking \textbf{S}ystem), a VLA foundation model that reformulates pretraining through Goal-Conditioned Reinforcement Learning. By treating language instructions as goals and employing contrastive reinforcement learning, PRTS learns a unified embedding space where the inner product of state-action and goal embeddings approximates the log-discounted goal occupancy, the probability of reaching the language-specified goal from the current state-action, quantitatively assessing physical feasibility beyond static semantic matching. PRTS draws this dense goal-reachability supervision directly from offline trajectories without reward annotations, and folds it into the VLM backbone via a role-aware causal mask, incurring negligible overhead over vanilla behavior cloning. This paradigm endows the high-level reasoning system with intrinsic goal reachability awareness, bridging semantic reasoning and temporal task progress, and further benefits goal-conditioned action prediction. Pretrained on 167B tokens of diverse manipulation and embodied-reasoning data, PRTS reaches state-of-the-art performance on LIBERO, LIBERO-Pro, LIBERO-Plus, SimplerEnv, and a real-world suite of 14 complex tasks, with particularly substantial gains on long-horizon, contact-rich, and zero-shot novel-instruction settings, confirming that injecting goal-reachability awareness significantly improves both execution success and long-horizon planning of general-purpose robotic foundation policies.

Yang Zhang, Jiangyuan Zhao, Chenyou Fan +11

Institute of Artificial Intelligence (TeleAI), China Telecom · Tsinghua University · Shanghai Jiao Tong University +1

Reinforcement Learning Long-Horizon Robotic Manipulation Language-Conditioned Robot Manipulation Vision-Language-Action Models Contrastive RL Goal-Conditioned RL

Feb 7, 2026cs.LG

Scalable Dexterous Robot Learning with AR-based Remote Human-Robot Interactions

This paper focuses on the scalable robot learning for manipulation in the dexterous robot arm-hand systems, where the remote human-robot interactions via augmented reality (AR) are established to collect the expert demonstration data for improving efficiency. In such a system, we present a novel method to address the general manipulation task problem. Specifically, the proposed method consists of two phases: i) In the first phase for pretraining, the policy is created in a behavior cloning (BC) manner, through leveraging the learning data from our AR-based remote human-robot interaction system; ii) In the second phase, a contrastive learning empowered reinforcement learning (RL) method is developed to obtain more efficient and robust policy than the BC, and thus a projection head is designed to accelerate the learning progress. An event-driven augmented reward is adopted for enhancing the safety. To validate the proposed method, both the physics simulations via PyBullet and real-world experiments are carried out. The results demonstrate that compared to the baselines, our method not only significantly speeds up the training process, but also achieves much better performance in terms of the success rate for fulfilling the manipulation tasks. By conducting the ablation study, it is confirmed that the proposed RL with contrastive learning overcomes policy collapse. Supplementary demonstrations are available at https://cyberyyc.github.io/.

Yicheng Yang, Ruijiao Li, Lifeng Wang +7

Reinforcement Learning Robotic RL Robotic Manipulation Robot Imitation Learning Contrastive RL

Jul 18, 2025cs.AI

CUDA-L1: Improving CUDA Optimization via Contrastive Reinforcement Learning

The exponential growth in demand for GPU computing resources has created an urgent need for automated CUDA optimization strategies. While recent advances in LLMs show promise for code generation, current SOTA models achieve low success rates in improving CUDA speed. In this paper, we introduce CUDA-L1, an automated reinforcement learning framework for CUDA optimization that employs a novel contrastive RL algorithm. CUDA-L1 achieves significant performance improvements on the CUDA optimization task: trained on A100, it delivers an average speedup of x3.12 with a median speedup of x1.42 against default baselines over across all 250 CUDA kernels of KernelBench, with peak speedups reaching x120. In addition to the default baseline provided by KernelBench, CUDA-L1 demonstrates x2.77 over Torch Compile, x2.88 over Torch Compile with reduce overhead, x2.81 over CUDA Graph implementations, and x7.72 over cuDNN libraries. Furthermore, the model also demonstrates portability across different GPU architectures. Beyond these benchmark results, CUDA-L1 demonstrates several properties: it 1) discovers a variety of CUDA optimization techniques and learns to combine them strategically to achieve optimal performance; 2) uncovers fundamental principles of CUDA optimization, such as the multiplicative nature of optimizations; 3) identifies non-obvious performance bottlenecks and rejects seemingly beneficial optimizations that actually harm performance. The capabilities demonstrate that, RL can transform an initially poor-performing LLM into an effective CUDA optimizer through speedup-based reward signals alone, without human expertise or domain knowledge. This paradigm opens possibilities for automated optimization of CUDA operations, and holds promise to substantially promote GPU efficiency and alleviate the rising pressure on GPU computing resources. Project: deepreinforce-ai.github.io/cudal1_blog

Xiaoya Li, Albert Wang, Guoyin Wang +2

DeepReinforce Team

Reinforcement Learning GPU Kernel Optimization GPU Acceleration Large Language Model-Guided Optimization Contrastive RL

Mar 19, 2025cs.LG

1000 Layer Networks for Self-Supervised RL: Scaling Depth Can Enable New Goal-Reaching Capabilities

Scaling up self-supervised learning has driven breakthroughs in language and vision, yet comparable progress has remained elusive in reinforcement learning (RL). In this paper, we study building blocks for self-supervised RL that unlock substantial improvements in scalability, with network depth serving as a critical factor. Whereas most RL papers in recent years have relied on shallow architectures (around 2 - 5 layers), we demonstrate that increasing the depth up to 1024 layers can significantly boost performance. Our experiments are conducted in an unsupervised goal-conditioned setting, where no demonstrations or rewards are provided, so an agent must explore (from scratch) and learn how to maximize the likelihood of reaching commanded goals. Evaluated on simulated locomotion and manipulation tasks, our approach increases performance on the self-supervised contrastive RL algorithm by

2\times

-

50\times

, outperforming other goal-conditioned baselines. Increasing the model depth not only increases success rates but also qualitatively changes the behaviors learned. The project webpage and code can be found here: https://wang-kevin3290.github.io/scaling-crl/.

Kevin Wang, Ishaan Javali, Michał Bortkiewicz +2

Princeton University · Warsaw University of Technology · Warsaw University of Technology, Tooploox, IDEAS Research Institute

Reinforcement Learning Robotic RL Contrastive RL Goal-Conditioned RL