Copertina

Where Rollouts Begin: Low-Load, High-Leverage First-Token Diversification for RLVR

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Kim, Soeun, No, Albert
Natura:	Preprint
Pubblicazione:	2026
Soggetti:	Artificial Intelligence Computation and Language Machine Learning
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Assigning Distinct Roles to Quantized and Low-Rank Matrices Toward Optimal Weight Decomposition
di: Cho, Yoonjun, et al.
Pubblicazione: (2025)

JudgeRLVR: Judge First, Generate Second for Efficient Reasoning
di: Duo, Jiangshan, et al.
Pubblicazione: (2026)

Exploring Multi-Temperature Strategies for Token- and Rollout-Level Control in RLVR
di: Zhuang, Haomin, et al.
Pubblicazione: (2025)

Low-rank Optimization Trajectories Modeling for LLM RLVR Acceleration
di: Chen, Zhipeng, et al.
Pubblicazione: (2026)

Sparse but Critical: A Token-Level Analysis of Distributional Shifts in RLVR Fine-Tuning of LLMs
di: Meng, Haoming, et al.
Pubblicazione: (2026)

Not All Rollouts are Useful: Down-Sampling Rollouts in LLM Reinforcement Learning
di: Xu, Yixuan Even, et al.
Pubblicazione: (2025)

CLIPO: Contrastive Learning in Policy Optimization Generalizes RLVR
di: Cui, Sijia, et al.
Pubblicazione: (2026)

SAGE: Shaping Anchors for Guided Exploration in RLVR of LLMs
di: Lee, Chanuk, et al.
Pubblicazione: (2026)

Asymmetric Advantage Modulation Calibrates Entropy Dynamics in RLVR
di: Gu, Hengrui, et al.
Pubblicazione: (2026)

RLPR: Extrapolating RLVR to General Domains without Verifiers
di: Yu, Tianyu, et al.
Pubblicazione: (2025)

Generalization of RLVR Using Causal Reasoning as a Testbed
di: Lu, Brian, et al.
Pubblicazione: (2025)

Token-Efficient Leverage Learning in Large Language Models
di: Zeng, Yuanhao, et al.
Pubblicazione: (2024)

Flexible Entropy Control in RLVR with a Gradient-Preserving Perspective
di: Chen, Kun, et al.
Pubblicazione: (2026)

IFCap: Image-like Retrieval and Frequency-based Entity Filtering for Zero-shot Captioning
di: Lee, Soeun, et al.
Pubblicazione: (2024)

ViPCap: Retrieval Text-Based Visual Prompts for Lightweight Image Captioning
di: Kim, Taewhan, et al.
Pubblicazione: (2024)

Adaptive Rollout Allocation for Online Reinforcement Learning with Verifiable Rewards
di: Nguyen, Hieu Trung, et al.
Pubblicazione: (2026)

SPEC-RL: Accelerating On-Policy Reinforcement Learning with Speculative Rollouts
di: Liu, Bingshuai, et al.
Pubblicazione: (2025)

The Invisible Leash: Why RLVR May or May Not Escape Its Origin
di: Wu, Fang, et al.
Pubblicazione: (2025)

TSR: Trajectory-Search Rollouts for Multi-Turn RL of LLM Agents
di: Djuhera, Aladin, et al.
Pubblicazione: (2026)

Countdown-Code: A Testbed for Studying The Emergence and Generalization of Reward Hacking in RLVR
di: Khalifa, Muhammad, et al.
Pubblicazione: (2026)

Exploration vs Exploitation: Rethinking RLVR through Clipping, Entropy, and Spurious Reward
di: Chen, Peter, et al.
Pubblicazione: (2025)

Knowledgeable Agents by Offline Reinforcement Learning from Large Language Model Rollouts
di: Pang, Jing-Cheng, et al.
Pubblicazione: (2024)

A State-of-the-Art SQL Reasoning Model using RLVR
di: Ali, Alnur, et al.
Pubblicazione: (2025)

$V_{0.5}$: Generalist Value Model as a Prior for Sparse RL Rollouts
di: Zhang, Yi-Kai, et al.
Pubblicazione: (2026)

Mitigating Distribution Sharpening in Math RLVR via Distribution-Aligned Hint Synthesis and Backward Hint Annealing
di: Xie, Pei-Xi, et al.
Pubblicazione: (2026)

Sparse-RL: Breaking the Memory Wall in LLM Reinforcement Learning via Stable Sparse Rollouts
di: Luo, Sijia, et al.
Pubblicazione: (2026)

Prismatic Synthesis: Gradient-based Data Diversification Boosts Generalization in LLM Reasoning
di: Jung, Jaehun, et al.
Pubblicazione: (2025)

Do Not Let Low-Probability Tokens Over-Dominate in RL for LLMs
di: Yang, Zhihe, et al.
Pubblicazione: (2025)

Where Hindsight Credit Can Reside: A Signed-Capacity View of Token Updates in RLVR
di: He, Yuhang, et al.
Pubblicazione: (2026)

Forget What Matters, Keep the Rest: Selective Unlearning of Informative Tokens
di: Koh, Seunghee, et al.
Pubblicazione: (2026)

When Every Token Counts: Optimal Segmentation for Low-Resource Language Models
di: Raj, Bharath, et al.
Pubblicazione: (2024)

Nested-ReFT: Efficient Reinforcement Learning for Large Language Model Fine-Tuning via Off-Policy Rollouts
di: Heuillet, Maxime, et al.
Pubblicazione: (2025)

Improving Data Efficiency for LLM Reinforcement Fine-tuning Through Difficulty-targeted Online Data Selection and Rollout Replay
di: Sun, Yifan, et al.
Pubblicazione: (2025)

Empowering Small-Scale Knowledge Graphs: A Strategy of Leveraging General-Purpose Knowledge Graphs for Enriched Embeddings
di: Sawczyn, Albert, et al.
Pubblicazione: (2024)

SynC: Synthetic Image Caption Dataset Refinement with One-to-many Mapping for Zero-shot Image Captioning
di: Kim, Si-Woo, et al.
Pubblicazione: (2025)

Looking to Learn: Token-wise Dynamic Gating for Low-Resource Vision-Language Modelling
di: Ganescu, Bianca-Mihaela, et al.
Pubblicazione: (2025)

Where Do Reasoning Models Refuse?
di: Yamaguchi, Kureha, et al.
Pubblicazione: (2025)

TokenButler: Token Importance is Predictable
di: Akhauri, Yash, et al.
Pubblicazione: (2025)

AI-Augmented Surveys: Leveraging Large Language Models and Surveys for Opinion Prediction
di: Kim, Junsol, et al.
Pubblicazione: (2023)

PaperSearchQA: Learning to Search and Reason over Scientific Papers with RLVR
di: Burgess, James, et al.
Pubblicazione: (2026)