POPO uses bounded importance sampling on positive rollouts and a siamese policy network to achieve implicit negative gradients and stable optimization, matching or exceeding GRPO on math benchmarks such as 36.67% on AIME 2025.
hub
Advances in Neural Information Processing Systems , volume=
17 Pith papers cite this work. Polarity classification is still indexing.
17
Pith papers citing it
hub tools
representative citing papers
Validity-calibrated reasoning distillation improves transfer of reasoning skills by modulating updates based on relative local validity of next steps instead of enforcing full trajectory imitation.
o1-like models overthink easy tasks; self-training reduces compute use without accuracy loss on GSM8K, MATH500, GPQA, and AIME.
HölderPO unifies token-level aggregation in GRPO via the Hölder mean with a tunable p parameter and annealing schedule, delivering 54.9% average accuracy on math benchmarks and 93.8% success on ALFWorld.
ReFlect is a harness that wraps LLMs to detect and recover from reasoning errors, achieving 7-29 pp gains over direct CoT on long-horizon tasks and improving code patch quality to 82-87%.
The power distribution is the target of power sampling, the closed-form solution to self-reward KL-regularized RL, and the basis for power self-distillation that matches sampling performance at lower cost.
MSD enables cross-lingual safety transfer in LLMs via self-distillation with Dual-Perspective Safety Weighting, improving safety in low-resource languages without target response data.
SSA uses learned gist tokens to score and selectively unfold relevant context chunks, achieving sparse attention without auxiliary KV caches or architectural changes.
A multi-agent AI system generates novel biomedical hypotheses that show promising experimental validation in drug repurposing for leukemia, new targets for liver fibrosis, and a bacterial gene transfer mechanism.
LIMO achieves 63.3% on AIME24 and 95.6% on MATH500 via supervised fine-tuning on roughly 1% of the data used by prior models, supporting the claim that minimal strategic examples suffice when pre-training has already encoded domain knowledge.
Agent Q integrates MCTS-guided search, self-critique, and off-policy DPO to train LLM agents that outperform behavior cloning and reinforced fine-tuning baselines in WebShop and achieve up to 95.4% success in real-world booking scenarios.
BALAR is a task-agnostic Bayesian loop that maintains structured beliefs over latent states, selects questions via expected mutual information, and expands its state space when needed, delivering 14.6-38.5% accuracy gains over baselines on detective, puzzle, and clinical diagnosis benchmarks.
ReMedi boosts LLM performance on EHR clinical predictions by up to 19.9% F1 through ground-truth-guided rationale regeneration and fine-tuning.
A co-evolving proposer-critic RL framework improves GUI grounding accuracy by letting the model critique its own proposals rendered on screenshots.
Fine-tuned 8B LLMs produce children's English reading stories with better difficulty control and safety than zero-shot GPT-4o and Llama 3.3 70B.
The paper delivers the first systematic review of self-evolving agents, structured around what components evolve, when adaptation occurs, and how it is implemented.
Oxygen AIIC is an industrial platform using LLMs and VLMs for scalable item knowledge production and service at JD.com, reporting 94.2% precision and 82.8% recall along with business metric improvements.
citing papers explorer
-
Beyond Negative Rollouts: Positive-Only Policy Optimization with Implicit Negative Gradients
POPO uses bounded importance sampling on positive rollouts and a siamese policy network to achieve implicit negative gradients and stable optimization, matching or exceeding GRPO on math benchmarks such as 36.67% on AIME 2025.
-
Validity-Calibrated Reasoning Distillation
Validity-calibrated reasoning distillation improves transfer of reasoning skills by modulating updates based on relative local validity of next steps instead of enforcing full trajectory imitation.
-
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.
-
Holder Policy Optimisation
HölderPO unifies token-level aggregation in GRPO via the Hölder mean with a tunable p parameter and annealing schedule, delivering 54.9% average accuracy on math benchmarks and 93.8% success on ALFWorld.
-
ReFlect: An Effective Harness System for Complex Long-Horizon LLM Reasoning
ReFlect is a harness that wraps LLMs to detect and recover from reasoning errors, achieving 7-29 pp gains over direct CoT on long-horizon tasks and improving code patch quality to 82-87%.
-
Power Distribution Bridges Sampling, Self-Reward RL, and Self-Distillation
The power distribution is the target of power sampling, the closed-form solution to self-reward KL-regularized RL, and the basis for power self-distillation that matches sampling performance at lower cost.
-
Multilingual Safety Alignment via Self-Distillation
MSD enables cross-lingual safety transfer in LLMs via self-distillation with Dual-Perspective Safety Weighting, improving safety in low-resource languages without target response data.
-
Simplified Sparse Attention via Gist Tokens
SSA uses learned gist tokens to score and selectively unfold relevant context chunks, achieving sparse attention without auxiliary KV caches or architectural changes.
-
Towards an AI co-scientist
A multi-agent AI system generates novel biomedical hypotheses that show promising experimental validation in drug repurposing for leukemia, new targets for liver fibrosis, and a bacterial gene transfer mechanism.
-
LIMO: Less is More for Reasoning
LIMO achieves 63.3% on AIME24 and 95.6% on MATH500 via supervised fine-tuning on roughly 1% of the data used by prior models, supporting the claim that minimal strategic examples suffice when pre-training has already encoded domain knowledge.
-
Agent Q: Advanced Reasoning and Learning for Autonomous AI Agents
Agent Q integrates MCTS-guided search, self-critique, and off-policy DPO to train LLM agents that outperform behavior cloning and reinforced fine-tuning baselines in WebShop and achieve up to 95.4% success in real-world booking scenarios.
-
BALAR : A Bayesian Agentic Loop for Active Reasoning
BALAR is a task-agnostic Bayesian loop that maintains structured beliefs over latent states, selects questions via expected mutual information, and expands its state space when needed, delivering 14.6-38.5% accuracy gains over baselines on detective, puzzle, and clinical diagnosis benchmarks.
-
ReMedi: Reasoner for Medical Clinical Prediction
ReMedi boosts LLM performance on EHR clinical predictions by up to 19.9% F1 through ground-truth-guided rationale regeneration and fine-tuning.
-
Measure Twice, Click Once: Co-evolving Proposer and Visual Critic via Reinforcement Learning for GUI Grounding
A co-evolving proposer-critic RL framework improves GUI grounding accuracy by letting the model critique its own proposals rendered on screenshots.
-
Children's English Reading Story Generation via Supervised Fine-Tuning of Compact LLMs with Controllable Difficulty and Safety
Fine-tuned 8B LLMs produce children's English reading stories with better difficulty control and safety than zero-shot GPT-4o and Llama 3.3 70B.
-
A Survey of Self-Evolving Agents: What, When, How, and Where to Evolve on the Path to Artificial Super Intelligence
The paper delivers the first systematic review of self-evolving agents, structured around what components evolve, when adaptation occurs, and how it is implemented.
-
JD Oxygen AI Item Center (Oxygen AIIC) V1: An Industrial-Scale LLM/VLM-Centric Solution for Item Understanding, Management, and Applications
Oxygen AIIC is an industrial platform using LLMs and VLMs for scalable item knowledge production and service at JD.com, reporting 94.2% precision and 82.8% recall along with business metric improvements.