EVE enables verifiable self-evolution of MLLMs by using a Challenger-Solver architecture to generate dynamic executable visual transformations that produce VQA problems with absolute execution-verified ground truth.
arXiv preprint arXiv:2603.08660 , year=
13 Pith papers cite this work. Polarity classification is still indexing.
13
Pith papers citing it
citation-role summary
background 2
citation-polarity summary
years
2026 13roles
background 2representative citing papers
Empirical evaluation on LiveCodeBench shows certainty-based RLIF yields early gains followed by output shortening and reasoning collapse, providing no advantage for RLVR initialization on code tasks.
OPRD performs distillation in hidden-state space on on-policy data for deterministic gradients and better math benchmark performance, plus OPRD-Bridge for cross-architecture transfer via low-rank projectors.
TTRL gains are reinterpreted as mostly sharpening rather than learning, with an identified extinction window causing net corruption; TTRL-Guard mitigates via FRS, MPS, and RCSU for improved pass@1.
Transferring a weak model’s RL-induced log-ratio policy shift on a strong student’s own rollouts raises AIME accuracy more cheaply than imitating the weak teacher or running matched-step RL on the student.
REINFORCE self-training on competitive programming tasks exhibits robust rise-then-collapse in pass@1; CARE, ES, and GRPO mitigate it in model-size-dependent ways across Qwen-2.5-3B/7B and a Gemma pilot.
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.
Neuron-OPSD uses neuron activations to guide data selection and teacher construction for annotation-free on-policy self-distillation in LLMs, claiming better in-domain results without harming cross-domain performance or calibration.
Lightweight modular latent memories trained on self-generated rewards enable continual self-improvement in LLMs, outperforming raw ICL and matching offline training on math benchmarks.
GeoMin uses geometric distribution modeling on labeled data to assess self-reward reliability, enabling better performance in semi-supervised RLVR with only 10% of typical annotations.
RLAVR uses the Corrective Advantage Gap metric and CARE policy to actively acquire ground-truth labels for key samples, stabilizing RLVR training and boosting performance with limited annotation budgets.
On-policy distillation gains efficiency from early foresight in module allocation and update directions, which the proposed EffOPD method exploits for 3x faster training with comparable performance.
citing papers explorer
-
EVE: Verifiable Self-Evolution of MLLMs via Executable Visual Transformations
EVE enables verifiable self-evolution of MLLMs by using a Challenger-Solver architecture to generate dynamic executable visual transformations that produce VQA problems with absolute execution-verified ground truth.
-
When Do Intrinsic Rewards Work for Code Reasoning? A Comprehensive Study
Empirical evaluation on LiveCodeBench shows certainty-based RLIF yields early gains followed by output shortening and reasoning collapse, providing no advantage for RLVR initialization on code tasks.
-
OPRD: On-Policy Representation Distillation
OPRD performs distillation in hidden-state space on on-policy data for deterministic gradients and better math benchmark performance, plus OPRD-Bridge for cross-architecture transfer via low-rank projectors.
-
Detecting and Mitigating the Correct-Answer Extinction Window in Test-Time Reinforcement Learning with Majority Voting
TTRL gains are reinterpreted as mostly sharpening rather than learning, with an identified extinction window causing net corruption; TTRL-Guard mitigates via FRS, MPS, and RCSU for improved pass@1.
-
Weak-to-Strong Generalization via Direct On-Policy Distillation
Transferring a weak model’s RL-induced log-ratio policy shift on a strong student’s own rollouts raises AIME accuracy more cheaply than imitating the weak teacher or running matched-step RL on the student.
-
Self-Improvement Can Self-Regress: The Rise-and-Collapse Failure Mode of LLM Self-Training
REINFORCE self-training on competitive programming tasks exhibits robust rise-then-collapse in pass@1; CARE, ES, and GRPO mitigate it in model-size-dependent ways across Qwen-2.5-3B/7B and a Gemma pilot.
-
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.
-
Neuron-Aware Data Selection for Annotation-Free LLM Self-Distillation
Neuron-OPSD uses neuron activations to guide data selection and teacher construction for annotation-free on-policy self-distillation in LLMs, claiming better in-domain results without harming cross-domain performance or calibration.
-
Continual Self-Improvement with Lightweight Experiential Latent Memories
Lightweight modular latent memories trained on self-generated rewards enable continual self-improvement in LLMs, outperforming raw ICL and matching offline training on math benchmarks.
-
GeoMin: Data-Efficient Semi-Supervised RLVR via Geometric Distribution Modeling
GeoMin uses geometric distribution modeling on labeled data to assess self-reward reliability, enabling better performance in semi-supervised RLVR with only 10% of typical annotations.
-
When Self-Belief Misleads: Active Label Acquisition for Reinforcement Learning with Verifiable Rewards
RLAVR uses the Corrective Advantage Gap metric and CARE policy to actively acquire ground-truth labels for key samples, stabilizing RLVR training and boosting performance with limited annotation budgets.
-
Learning to Foresee: Unveiling the Unlocking Efficiency of On-Policy Distillation
On-policy distillation gains efficiency from early foresight in module allocation and update directions, which the proposed EffOPD method exploits for 3x faster training with comparable performance.
- RemoteZero: Geospatial Reasoning with Zero Labels