TAC is a bandit curriculum for multi-domain RLVR that prioritizes domains whose gradient updates align with and benefit other domains, yielding up to 2.8-point macro accuracy gains over learnability-only baselines on Qwen3-1.7B and Llama3.2-3B.
Omni-think: Scaling cross-domain generalization in llms via multi-task rl with hybrid rewards
7 Pith papers cite this work. Polarity classification is still indexing.
7
Pith papers citing it
representative citing papers
Later-domain RL training harms earlier domains via second-order damage concentrated in a low-dimensional shared conflict subspace; brief domain refresh contracts this component to enable selective recovery.
A prompt-level reward specification framework constructs reusable rubrics and executable checkers from prompts alone to deliver hybrid rewards combining requirement satisfaction, holistic quality, and deterministic constraints for LLM post-training.
A new benchmark for speculative decoding that maximizes semantic diversity and supports throughput evaluation across input lengths, exposing biases in synthetic benchmarks.
TrOPD stabilizes on-policy distillation for LLMs with trust-region learning, outlier estimation, and off-policy guidance, outperforming prior OPD methods on reasoning and code benchmarks.
SPARD dynamically tunes multi-objective reward weights and data importance in LLM reinforcement learning alignment using a self-paced curriculum driven by reward dynamics and data utility.
A survey compiling RL methods, challenges, data resources, and applications for enhancing reasoning in large language models and large reasoning models since DeepSeek-R1.
citing papers explorer
-
Transferability for General Reasoning: An Automated Curriculum for Multi-Domain RLVR
TAC is a bandit curriculum for multi-domain RLVR that prioritizes domains whose gradient updates align with and benefit other domains, yielding up to 2.8-point macro accuracy gains over learnability-only baselines on Qwen3-1.7B and Llama3.2-3B.
-
A Local Perturbation Theory for Cross-Domain Interference and Recovery in Multi-Domain RL
Later-domain RL training harms earlier domains via second-order damage concentrated in a low-dimensional shared conflict subspace; brief domain refresh contracts this component to enable selective recovery.
-
Prompt-Level Reward Specifications for Open-Ended Post-Training
A prompt-level reward specification framework constructs reusable rubrics and executable checkers from prompts alone to deliver hybrid rewards combining requirement satisfaction, holistic quality, and deterministic constraints for LLM post-training.
-
SPEED-Bench: A Unified and Diverse Benchmark for Speculative Decoding
A new benchmark for speculative decoding that maximizes semantic diversity and supports throughput evaluation across input lengths, exposing biases in synthetic benchmarks.
-
Trust Region On-Policy Distillation
TrOPD stabilizes on-policy distillation for LLMs with trust-region learning, outlier estimation, and off-policy guidance, outperforming prior OPD methods on reasoning and code benchmarks.
-
SPARD: Self-Paced Curriculum for RL Alignment via Integrating Reward Dynamics and Data Utility
SPARD dynamically tunes multi-objective reward weights and data importance in LLM reinforcement learning alignment using a self-paced curriculum driven by reward dynamics and data utility.
-
A Survey of Reinforcement Learning for Large Reasoning Models
A survey compiling RL methods, challenges, data resources, and applications for enhancing reasoning in large language models and large reasoning models since DeepSeek-R1.