The Mirage of Optimizing Training Policies: Monotonic Inference Policies as the Real Objective for LLM Reinforcement Learning
Paper • 2606.29526 • Published • 170
Amara Ndlovu
test-99
·
AI & ML interests
None yet
Recent Activity
liked a dataset 4 days ago
cubelawm/alihunn liked a model 9 days ago
Wangchy226/dqn-SpaceInvadersNoFrameskip-v4 liked a dataset 10 days ago
umwe/sabyinyo-v04-datasetOrganizations
None yet
upvoted a paper 22 days ago
upvoted 4
papers about 2 months ago
Imaginative Perception Tokens Enhance Spatial Reasoning in Multimodal Language Models
Paper • 2606.03988 • Published • 126
GRAIL: Gradient-Reweighted Advantages for Reinforcement Learning with Verifiable Rewards
Paper • 2606.04889 • Published • 4
On the Scaling of PEFT: Towards Million Personal Models of Trillion Parameters
Paper • 2606.02437 • Published • 241
Gamma-World: Generative Multi-Agent World Modeling Beyond Two Players
Paper • 2605.28816 • Published • 433
upvoted 3
papers 2 months ago
AutoResearchClaw: Self-Reinforcing Autonomous Research with Human-AI Collaboration
Paper • 2605.20025 • Published • 191
Learning to Foresee: Unveiling the Unlocking Efficiency of On-Policy Distillation
Paper • 2605.11739 • Published • 61
CiteVQA: Benchmarking Evidence Attribution for Trustworthy Document Intelligence
Paper • 2605.12882 • Published • 274
upvoted 4
papers 4 months ago
GEMS: Agent-Native Multimodal Generation with Memory and Skills
Paper • 2603.28088 • Published • 88
Out of Sight but Not Out of Mind: Hybrid Memory for Dynamic Video World Models
Paper • 2603.25716 • Published • 157
SQuTR: A Robustness Benchmark for Spoken Query to Text Retrieval under Acoustic Noise
Paper • 2602.12783 • Published • 246
MinerU-Diffusion: Rethinking Document OCR as Inverse Rendering via Diffusion Decoding
Paper • 2603.22458 • Published • 139
upvoted 6
papers 5 months ago
Bootstrapping Exploration with Group-Level Natural Language Feedback in Reinforcement Learning
Paper • 2603.04597 • Published • 211
A Very Big Video Reasoning Suite
Paper • 2602.20159 • Published • 526
VESPO: Variational Sequence-Level Soft Policy Optimization for Stable Off-Policy LLM Training
Paper • 2602.10693 • Published • 221
TermiGen: High-Fidelity Environment and Robust Trajectory Synthesis for Terminal Agents
Paper • 2602.07274 • Published • 212
Less is Enough: Synthesizing Diverse Data in Feature Space of LLMs
Paper • 2602.10388 • Published • 246
NarraScore: Bridging Visual Narrative and Musical Dynamics via Hierarchical Affective Control
Paper • 2602.09070 • Published • 46