Every processed story in chronological order, with the newest coverage first. Filter by tag, source, or score to drill in.
The contrastive context-selection objective demonstrably outperforms simply adding more contrastive data, showing that how the training signal is structured — not just what data is used — drives grounding improvements in both agentic and multimodal LLM settings.
RefGRPO demonstrates that agent self-assessment can be substantially improved without any external annotation or reward model, enabling agents to act as their own verifiers grounded in environment feedback.
InterleaveThinker removes the architectural barrier that has prevented existing image generators from producing interleaved text-image sequences, extending a capability previously limited to frontier models like GPT-5 to any image generator via a plug-in multi-agent pipeline.
The work removes the rollout stage as the key bottleneck in RL training pipelines by showing that a pre-RL MTP training recipe with TV loss and rejection sampling sustains high acceptance rates throughout RL without costly online updates, delivering up to 1.8x end-to-end acceleration.
EvoDS directly addresses two core failure modes of current LLM-based data science automation — static skill sets and context overflow — with a system that learns to expand its own capabilities and manage long-horizon context, achieving a 28.9% average improvement over existing open-source agents across four benchmarks.
TMEM demonstrates that agent parameters can be updated within a single episode via online LoRA adaptation, overcoming the permanent information loss that affects all prompt-only memory approaches.
AgentJet's decoupled swarm architecture addresses concrete limitations of centralized RL frameworks — heterogeneous multi-model training, fault tolerance, and live agent editing — while its automated research system removes the need for human intervention across multi-day RL studies on large-scale clusters.
AMC demonstrates that principled RL-style optimization of black-box LLM agents is feasible at test time, opening a path to improving proprietary API-only agents without requiring access to model weights.
Teams building AI-powered web development tools can use WebGen-R1's RL approach and multimodal reward design as a blueprint for training small, efficient models to handle full project-level code generation without relying on expensive proprietary APIs.
Developers building AI-powered educational or presentation tools can use the ManimTrainer/ManimAgent framework as a blueprint for combining fine-tuning and agentic inference to reliably generate high-quality programmatic animations from text prompts.