Tag: post-training
Browse our exclusive articles!
Agent² RL-Bench: Evaluating LLM Agents in RL Post-Training
Discover how Agent² RL-Bench tests LLM agents' ability to engineer agentic reinforcement learning post-training with dynamic, interactive benchmarks.
Enhance Video Understanding with Visually Grounded Training
Discover how visually grounded post-training boosts vision-language models for better video understanding and efficiency.
User Turn Generation Reveals Interaction Awareness in LLMs
Discover how user turn generation probes interaction awareness in language models, uncovering deeper conversational understanding beyond assistant response...
Hallucination’s Impact on Reinforcement Training in Multimodal Models
Explore how hallucination influences reinforcement post-training in multimodal reasoning models, boosting visual and multimodal AI performance.
Reactivating Hidden Safety in Post-Trained Large Language Models
Discover how SafeReAct restores suppressed safety mechanisms in post-trained LLMs, enhancing security without losing reasoning power.