Every processed story in chronological order, with the newest coverage first. Filter by tag, source, or score to drill in.
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.
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.