Diff-Instruct with Diffused Reward: Towards Principled One-step Generator RL · 相关技术
相关技术
远程代码执行（RCE）reinforcement learningimage generationdivide-and-conquer partitioningalignmentUCTText-to-imageTerminalStraight-Through EstimatorSpatial Pivot-Aligned Coordinate-free Embedding (SPACE)SULSPAReferring expression comprehension (REC)RLHFParts-of-Speech (POS) tagsNarrative Abstraction BenchmarkHISForFFIEffort Metric AttentionENADiffusionANN