Mode Collapse
A failure mode in generative models in which the model produces only a small number of sample types and fails to cover the multimodal characteristics of the data distribution, resulting in insufficient diversity.
Timeline (last 90 days)
When fine-tuning precipitation diffusion models with reinforcement learning, improper reward design can lead to mode collapse (reduced generation diversity) or reward hacking (optimizing only the score at the expense of physical consistency), and training requires extensive sampling, resulting in high computational cost.
The reverse KL divergence is zero-forcing and tends toward mode seeking, causing severe mode collapse when training diffusion samplers with variational inference.
All Facts (2)
When fine-tuning precipitation diffusion models with reinforcement learning, improper reward design can lead to mode collapse (reduced generation diversity) or reward hacking (optimizing only the score at the expense of physical consistency), and training requires extensive sampling, resulting in high computational cost.
50%UnverifiedThe reverse KL divergence is zero-forcing and tends toward mode seeking, causing severe mode collapse when training diffusion samplers with variational inference.
50%