Overthinking
A phenomenon in which reasoning models generate unnecessarily long chain-of-thought sequences for simple tasks, resulting in wasted tokens and increased latency without a corresponding improvement in answer accuracy.
Core Facts
Timeline (last 90 days)
RareTrap is sampling-based: it randomly samples latent variables in a continuous latent space and decodes them into prompts, searching for samples that can trigger abnormal behaviors such as overthinking; ThinkTrap is optimization-based: it uses gradient estimation or evolutionary strategies to optimize latent variables so that the model produces high-risk internal reasoning.
Models may fall into the trap of overthinking, as longer reasoning time does not always lead to better results.
All Facts (2)
Models may fall into the trap of overthinking, as longer reasoning time does not always lead to better results.
70%UnverifiedRareTrap is sampling-based: it randomly samples latent variables in a continuous latent space and decodes them into prompts, searching for samples that can trigger abnormal behaviors such as overthinking; ThinkTrap is optimization-based: it uses gradient estimation or evolutionary strategies to optimize latent variables so that the model produces high-risk internal reasoning.
50%