Direct Prediction Paradigm
A working paradigm for large language models in which the model generates answers directly through token-by-token completion. This approach is essentially pattern-matching continuation based on the current context and lacks a systematic planning mechanism. The terminology is borrowed from Vladimir Vapnik's statistical learning theory.
Timeline (last 90 days)
The transductive paradigm refers to a model directly generating answers in a token-by-token completion manner, continuing word by word based on the existing context.
In the transductive paradigm, computation is evenly distributed across each output token; in the inductive paradigm, the model can invest large amounts of computation in reasoning tokens while the final answer tokens are relatively lightweight.
Programming tasks are the area where the transductive paradigm is most prone to errors, as they are multi-step, strongly logical tasks where errors accumulate across subsequent tokens.
All Facts (3)
The transductive paradigm refers to a model directly generating answers in a token-by-token completion manner, continuing word by word based on the existing context.
50%UnverifiedIn the transductive paradigm, computation is evenly distributed across each output token; in the inductive paradigm, the model can invest large amounts of computation in reasoning tokens while the final answer tokens are relatively lightweight.
50%UnverifiedProgramming tasks are the area where the transductive paradigm is most prone to errors, as they are multi-step, strongly logical tasks where errors accumulate across subsequent tokens.
50%