Unverified50% confidenceFactExact time
Most existing generative recommendation methods only use the standard causal cross-entropy objective to optimize semantic ID prediction, ignoring the hierarchical structure of the semantic IDs themselves.
1
Sources
50%
Confidence
Long-term
Relevance
10/14/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedStandard causal cross-entropy treats the token at every position in the sequence equally, without distinguishing which level of the semantic ID hierarchy a token is at, flattening a prediction task that could be decomposed into a tree structure into mutually independent classification problems.77% similarUnverifiedThe semantic entropy method does not rely on the internal structure of a specific model; it infers uncertainty through semantic consistency at the output level, giving it relatively strong generalization ability.74% similarVerifiedThe semantic entropy method is technically implemented by obtaining candidate answers through multiple samplings, using an NLI model to determine semantic equivalence for clustering, and then computing entropy over the semantic categories.74% similarUnverifiedThe objective function for pre-training is typically the cross-entropy loss in causal language modeling, i.e., maximizing the probability of correctly predicting the next token given all preceding tokens.70% similarUnverifiedThe quality of semantic IDs directly determines the upper bound of generative recommendation, because they are both the input representation and the generation target.65% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1023833API
curl https://kongchang.com/api/v1/knowledge/claims/1023833MCP
get_claim(id=1023833)