Unverified50% confidenceFactExact time
The first generation of embedding models was word embeddings (such as Word2Vec and GloVe), which take a single word as input and output a vector, with the limitation of being unable to distinguish the meanings of polysemous words in different contexts.
1
Sources
50%
Confidence
Long-term
Relevance
11/3/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedThe representational approach of LLMs follows the same lineage as traditional word embedding models such as Word2Vec and GloVe, but is capable of capturing more complex features such as context, tone, and implicit semantics.76% similarUnverifiedMethods such as Word2Vec and GloVe produce static embeddings, where the vector for a given word is the same regardless of context, leading to ambiguity problems.74% similarUnverifiedIf a multilingual model's vocabulary has abnormally high fertility on low-resource languages, the problem may not lie in vocabulary size but in the construction objective.72% similarUnverifiedVector embedding models rely on general language distributions; internal enterprise abbreviations, project codenames, and personal name shortcuts are nearly absent from pre-training corpora, causing semantic similarity calculations to be inaccurate.72% similarUnverifiedWord2Vec, proposed in 2013, and GloVe, proposed in 2014, can only generate fixed vectors for individual words and cannot handle the polysemy problem.72% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1131418API
curl https://kongchang.com/api/v1/knowledge/claims/1131418MCP
get_claim(id=1131418)