Embeddings
An embedding maps discrete or complex objects into vectors whose geometry can support learned comparisons. Distance is meaningful only relative to a model, training objective, normalization choice, and downstream task.
Overview. An embedding maps discrete or complex objects into vectors whose geometry can support learned comparisons.
Context. Distance is meaningful only relative to a model, training objective, normalization choice, and downstream task.
Claims and evidence
Claim ledger, 2An embedding maps discrete or complex objects into vectors whose geometry can support learned comparisons.
Distance is meaningful only relative to a model, training objective, normalization choice, and downstream task.