Model card

astris-embeddings-v1

Multilingual transformer sentence embeddings (intfloat e5) with an offline character-n-gram hashing fallback when the transformer is unavailable. L2-normalized vectors power semantic skill/title matching and approximate-nearest-neighbor candidate recall.

At a glance

Kindembedding
Version1.0.0
StatusActive — contributing to scores
Ensemble weight0
Trained atNot trained yet
Trained onn/a
ArtifactNo artifact (heuristic or pre-training)

Metrics

{
  "backend": "e5_transformer",
  "fallback": "char_ngram_hash",
  "validation": "synthetic_seed",
  "real_cohort": "pending",
  "normalization": "l2"
}

Notes

Capability layer, not a retention voter — weight 0. Degrades to the hashing fallback offline; see api/ml/embeddings.py.