diff options
| author | Jonas Knobloch <jonas.knobloch@t-online.de> | 2026-05-04 15:39:45 +0200 |
|---|---|---|
| committer | Jonas Knobloch <jonas.knobloch@t-online.de> | 2026-05-04 15:39:45 +0200 |
| commit | 29c52bf0f0981d4ea8e12e86237365da2b225ef7 (patch) | |
| tree | 6ac95a78f2d0ad59fbafa760c216835f38f462e1 /research/sander/analyze.go | |
| parent | 281e82eed24c1f871e868afbd23cb6811333c2f4 (diff) | |
Add parameter to set hidden dimension
Diffstat (limited to 'research/sander/analyze.go')
| -rw-r--r-- | research/sander/analyze.go | 10 |
1 files changed, 5 insertions, 5 deletions
diff --git a/research/sander/analyze.go b/research/sander/analyze.go index c58195d..6abc9b8 100644 --- a/research/sander/analyze.go +++ b/research/sander/analyze.go @@ -6,11 +6,10 @@ import ( "fmt" ) -var ( - sqlSimilarity = ` +var sqlSimilarity = ` CREATE TABLE similarity_buffer AS WITH centroid AS ( - SELECT list(avg_val ORDER BY idx)::FLOAT[768] AS vec + SELECT list(avg_val ORDER BY idx)::FLOAT[%d] AS vec FROM ( SELECT idx, avg(val) AS avg_val FROM embeddings @@ -22,7 +21,6 @@ var ( SELECT e.token_id, array_distance(e.embedding, c.vec) AS distance FROM embeddings e, centroid c ` -) func (e *Experiment) Analyze(db *sql.DB) error { ctx := context.Background() @@ -41,7 +39,9 @@ func (e *Experiment) Analyze(db *sql.DB) error { return err } - if _, err := tx.ExecContext(ctx, sqlSimilarity); err != nil { + query := fmt.Sprintf(sqlSimilarity, e.hiddenDim) + + if _, err := tx.ExecContext(ctx, query); err != nil { return fmt.Errorf("similarity: %w", err) } |
