Skip to content
fialr

similar

fialr similar <file> [options]

Find files in the corpus that are semantically similar to the given file, ranked by cosine similarity. Uses stored vector embeddings.

Arguments

Section titled “Arguments”
ArgumentDescription
filePath to the reference file

Options

Section titled “Options”
OptionDescription
--limit NMaximum results (default: 10)
--threshold NMinimum similarity score, 0.0–1.0 (default: 0.0)
-v, --verboseShow debug output

What it does

Section titled “What it does”

Retrieves the stored embedding for the given file and compares it against all other embeddings in the corpus using cosine similarity. Results are ranked by similarity score and returned with file metadata.

If the reference file does not have an embedding, run fialr embed first.

Output

Section titled “Output”
similar ~/Documents/2024-q1_report.pdf


  SIMILAR   8 files (threshold: 0.0)
────────────────────────────────────────────────────────
  0.94  financial/2024-q2_report.pdf
          quarterly, financial, revenue
  0.91  financial/2024-q3_report.pdf
          quarterly, financial, revenue
  0.87  financial/2023-annual_report.pdf
          annual, financial, revenue
  0.82  work/quarterly_review_template.docx
          quarterly, template, review
  0.74  financial/2024-budget_forecast.xlsx
          budget, forecast, financial

Examples

Section titled “Examples”
Terminal window
# Find files similar to a specific document
fialr similar ~/Documents/contract.pdf


# Limit to top 5 results above 0.8 similarity
fialr similar ~/Documents/contract.pdf --limit 5 --threshold 0.8

See also

Section titled “See also”
  • embed — generate embeddings for the corpus
  • search — semantic search with --semantic flag
  • Deduplication guide — embedding-based near-duplicate detection