calibrate_threshold
Auto-calibrate vector search threshold by sampling random memory pairs and setting threshold above the null cosine distribution mean, rejecting unrelated pairs. Uses percentile or zscore method.
Instructions
Auto-calibrate the vector search threshold from the null (random-pair) cosine distribution. Samples random memory pairs and places the threshold ABOVE the null mean so unrelated pairs are rejected. method='percentile' (default) uses a quantile of the null distribution (robust to anisotropic models such as bge-m3); method='zscore' uses mean + z*std. No labels used, purely statistical. Adapts to both embedding model and corpus characteristics.
Input Schema
TableJSON Schema
| Name | Required | Description | Default |
|---|---|---|---|
| method | No | 'percentile' (default), 'zscore', or 'separation' (two-population, learns the operating point from null vs nearest-neighbour positives) | |
| agent_id | Yes | Agent ID whose memories to sample | |
| z_factor | No | Z-score multiplier for method='zscore' (default: 1.0, higher = stricter) | |
| percentile | No | Null-distribution quantile for method='percentile' (default: 0.95, higher = stricter) | |
| sample_size | No | Number of embeddings to sample (default: 200) |