Skip to content

v1.3.2 — Auto-Config Robustness

Choose a tag to compare

View all tags
@benzsevern benzsevern released this 03 Apr 23:27
v1.3.2
852c4a4

Fixes

Cross-Source Blocking Overlap Detection

Auto-config now checks whether blocking key values overlap between sources when __source__ column is present. Keys with zero overlap are skipped with a warning.

Before: DBLP-ACM dataset — auto-config picked venue for blocking, but DBLP uses "VLDB" while ACM uses "Very Large Data Bases". Zero overlap = zero cross-source matches.

After: venue skipped with warning, year selected instead. Pipeline completes with 92% F1.

Embedding Scorer Fallback

When embedding models fail to load (HuggingFace auth, Vertex AI quota, missing dependency, CUDA OOM), the scorer now falls back to token_sort instead of crashing. Both _fuzzy_score_matrix and record_embedding paths are covered.

Full Changelog: v1.3.1...v1.3.2

Assets 2
Loading