CorpusGuard audit

examples/demo.jsonl

64/100 ยท D

Metrics

Rows5
Valid Rows5
Invalid Rows0
Empty Outputs0
Short Outputs0
Exact Duplicate Groups1
Exact Duplicate Rows1
Near Duplicate Pairs0
Pii Hits1
High Repetition Rows0
Latin Chars325
Cyrillic Chars0
Other Letter Chars0
Avg Input Chars36.6
Avg Output Chars49.4
P95 Output Chars101
Leakage Matches0

Recommendations

  • Deduplicate repeated examples to reduce memorization bias.
  • Review and redact potential PII before publishing or training.

Findings

SeverityCodeRowMessage
errorpii4Potential sensitive data: email, phone
warningexact_duplicate2Exact duplicate across rows [2, 5]