A contrastive learning penalty that uses synthetic queries for negative documents improves retrieval nDCG by about 4 points on average over the BGE-M3 baseline across Korean, Persian, and Hindi.
Is this document relevant?
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.IR 1years
2024 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Efficient fine-tuning methodology of text embedding models for information retrieval: contrastive learning penalty (clp)
A contrastive learning penalty that uses synthetic queries for negative documents improves retrieval nDCG by about 4 points on average over the BGE-M3 baseline across Korean, Persian, and Hindi.