For English-to-Czech MT evaluation, embeddings that win intrinsic semantic similarity benchmarks (SimCSE) perform worst after fine-tuning, while intrinsically poor models (XLM-R, FERNET) perform best.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Intrinsic vs. Extrinsic Evaluation of Czech Sentence Embeddings: Semantic Relevance Doesn't Help with MT Evaluation
For English-to-Czech MT evaluation, embeddings that win intrinsic semantic similarity benchmarks (SimCSE) perform worst after fine-tuning, while intrinsically poor models (XLM-R, FERNET) perform best.