Collected molecules will appear here. Add from search or explore.
A language-independent machine translation evaluation metric that incorporates multiple factors like length penalty, position difference, and word alignment to score translation quality against references.
Defensibility
stars
9
forks
2
hLEPOR is a legacy academic project originally designed to improve upon basic metrics like BLEU by adding linguistic factors. With only 9 stars and no activity in nearly 13 years (4636 days), it is effectively a 'zombie' repository. In the current competitive landscape, machine translation evaluation has shifted almost entirely to neural-based metrics like COMET, BERTScore, and BLEURT, or 'LLM-as-a-judge' paradigms using GPT-4o. The project lacks any modern moat, community support, or technical edge that would prevent it from being entirely bypassed by modern standard libraries or frontier lab evaluation suites. It serves primarily as a historical reference for MT research rather than a viable tool for contemporary production pipelines.
TECH STACK
INTEGRATION
cli_tool
READINESS