Post
38
πΉπ· **You tried to break our Turkish NER model.**
And that's exactly what we wanted.
After sharing Werea-TR-NER, we asked the Hugging Face community to challenge our ~110M parameter model with difficult Turkish sentences.
Some worked.
Some exposed weaknesses.
And that's more valuable than pretending a benchmark score tells the whole story.
Our published WikiANN-tr result:
**91.7% Entity F1**
π€ PERSON β 94.2%
π LOCATION β 91.4%
π’ ORGANIZATION β 89.2%
But now we want to go further.
π₯ **ROUND 2**
Send me a Turkish sentence designed specifically to break the model.
Ambiguous names.
Companies that sound like people.
Locations hidden inside organization names.
Turkish suffixes.
Slang.
Anything nasty.
**Try to make it fail.**
I'll collect the hardest examples and use them to build a public adversarial Turkish NER evaluation set.
π€ Model:
Werea-co/Werea-TR-NER
πΉπ· Werea:
Werea-co
Follow me if you want to see whether the community can break it β and what we build from the failures.
#TurkishNLP #HuggingFace #NER #OpenSourceAI
And that's exactly what we wanted.
After sharing Werea-TR-NER, we asked the Hugging Face community to challenge our ~110M parameter model with difficult Turkish sentences.
Some worked.
Some exposed weaknesses.
And that's more valuable than pretending a benchmark score tells the whole story.
Our published WikiANN-tr result:
**91.7% Entity F1**
π€ PERSON β 94.2%
π LOCATION β 91.4%
π’ ORGANIZATION β 89.2%
But now we want to go further.
π₯ **ROUND 2**
Send me a Turkish sentence designed specifically to break the model.
Ambiguous names.
Companies that sound like people.
Locations hidden inside organization names.
Turkish suffixes.
Slang.
Anything nasty.
**Try to make it fail.**
I'll collect the hardest examples and use them to build a public adversarial Turkish NER evaluation set.
π€ Model:
Werea-co/Werea-TR-NER
πΉπ· Werea:
Follow me if you want to see whether the community can break it β and what we build from the failures.
#TurkishNLP #HuggingFace #NER #OpenSourceAI