Open sadettindemirel opened 4 months ago
Hi there! Our multilingual classifier is based on xlm-roberta, so the model potentially work on any languages included in the xlm-roberta training corpus. However, we fine tuned it on a classification dataset, which is limited to a few languages https://huggingface.co/datasets/cardiffnlp/tweet_sentiment_multilingual, and Turkish is not in the fine tuning dataset.
With that being said, you can’t expect it to do a great job but could work to some extent.
On Thu, 23 May 2024 at 22:46, Sadettin Demirel @.***> wrote:
This is not an issue but a question. I discovered that the TweetNLP demo can classify multilingual texts, including Turkish. Can I classify Turkish texts with this version? I haven't tried it yet, so my question is whether the TweetNLP multilingual support is exclusive to the TweetNLP Demo.
— Reply to this email directly, view it on GitHub https://github.com/cardiffnlp/tweetnlp/issues/23, or unsubscribe https://github.com/notifications/unsubscribe-auth/AEEXCDBOLST72ES5EU5HPYLZDXXJTAVCNFSM6AAAAABIFWJ2WGVHI2DSMVQWIX3LMV43ASLTON2WKOZSGMYTEOJXGQZTAMQ . You are receiving this because you are subscribed to this thread.Message ID: @.***>
Hi, Then I will be looking for other models that are fine tuned for Turkish texts. Thank you for the prompt response.
This is not an issue but a question. I discovered that the TweetNLP demo can classify multilingual texts, including Turkish. Can I classify Turkish texts with this version? I haven't tried it yet, so my question is whether the TweetNLP multilingual support is exclusive to the TweetNLP Demo.