My latest work on transliteration has been accepted to @emnlpmeeting.bsky.social Findings.
We developed Arabic→Maltese transliteration systems and evaluated it as a tool for data augmentation for #Maltese #NLP.
📄 Preprint: arxiv.org/abs/2509.12853
@nyhabash.bsky.social
We developed Arabic→Maltese transliteration systems and evaluated it as a tool for data augmentation for #Maltese #NLP.
📄 Preprint: arxiv.org/abs/2509.12853
@nyhabash.bsky.social
September 18, 2025 at 10:58 AM
My latest work on transliteration has been accepted to @emnlpmeeting.bsky.social Findings.
We developed Arabic→Maltese transliteration systems and evaluated it as a tool for data augmentation for #Maltese #NLP.
📄 Preprint: arxiv.org/abs/2509.12853
@nyhabash.bsky.social
We developed Arabic→Maltese transliteration systems and evaluated it as a tool for data augmentation for #Maltese #NLP.
📄 Preprint: arxiv.org/abs/2509.12853
@nyhabash.bsky.social
I'll be presenting our work on benchmarking #LLMs on #Maltese at @aclmeeting.bsky.social 🇦🇹
We show how LLMs struggle with Maltese, performing worse than smaller fine-tuned models, and how fine-tuning remains viable and more efficient for low-resource languages.
arxiv.org/abs/2506.043...
We show how LLMs struggle with Maltese, performing worse than smaller fine-tuned models, and how fine-tuning remains viable and more efficient for low-resource languages.
arxiv.org/abs/2506.043...
July 27, 2025 at 7:22 AM
I'll be presenting our work on benchmarking #LLMs on #Maltese at @aclmeeting.bsky.social 🇦🇹
We show how LLMs struggle with Maltese, performing worse than smaller fine-tuned models, and how fine-tuning remains viable and more efficient for low-resource languages.
arxiv.org/abs/2506.043...
We show how LLMs struggle with Maltese, performing worse than smaller fine-tuned models, and how fine-tuning remains viable and more efficient for low-resource languages.
arxiv.org/abs/2506.043...
Reposted by Kurt
Court filings show Meta staffers discussed using copyrighted content for AI training
Court filings show Meta staffers discussed using copyrighted content for AI training
For years, Meta employees have internally discussed using copyrighted works obtained through legally questionable means to train the company’s AI models, according to court documents unsealed on Thursday. The documents were submitted by plaintiffs in the…
tcrn.ch
February 21, 2025 at 11:16 PM
Court filings show Meta staffers discussed using copyrighted content for AI training
