Orai NLP Teknologiak
@orainlp.bsky.social
· NLP research and technological solutions for industry and companies
· NLP ikerketa eta soluzio teknologikoak industriaren eta enpresen zerbitzura
· Investigación y soluciones tecnológicas NLP al servicio de la industria y las empresas
· NLP ikerketa eta soluzio teknologikoak industriaren eta enpresen zerbitzura
· Investigación y soluciones tecnológicas NLP al servicio de la industria y las empresas
📈 This work investigates the effectiveness of #SLMs with up to one billion parameters (sub-1B) for #NLP tasks in low-resource languages, focusing on Basque. We analyze optimal training strategies by comparing training from scratch and continual pre-training using state-of-the-art SLM architectures.
November 10, 2025 at 7:33 AM
📈 #DIPLomA: Efficient Adaptation of Instructed LLMs to Low-Resource Languages via Post-Training Delta Merging
❓How open-weight instruction-tuned large language models ( #LLMs ) can be efficiently adapted to low-resource languages without requiring costly large-scale post-training.
❓How open-weight instruction-tuned large language models ( #LLMs ) can be efficiently adapted to low-resource languages without requiring costly large-scale post-training.
November 4, 2025 at 12:53 PM
More info ⤵️⤵️
www.orai.eus/en/blog/new-...
www.orai.eus/en/blog/new-...
A new chatbot in Basque that can be installed on in-house servers: Kimu | Orai
Orai has created a lightweight model that works well in Basque and can easily be adapted to the needs of companies and institutions. It can be used for a wide range of tasks: to answer queries about d...
www.orai.eus
October 14, 2025 at 11:52 AM
More info ⤵️⤵️
www.orai.eus/en/blog/new-...
www.orai.eus/en/blog/new-...
✨ Results:
Significant improvements in #Basque for instruction following, safety, and linguistic proficiency👏
🔗 kimu.orai.eus
Significant improvements in #Basque for instruction following, safety, and linguistic proficiency👏
🔗 kimu.orai.eus
October 14, 2025 at 8:21 AM
✨ Results:
Significant improvements in #Basque for instruction following, safety, and linguistic proficiency👏
🔗 kimu.orai.eus
Significant improvements in #Basque for instruction following, safety, and linguistic proficiency👏
🔗 kimu.orai.eus
🧠 The Basque adaptation was carried out using #ZelaiHaundi, a 500M-word corpus, without relying on any real or synthetic instructions or preferences for post-training.
We then combined the foundational adapted model with the instruction-tuned base model.
We then combined the foundational adapted model with the instruction-tuned base model.
October 14, 2025 at 8:21 AM
🧠 The Basque adaptation was carried out using #ZelaiHaundi, a 500M-word corpus, without relying on any real or synthetic instructions or preferences for post-training.
We then combined the foundational adapted model with the instruction-tuned base model.
We then combined the foundational adapted model with the instruction-tuned base model.