Hanna Yukhymenko
@ayukh.bsky.social
Statistics MSc @ ETH Zurich
Multilingual LLM training/eval/safety @ SRI lab
ayukh.com
Multilingual LLM training/eval/safety @ SRI lab
ayukh.com
It's just a game to these people
March 8, 2025 at 2:02 AM
It's just a game to these people
Yes, I should have specified it before maybe :)
December 4, 2024 at 2:13 AM
Yes, I should have specified it before maybe :)
This means basically more publicly available materials, yes
For example, more ready-to-use data (e.g. web scraped texts) for LLM fine-tuning, more Ukrainian-native benchmarks for evals etc. This screenshot is from INCLUDE paper by Cohere which has Ukrainian exams in it, thus a new resource for eval🙂
For example, more ready-to-use data (e.g. web scraped texts) for LLM fine-tuning, more Ukrainian-native benchmarks for evals etc. This screenshot is from INCLUDE paper by Cohere which has Ukrainian exams in it, thus a new resource for eval🙂
December 4, 2024 at 2:12 AM
This means basically more publicly available materials, yes
For example, more ready-to-use data (e.g. web scraped texts) for LLM fine-tuning, more Ukrainian-native benchmarks for evals etc. This screenshot is from INCLUDE paper by Cohere which has Ukrainian exams in it, thus a new resource for eval🙂
For example, more ready-to-use data (e.g. web scraped texts) for LLM fine-tuning, more Ukrainian-native benchmarks for evals etc. This screenshot is from INCLUDE paper by Cohere which has Ukrainian exams in it, thus a new resource for eval🙂
Before I have seen many papers claiming Ukrainian language to be low-resource, even though there are ~40 mil UA speakers worldwide, so there should be a lot of proof to that
Since 2022 Ukrainian NLP effort has dramatically increased and the number of Ukrainian texts available online has increased
Since 2022 Ukrainian NLP effort has dramatically increased and the number of Ukrainian texts available online has increased
December 4, 2024 at 2:09 AM
Before I have seen many papers claiming Ukrainian language to be low-resource, even though there are ~40 mil UA speakers worldwide, so there should be a lot of proof to that
Since 2022 Ukrainian NLP effort has dramatically increased and the number of Ukrainian texts available online has increased
Since 2022 Ukrainian NLP effort has dramatically increased and the number of Ukrainian texts available online has increased
