Gorka Azkune
@gazkune.bsky.social
Associate professor at the University of the Basque Country (UPV/EHU) and researcher at HiTZ Zentroa. Mainly interested on multimodal and multilingual deep learning.
Reposted by Gorka Azkune
Yesterday, Imanol Pradales, the Lehendakari, together with Ibone Bengoetxea, First Vice Lehendakari and Minister of Culture and Language Policy, and Juan Ignacio Pérez Iglesias, Minister of Science, University and Innovation, visited our HiTZ Centre.
www.irekia.euskadi.eus/en/news/1031...
www.irekia.euskadi.eus/en/news/1031...
Irekia - El Lehendakari visita HiTZ, el Centro Vasco de Tecnología de la Lengua
El Lehendakari Imanol Pradales, acompañado de la vicelehendakari primera y consejera de Cultura y Política Lingüística, Ibone Bengoetxea, y del consejero de Ciencia, …
www.irekia.euskadi.eus
July 3, 2025 at 10:02 AM
Yesterday, Imanol Pradales, the Lehendakari, together with Ibone Bengoetxea, First Vice Lehendakari and Minister of Culture and Language Policy, and Juan Ignacio Pérez Iglesias, Minister of Science, University and Innovation, visited our HiTZ Centre.
www.irekia.euskadi.eus/en/news/1031...
www.irekia.euskadi.eus/en/news/1031...
Reposted by Gorka Azkune
Atzo goizean, Imanol Pradales lehendakaria, Ibone Bengoetxea lehen lehendakariorde eta Kultura eta Hizkuntza Politikako sailburua, eta @jipereziglesias.bsky.social @zubsaila.bsky.social sailburua, HiTZ Zentroan izan ziren.
www.irekia.euskadi.eus/eu/news/1031...
www.irekia.euskadi.eus/eu/news/1031...
Irekia - Lehendakariak HiTZ bisitatu du, Hizkuntza Teknologiako Euskal Zentroa
Imanol Pradales Lehendakaria, Ibone Bengoetxea lehen lehendakariorde eta Kultura eta Hizkuntza Politikako sailburuarekin eta Juan Ignacio Pérez Iglesias Zientzia, Unibertsitate eta …
www.irekia.euskadi.eus
July 3, 2025 at 9:56 AM
Atzo goizean, Imanol Pradales lehendakaria, Ibone Bengoetxea lehen lehendakariorde eta Kultura eta Hizkuntza Politikako sailburua, eta @jipereziglesias.bsky.social @zubsaila.bsky.social sailburua, HiTZ Zentroan izan ziren.
www.irekia.euskadi.eus/eu/news/1031...
www.irekia.euskadi.eus/eu/news/1031...
Reposted by Gorka Azkune
Key takeaway: Adding simple structure at inference-time, through image crops and text segments, is a powerful, training-free way to improve Vision-Language Compositionality performance.
Joint work with @Ander Salaberria @eagirre.bsky.social @gazkune.bsky.social @hitz-zentroa.bsky.social
Joint work with @Ander Salaberria @eagirre.bsky.social @gazkune.bsky.social @hitz-zentroa.bsky.social
June 18, 2025 at 11:28 AM
Key takeaway: Adding simple structure at inference-time, through image crops and text segments, is a powerful, training-free way to improve Vision-Language Compositionality performance.
Joint work with @Ander Salaberria @eagirre.bsky.social @gazkune.bsky.social @hitz-zentroa.bsky.social
Joint work with @Ander Salaberria @eagirre.bsky.social @gazkune.bsky.social @hitz-zentroa.bsky.social
Reposted by Gorka Azkune
Our analysis shows that:
1. There is room to improve the quality of extracted text segments.
2. Our method achieves significant performance gains in Winoground's non-trivial instances.
3. Isolated image crops can lose size and quantity information, leaving room for improvement.
1. There is room to improve the quality of extracted text segments.
2. Our method achieves significant performance gains in Winoground's non-trivial instances.
3. Isolated image crops can lose size and quantity information, leaving room for improvement.
June 18, 2025 at 11:28 AM
Our analysis shows that:
1. There is room to improve the quality of extracted text segments.
2. Our method achieves significant performance gains in Winoground's non-trivial instances.
3. Isolated image crops can lose size and quantity information, leaving room for improvement.
1. There is room to improve the quality of extracted text segments.
2. Our method achieves significant performance gains in Winoground's non-trivial instances.
3. Isolated image crops can lose size and quantity information, leaving room for improvement.
Reposted by Gorka Azkune
Why are image crops crucial? 🤔 We found that simply adding text segments isn't enough. The biggest performance gains come when text segments are paired with image crops, proving the power of serial image computing.
June 18, 2025 at 11:28 AM
Why are image crops crucial? 🤔 We found that simply adding text segments isn't enough. The biggest performance gains come when text segments are paired with image crops, proving the power of serial image computing.
Reposted by Gorka Azkune
Our approach is straightforward yet effective:
1. Divide the image into smaller crops.
2. Extract text segments capturing objects, attributes and relations.
3. Use the VLM to find image crops that best fit the text segments.
4. Aggregate matching similarities for the final score.
1. Divide the image into smaller crops.
2. Extract text segments capturing objects, attributes and relations.
3. Use the VLM to find image crops that best fit the text segments.
4. Aggregate matching similarities for the final score.
June 18, 2025 at 11:28 AM
Our approach is straightforward yet effective:
1. Divide the image into smaller crops.
2. Extract text segments capturing objects, attributes and relations.
3. Use the VLM to find image crops that best fit the text segments.
4. Aggregate matching similarities for the final score.
1. Divide the image into smaller crops.
2. Extract text segments capturing objects, attributes and relations.
3. Use the VLM to find image crops that best fit the text segments.
4. Aggregate matching similarities for the final score.
Reposted by Gorka Azkune
#newHitzPaper
Can a simple inference-time approach unlock better Vision-Language Compositionality?🤯
Our latest paper shows how adding structure at inference significantly boosts performance in popular dual-encoder VLMs on different datasets.
Read more: arxiv.org/abs/2506.09691
Can a simple inference-time approach unlock better Vision-Language Compositionality?🤯
Our latest paper shows how adding structure at inference significantly boosts performance in popular dual-encoder VLMs on different datasets.
Read more: arxiv.org/abs/2506.09691
June 18, 2025 at 11:28 AM
#newHitzPaper
Can a simple inference-time approach unlock better Vision-Language Compositionality?🤯
Our latest paper shows how adding structure at inference significantly boosts performance in popular dual-encoder VLMs on different datasets.
Read more: arxiv.org/abs/2506.09691
Can a simple inference-time approach unlock better Vision-Language Compositionality?🤯
Our latest paper shows how adding structure at inference significantly boosts performance in popular dual-encoder VLMs on different datasets.
Read more: arxiv.org/abs/2506.09691
Reposted by Gorka Azkune
While the experiments were not complicated, they required the collaboration of amazing co-authors, many compute hours, and of course, the impressive collaboration of the Basque community that was involved in manually assessing the models on an arena style evaluation.
Thank you!
Thank you!
June 11, 2025 at 6:01 PM
While the experiments were not complicated, they required the collaboration of amazing co-authors, many compute hours, and of course, the impressive collaboration of the Basque community that was involved in manually assessing the models on an arena style evaluation.
Thank you!
Thank you!
Reposted by Gorka Azkune
In this work we face the challenge of developing instruct models for Basque, a low-resource language.
Continue pretraining base models is intuitive, but what about instructed models? We analyze systematically all different approaches to find the best solution.
2/3
Continue pretraining base models is intuitive, but what about instructed models? We analyze systematically all different approaches to find the best solution.
2/3
June 11, 2025 at 6:01 PM
In this work we face the challenge of developing instruct models for Basque, a low-resource language.
Continue pretraining base models is intuitive, but what about instructed models? We analyze systematically all different approaches to find the best solution.
2/3
Continue pretraining base models is intuitive, but what about instructed models? We analyze systematically all different approaches to find the best solution.
2/3
Reposted by Gorka Azkune
Do you know that you can continue pretraining Instructed LLMs without losing their instruction following capabilities?
We did so to teach Basque to Llama models with promising results!
Interestingly, you only need English instructions and target language corpora 🤯
1/3
We did so to teach Basque to Llama models with promising results!
Interestingly, you only need English instructions and target language corpora 🤯
1/3
[4/7]
Key findings:
1️⃣Language corpora is essential: models need exposure to plain Basque text
2️⃣Starting from instructed models beats the standard base→instruct pipeline
3️⃣English-only instructions work well, but combining with Basque instructions yields the most robust models
Key findings:
1️⃣Language corpora is essential: models need exposure to plain Basque text
2️⃣Starting from instructed models beats the standard base→instruct pipeline
3️⃣English-only instructions work well, but combining with Basque instructions yields the most robust models
June 11, 2025 at 6:01 PM
Do you know that you can continue pretraining Instructed LLMs without losing their instruction following capabilities?
We did so to teach Basque to Llama models with promising results!
Interestingly, you only need English instructions and target language corpora 🤯
1/3
We did so to teach Basque to Llama models with promising results!
Interestingly, you only need English instructions and target language corpora 🤯
1/3
Reposted by Gorka Azkune
Ostegunero, HiTZ zentroko kideak biltzen gara gure ikerketen berri emateko HiTZ mintegian.
Aste honetan, bi tesi proiektu aurkeztu dira: Irune Zubiagak" Learning to Judge: Automated Multilingual Evaluation of LLM-Generated Text" eta @blanca-calvo-figue.bsky.social-ek "Critical Questions Generation"
Aste honetan, bi tesi proiektu aurkeztu dira: Irune Zubiagak" Learning to Judge: Automated Multilingual Evaluation of LLM-Generated Text" eta @blanca-calvo-figue.bsky.social-ek "Critical Questions Generation"
June 6, 2025 at 1:36 PM
Ostegunero, HiTZ zentroko kideak biltzen gara gure ikerketen berri emateko HiTZ mintegian.
Aste honetan, bi tesi proiektu aurkeztu dira: Irune Zubiagak" Learning to Judge: Automated Multilingual Evaluation of LLM-Generated Text" eta @blanca-calvo-figue.bsky.social-ek "Critical Questions Generation"
Aste honetan, bi tesi proiektu aurkeztu dira: Irune Zubiagak" Learning to Judge: Automated Multilingual Evaluation of LLM-Generated Text" eta @blanca-calvo-figue.bsky.social-ek "Critical Questions Generation"
Reposted by Gorka Azkune
@oierldl.bsky.social and I are hiring for a post-doc position starting in September within the LUMINOUS project, where we focus on multi-modal LLMs. If you are interested or have questions, feel free to get in contact with us or apply directly at the link below.
link: lnkd.in/d-Zw6Ysh
link: lnkd.in/d-Zw6Ysh
Universidad del País Vasco/Euskal Herriko Unibertsitatea hiring Fully-funded position in Multimodal NLP in Greater Bilbao Metropolitan Area | LinkedIn
Posted 11:36:54 AM. The Hitz Center for Language Technology (www.hitz.eus) at the University of the Basque Country…See this and similar jobs on LinkedIn.
lnkd.in
May 20, 2025 at 12:00 PM
@oierldl.bsky.social and I are hiring for a post-doc position starting in September within the LUMINOUS project, where we focus on multi-modal LLMs. If you are interested or have questions, feel free to get in contact with us or apply directly at the link below.
link: lnkd.in/d-Zw6Ysh
link: lnkd.in/d-Zw6Ysh
Reposted by Gorka Azkune
Apirilaren 30ean, EHUren Zientzia eta Teknologia Fakultateko Paraninfoan, Xabier Arregik "EUSKARAREN ERRONKAK UNIBERTSO DIGITALEAN: Adimen Artifiziala" izeneko mahai-inguruan parte hartuko du.
April 28, 2025 at 6:34 AM
Apirilaren 30ean, EHUren Zientzia eta Teknologia Fakultateko Paraninfoan, Xabier Arregik "EUSKARAREN ERRONKAK UNIBERTSO DIGITALEAN: Adimen Artifiziala" izeneko mahai-inguruan parte hartuko du.
Tomorrow we will have @ebugliarello.bsky.social in our webinar series at @hitz-zentroa.bsky.social . He will talk about vision-language models in the multilingual and multicultural context, exploring evaluation, training and generation. Info here: www.hitz.eus/webinars
HiTZ Chair of AI<: webinar series | HiTZ Zentroa
www.hitz.eus
April 2, 2025 at 8:23 AM
Tomorrow we will have @ebugliarello.bsky.social in our webinar series at @hitz-zentroa.bsky.social . He will talk about vision-language models in the multilingual and multicultural context, exploring evaluation, training and generation. Info here: www.hitz.eus/webinars
Reposted by Gorka Azkune
📊 Ebaluatoiaren lehen 5 egunetako datuak! 📊
775+ erabiltzaile eta 6000+ bidalketa! 🚀
Mila esker guztioi! 💕
Erronka: 20000 bidalketa lortzea apirilaren 2a baino lehen! 🕒
Sartu ebaluatoia.hitz.eus eta egin zure galdera!
775+ erabiltzaile eta 6000+ bidalketa! 🚀
Mila esker guztioi! 💕
Erronka: 20000 bidalketa lortzea apirilaren 2a baino lehen! 🕒
Sartu ebaluatoia.hitz.eus eta egin zure galdera!
Txatbot ebaluatoia
ebaluatoia.hitz.eus
March 24, 2025 at 1:05 PM
📊 Ebaluatoiaren lehen 5 egunetako datuak! 📊
775+ erabiltzaile eta 6000+ bidalketa! 🚀
Mila esker guztioi! 💕
Erronka: 20000 bidalketa lortzea apirilaren 2a baino lehen! 🕒
Sartu ebaluatoia.hitz.eus eta egin zure galdera!
775+ erabiltzaile eta 6000+ bidalketa! 🚀
Mila esker guztioi! 💕
Erronka: 20000 bidalketa lortzea apirilaren 2a baino lehen! 🕒
Sartu ebaluatoia.hitz.eus eta egin zure galdera!
Reposted by Gorka Azkune
Euskal Herriko Unibertsitateak #HITZ bezalako erreferentziazko ikerketa zentroak izateak harrotasunez betetzen gaitu. Eneko Agirre zuzendaria eta bere taldekideekin izan gara goizean.
Jarrai dezazuela bide berriak urratzen hizkuntza eta ahotsa ardatz duen AAren inguruan, @hitz-zentroa.bsky.social.
Jarrai dezazuela bide berriak urratzen hizkuntza eta ahotsa ardatz duen AAren inguruan, @hitz-zentroa.bsky.social.
March 24, 2025 at 11:39 AM
Euskal Herriko Unibertsitateak #HITZ bezalako erreferentziazko ikerketa zentroak izateak harrotasunez betetzen gaitu. Eneko Agirre zuzendaria eta bere taldekideekin izan gara goizean.
Jarrai dezazuela bide berriak urratzen hizkuntza eta ahotsa ardatz duen AAren inguruan, @hitz-zentroa.bsky.social.
Jarrai dezazuela bide berriak urratzen hizkuntza eta ahotsa ardatz duen AAren inguruan, @hitz-zentroa.bsky.social.
Reposted by Gorka Azkune
📢#ebaluatoia
Euskarazko txatbota hobetzen laguntzeko deia egin du @hitz-zentroa.bsky.social -ak. Martxoaren 19tik apirilaren 2ra bitartean galdetu #Latxa ri eta baloratu erantzuna.
Eta parte hartu: www.ehu.eus/eu/web/campu...
Umorea ere baduela konprobatu dugu! Begira👇
Euskarazko txatbota hobetzen laguntzeko deia egin du @hitz-zentroa.bsky.social -ak. Martxoaren 19tik apirilaren 2ra bitartean galdetu #Latxa ri eta baloratu erantzuna.
Eta parte hartu: www.ehu.eus/eu/web/campu...
Umorea ere baduela konprobatu dugu! Begira👇
March 19, 2025 at 12:59 PM
📢#ebaluatoia
Euskarazko txatbota hobetzen laguntzeko deia egin du @hitz-zentroa.bsky.social -ak. Martxoaren 19tik apirilaren 2ra bitartean galdetu #Latxa ri eta baloratu erantzuna.
Eta parte hartu: www.ehu.eus/eu/web/campu...
Umorea ere baduela konprobatu dugu! Begira👇
Euskarazko txatbota hobetzen laguntzeko deia egin du @hitz-zentroa.bsky.social -ak. Martxoaren 19tik apirilaren 2ra bitartean galdetu #Latxa ri eta baloratu erantzuna.
Eta parte hartu: www.ehu.eus/eu/web/campu...
Umorea ere baduela konprobatu dugu! Begira👇
Reposted by Gorka Azkune
Adimen artifizialeko adituen bila zabiltza? HiTZ zentroan badituzu hamaika emakume! #M8 #martxoak8
March 8, 2025 at 3:06 PM
Adimen artifizialeko adituen bila zabiltza? HiTZ zentroan badituzu hamaika emakume! #M8 #martxoak8
Reposted by Gorka Azkune
🧙♂️ New paper 🧙♀️:
Presenting Wicked: a simple automated method to make MCQA benchmarks more challenging. Wicked shook up 18 open-weight LLMs on 6 benchmarks, with up to 19.7% performance drop with direct prompting 🤯
Paper: shorturl.at/1CGq0
Code: shorturl.at/n2nCU
Presenting Wicked: a simple automated method to make MCQA benchmarks more challenging. Wicked shook up 18 open-weight LLMs on 6 benchmarks, with up to 19.7% performance drop with direct prompting 🤯
Paper: shorturl.at/1CGq0
Code: shorturl.at/n2nCU
February 26, 2025 at 11:51 AM
🧙♂️ New paper 🧙♀️:
Presenting Wicked: a simple automated method to make MCQA benchmarks more challenging. Wicked shook up 18 open-weight LLMs on 6 benchmarks, with up to 19.7% performance drop with direct prompting 🤯
Paper: shorturl.at/1CGq0
Code: shorturl.at/n2nCU
Presenting Wicked: a simple automated method to make MCQA benchmarks more challenging. Wicked shook up 18 open-weight LLMs on 6 benchmarks, with up to 19.7% performance drop with direct prompting 🤯
Paper: shorturl.at/1CGq0
Code: shorturl.at/n2nCU
Reposted by Gorka Azkune
Iñigo Alonsok “Improving Fidelity and Table Representation in Table Understanding and Table-to-Text Generation” doktore-tesia defendatu du gaur. Zorionak, Iñigo!
February 27, 2025 at 12:16 PM
Iñigo Alonsok “Improving Fidelity and Table Representation in Table Understanding and Table-to-Text Generation” doktore-tesia defendatu du gaur. Zorionak, Iñigo!
Reposted by Gorka Azkune
German Rigau zuzendariordea Parisen dago "Language Technologies for All" konferentzian parte hartzen.
www.lt4all2025.eu
www.lt4all2025.eu
February 24, 2025 at 9:16 AM
German Rigau zuzendariordea Parisen dago "Language Technologies for All" konferentzian parte hartzen.
www.lt4all2025.eu
www.lt4all2025.eu
Reposted by Gorka Azkune
Bisita izan dugu! Eusko Jaurlaritzako Hizkuntza Politikarako Sailburuordetzako talde zabalarekin, Aitor Aldasoro sailburuordea, Josune Irabien eta Sonia Rodriguez barne, arratsalde ederra pasa genuen zentroa erakusten eta Ikergaitu proiektuko azken emaitzak erakusten https://www.hitz.eus/iker-gaitu/
February 17, 2025 at 9:39 AM
Bisita izan dugu! Eusko Jaurlaritzako Hizkuntza Politikarako Sailburuordetzako talde zabalarekin, Aitor Aldasoro sailburuordea, Josune Irabien eta Sonia Rodriguez barne, arratsalde ederra pasa genuen zentroa erakusten eta Ikergaitu proiektuko azken emaitzak erakusten https://www.hitz.eus/iker-gaitu/
Reposted by Gorka Azkune
Iker Garcia-Ferrerok "Cross-Lingual Transfer for Low-Resource Natural Language Processing" doktore-tesia defendatu du gaur. Zorionak, Iker!
February 12, 2025 at 1:34 PM
Iker Garcia-Ferrerok "Cross-Lingual Transfer for Low-Resource Natural Language Processing" doktore-tesia defendatu du gaur. Zorionak, Iker!
Reposted by Gorka Azkune
Gogoratu otsailaren 17rarte aukera duzuela euskarazko txatbotak ebaluatu eta ikerkuntzan laguntzeko:
ebaluatoia.hitz.eus
Sartu eta parte hartu, erraza eta dibertigarria izateaz gain, sariak ere badaude!
ebaluatoia.hitz.eus
Sartu eta parte hartu, erraza eta dibertigarria izateaz gain, sariak ere badaude!
February 12, 2025 at 8:57 AM
Gogoratu otsailaren 17rarte aukera duzuela euskarazko txatbotak ebaluatu eta ikerkuntzan laguntzeko:
ebaluatoia.hitz.eus
Sartu eta parte hartu, erraza eta dibertigarria izateaz gain, sariak ere badaude!
ebaluatoia.hitz.eus
Sartu eta parte hartu, erraza eta dibertigarria izateaz gain, sariak ere badaude!
Reposted by Gorka Azkune
Atzo, otsailak 11, ETB1eko Egun On saioan aritu zen @eagirre.bsky.social HiTZeko zuzendaria Ebaluatoiari buruz hizketan.
Hemen topatu dezakezue saio osoaren grabazioa, bere parte-hartzea 10:46 inguruan izan zen (grabazioaren ~1:46): www.eitb.eus/eu/nahieran/...
Hemen topatu dezakezue saio osoaren grabazioa, bere parte-hartzea 10:46 inguruan izan zen (grabazioaren ~1:46): www.eitb.eus/eu/nahieran/...
Egun On
Odei Esnaolak eta Irati Errandoneak aurkeztutako goizeko albistegia.
www.eitb.eus
February 12, 2025 at 8:56 AM
Atzo, otsailak 11, ETB1eko Egun On saioan aritu zen @eagirre.bsky.social HiTZeko zuzendaria Ebaluatoiari buruz hizketan.
Hemen topatu dezakezue saio osoaren grabazioa, bere parte-hartzea 10:46 inguruan izan zen (grabazioaren ~1:46): www.eitb.eus/eu/nahieran/...
Hemen topatu dezakezue saio osoaren grabazioa, bere parte-hartzea 10:46 inguruan izan zen (grabazioaren ~1:46): www.eitb.eus/eu/nahieran/...