home.social

#nlproc — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #nlproc, aggregated by home.social.

fetched live
  1. I'm recommending more and more to my #NLProc and #ComputationalLiteraryStudies mentees this book by @[email protected]

    Chapter 8 on measurement is mind blowing (and scary!) for many of them 😅

    experimentology.io/008-measurem...

    8  Measurement – Experimentolo...

  2. I'm recommending more and more to my #NLProc and #ComputationalLiteraryStudies mentees this book by @[email protected]

    Chapter 8 on measurement is mind blowing (and scary!) for many of them 😅

    experimentology.io/008-measurem...

    8  Measurement – Experimentolo...

  3. I'm recommending more and more to my #NLProc and #ComputationalLiteraryStudies mentees this book by @[email protected]

    Chapter 8 on measurement is mind blowing (and scary!) for many of them 😅

    experimentology.io/008-measurem...

    8  Measurement – Experimentolo...

  4. I'm recommending more and more to my #NLProc and #ComputationalLiteraryStudies mentees this book by @[email protected]

    Chapter 8 on measurement is mind blowing (and scary!) for many of them 😅

    experimentology.io/008-measurem...

    8  Measurement – Experimentolo...

  5. По поводу вот этого вот взлома структур #HuggingFace моделью от OpenAI очень правильные вопросы задаёт Брюс Шнайер:

    Hypothetical: Imagine that this wasn’t an #OpenAI model. Imagine that it was a Chinese model from a Chinese company. This would be an international crisis.

    Question: Why aren’t we bringing OpenAI up on charges under the Computer Fraud and Abuse Act? How is this different from the Morris Worm? That was also an experiment that escaped the lab.

    schneier.com/blog/archives/202

    #Schneier #БрюсШнайер #LLM #NLProc

  6. По поводу вот этого вот взлома структур #HuggingFace моделью от OpenAI очень правильные вопросы задаёт Брюс Шнайер:

    Hypothetical: Imagine that this wasn’t an #OpenAI model. Imagine that it was a Chinese model from a Chinese company. This would be an international crisis.

    Question: Why aren’t we bringing OpenAI up on charges under the Computer Fraud and Abuse Act? How is this different from the Morris Worm? That was also an experiment that escaped the lab.

    schneier.com/blog/archives/202

    #Schneier #БрюсШнайер #LLM #NLProc

  7. По поводу вот этого вот взлома структур #HuggingFace моделью от OpenAI очень правильные вопросы задаёт Брюс Шнайер:

    Hypothetical: Imagine that this wasn’t an #OpenAI model. Imagine that it was a Chinese model from a Chinese company. This would be an international crisis.

    Question: Why aren’t we bringing OpenAI up on charges under the Computer Fraud and Abuse Act? How is this different from the Morris Worm? That was also an experiment that escaped the lab.

    schneier.com/blog/archives/202

    #Schneier #БрюсШнайер #LLM #NLProc

  8. По поводу вот этого вот взлома структур #HuggingFace моделью от OpenAI очень правильные вопросы задаёт Брюс Шнайер:

    Hypothetical: Imagine that this wasn’t an #OpenAI model. Imagine that it was a Chinese model from a Chinese company. This would be an international crisis.

    Question: Why aren’t we bringing OpenAI up on charges under the Computer Fraud and Abuse Act? How is this different from the Morris Worm? That was also an experiment that escaped the lab.

    schneier.com/blog/archives/202

    #Schneier #БрюсШнайер #LLM #NLProc

  9. По поводу вот этого вот взлома структур #HuggingFace моделью от OpenAI очень правильные вопросы задаёт Брюс Шнайер:

    Hypothetical: Imagine that this wasn’t an #OpenAI model. Imagine that it was a Chinese model from a Chinese company. This would be an international crisis.

    Question: Why aren’t we bringing OpenAI up on charges under the Computer Fraud and Abuse Act? How is this different from the Morris Worm? That was also an experiment that escaped the lab.

    schneier.com/blog/archives/202

    #Schneier #БрюсШнайер #LLM #NLProc

  10. New #NLProc paper! @[email protected] and I propose a constraining method for decoding from LLMs that combines the advantages of grid beam search (fast) and DFA-constraining (not biased towards satisfying easier constraints earlier): "fair grid beam search": www.romanklinger.de/publications...

    romanklinger.de/publications/P...

  11. New #NLProc paper! @[email protected] and I propose a constraining method for decoding from LLMs that combines the advantages of grid beam search (fast) and DFA-constraining (not biased towards satisfying easier constraints earlier): "fair grid beam search": www.romanklinger.de/publications...

    romanklinger.de/publications/P...

  12. New #NLProc paper! @[email protected] and I propose a constraining method for decoding from LLMs that combines the advantages of grid beam search (fast) and DFA-constraining (not biased towards satisfying easier constraints earlier): "fair grid beam search": www.romanklinger.de/publications...

    romanklinger.de/publications/P...

  13. New #NLProc paper! @[email protected] and I propose a constraining method for decoding from LLMs that combines the advantages of grid beam search (fast) and DFA-constraining (not biased towards satisfying easier constraints earlier): "fair grid beam search": www.romanklinger.de/publications...

    romanklinger.de/publications/P...

  14. New #NLProc paper! @[email protected] and I propose a constraining method for decoding from LLMs that combines the advantages of grid beam search (fast) and DFA-constraining (not biased towards satisfying easier constraints earlier): "fair grid beam search": www.romanklinger.de/publications...

    romanklinger.de/publications/P...

  15. This week the decisions for #COLM are out (8th of July) and the #EMNLP author response period has started (July 7-13).

    If you're looking for an alternative #NLProc venue: please consider #INLG! (Either via ARR or direct submissions.)

    INLG focuses on all aspects of language generation, and is an appropriate venue for work on LLMs, rule-based generation, as well as human language production.

    Location: Utrecht (NL), just before EMNLP.

    Submission deadline is July 15.

    See: 2026.inlgmeeting.org

  16. This week the decisions for #COLM are out (8th of July) and the #EMNLP author response period has started (July 7-13).

    If you're looking for an alternative #NLProc venue: please consider #INLG! (Either via ARR or direct submissions.)

    INLG focuses on all aspects of language generation, and is an appropriate venue for work on LLMs, rule-based generation, as well as human language production.

    Location: Utrecht (NL), just before EMNLP.

    Submission deadline is July 15.

    See: 2026.inlgmeeting.org

  17. This week the decisions for #COLM are out (8th of July) and the #EMNLP author response period has started (July 7-13).

    If you're looking for an alternative #NLProc venue: please consider #INLG! (Either via ARR or direct submissions.)

    INLG focuses on all aspects of language generation, and is an appropriate venue for work on LLMs, rule-based generation, as well as human language production.

    Location: Utrecht (NL), just before EMNLP.

    Submission deadline is July 15.

    See: 2026.inlgmeeting.org

  18. This week the decisions for #COLM are out (8th of July) and the #EMNLP author response period has started (July 7-13).

    If you're looking for an alternative #NLProc venue: please consider #INLG! (Either via ARR or direct submissions.)

    INLG focuses on all aspects of language generation, and is an appropriate venue for work on LLMs, rule-based generation, as well as human language production.

    Location: Utrecht (NL), just before EMNLP.

    Submission deadline is July 15.

    See: 2026.inlgmeeting.org

  19. Check NorOLMo 1.0 - our new 13B language model aimed at Norwegian language tasks.

    huggingface.co/HPLT/NorOLMo-13B

    We took English OLMo2-13B as the base model and continually pre-trained it on Norwegian data.
    NorOLMo is the only modern fully open #LLM specifically adapted for #Norwegian. "Fully open" here means that the entirety of the training data and training pipeline is transparent and reproducible - and of course the model weights are available under a permissive license.

    #NLProc #Norsk

  20. Check NorOLMo 1.0 - our new 13B language model aimed at Norwegian language tasks.

    huggingface.co/HPLT/NorOLMo-13B

    We took English OLMo2-13B as the base model and continually pre-trained it on Norwegian data.
    NorOLMo is the only modern fully open #LLM specifically adapted for #Norwegian. "Fully open" here means that the entirety of the training data and training pipeline is transparent and reproducible - and of course the model weights are available under a permissive license.

    #NLProc #Norsk

  21. Check NorOLMo 1.0 - our new 13B language model aimed at Norwegian language tasks.

    huggingface.co/HPLT/NorOLMo-13B

    We took English OLMo2-13B as the base model and continually pre-trained it on Norwegian data.
    NorOLMo is the only modern fully open #LLM specifically adapted for #Norwegian. "Fully open" here means that the entirety of the training data and training pipeline is transparent and reproducible - and of course the model weights are available under a permissive license.

    #NLProc #Norsk

  22. Check NorOLMo 1.0 - our new 13B language model aimed at Norwegian language tasks.

    huggingface.co/HPLT/NorOLMo-13B

    We took English OLMo2-13B as the base model and continually pre-trained it on Norwegian data.
    NorOLMo is the only modern fully open #LLM specifically adapted for #Norwegian. "Fully open" here means that the entirety of the training data and training pipeline is transparent and reproducible - and of course the model weights are available under a permissive license.

    #NLProc #Norsk

  23. Check NorOLMo 1.0 - our new 13B language model aimed at Norwegian language tasks.

    huggingface.co/HPLT/NorOLMo-13B

    We took English OLMo2-13B as the base model and continually pre-trained it on Norwegian data.
    NorOLMo is the only modern fully open #LLM specifically adapted for #Norwegian. "Fully open" here means that the entirety of the training data and training pipeline is transparent and reproducible - and of course the model weights are available under a permissive license.

    #NLProc #Norsk

  24. #Integreat and #TRUST have jointly announced several new PhD projects on AI / ML.

    One of the projects will focus on analyzing ideological bias in multimodal text–image models.

    Supervisors from LTG Oslo include Erik Velldal and Lilja Øvrelid. Other collaborators and co-supervisors include Ali Ramezani-Kebrya, Yael Friedman, and Benjamin Ricaud.
    Come join us!

    integreat.no/about/vacancies/p

    Closing date: 9th August 2026

    #academia #hiring #NLProc #Norway

  25. #Integreat and #TRUST have jointly announced several new PhD projects on AI / ML.

    One of the projects will focus on analyzing ideological bias in multimodal text–image models.

    Supervisors from LTG Oslo include Erik Velldal and Lilja Øvrelid. Other collaborators and co-supervisors include Ali Ramezani-Kebrya, Yael Friedman, and Benjamin Ricaud.
    Come join us!

    integreat.no/about/vacancies/p

    Closing date: 9th August 2026

    #academia #hiring #NLProc #Norway

  26. #Integreat and #TRUST have jointly announced several new PhD projects on AI / ML.

    One of the projects will focus on analyzing ideological bias in multimodal text–image models.

    Supervisors from LTG Oslo include Erik Velldal and Lilja Øvrelid. Other collaborators and co-supervisors include Ali Ramezani-Kebrya, Yael Friedman, and Benjamin Ricaud.
    Come join us!

    integreat.no/about/vacancies/p

    Closing date: 9th August 2026

    #academia #hiring #NLProc #Norway

  27. #Integreat and #TRUST have jointly announced several new PhD projects on AI / ML.

    One of the projects will focus on analyzing ideological bias in multimodal text–image models.

    Supervisors from LTG Oslo include Erik Velldal and Lilja Øvrelid. Other collaborators and co-supervisors include Ali Ramezani-Kebrya, Yael Friedman, and Benjamin Ricaud.
    Come join us!

    integreat.no/about/vacancies/p

    Closing date: 9th August 2026

    #academia #hiring #NLProc #Norway

  28. #Integreat and #TRUST have jointly announced several new PhD projects on AI / ML.

    One of the projects will focus on analyzing ideological bias in multimodal text–image models.

    Supervisors from LTG Oslo include Erik Velldal and Lilja Øvrelid. Other collaborators and co-supervisors include Ali Ramezani-Kebrya, Yael Friedman, and Benjamin Ricaud.
    Come join us!

    integreat.no/about/vacancies/p

    Closing date: 9th August 2026

    #academia #hiring #NLProc #Norway

  29. #INLG2026 submissions are now open!

    Please submit all of your work on Natural Language Generation.

    Submit systems, demos, experiments, surveys, position papers, squibs, linguistic analyses, etc.

    For more information, see: 2026.inlgmeeting.org

    Good to know
    * Organised by @siggen_acl
    * INLG also welcomes dual submissions.
    * The conference is held right before & relatively close to EMNLP.
    * You can either submit directly or commit a reviewed paper from ARR.

    #NLProc #linguistics

  30. #INLG2026 submissions are now open!

    Please submit all of your work on Natural Language Generation.

    Submit systems, demos, experiments, surveys, position papers, squibs, linguistic analyses, etc.

    For more information, see: 2026.inlgmeeting.org

    Good to know
    * Organised by @siggen_acl
    * INLG also welcomes dual submissions.
    * The conference is held right before & relatively close to EMNLP.
    * You can either submit directly or commit a reviewed paper from ARR.

    #NLProc #linguistics

  31. #INLG2026 submissions are now open!

    Please submit all of your work on Natural Language Generation.

    Submit systems, demos, experiments, surveys, position papers, squibs, linguistic analyses, etc.

    For more information, see: 2026.inlgmeeting.org

    Good to know
    * Organised by @siggen_acl
    * INLG also welcomes dual submissions.
    * The conference is held right before & relatively close to EMNLP.
    * You can either submit directly or commit a reviewed paper from ARR.

    #NLProc #linguistics

  32. #INLG2026 submissions are now open!

    Please submit all of your work on Natural Language Generation.

    Submit systems, demos, experiments, surveys, position papers, squibs, linguistic analyses, etc.

    For more information, see: 2026.inlgmeeting.org

    Good to know
    * Organised by @siggen_acl
    * INLG also welcomes dual submissions.
    * The conference is held right before & relatively close to EMNLP.
    * You can either submit directly or commit a reviewed paper from ARR.

    #NLProc #linguistics

  33. This is just the first layer. Over the coming months we'll release further annotation layers on the same corpus — keep an eye here: github.com/GOLEM-lab/GOLEM-multilingual-annotated-corpus 🌱 6/ #DigitalHumanities #ComputationalLiteraryStudies #NLProc

    GitHub - GOLEM-lab/GOLEM-multi...

  34. Interdisciplinary collaboration can be so eye-opening. I just heard a speaker say off-hand "Oh yeah, this is really new work, from 2017" 😅 I think my colleagues in #NLProc wouldn't even touch such historic sources #academicChatter

  35. Interdisciplinary collaboration can be so eye-opening. I just heard a speaker say off-hand "Oh yeah, this is really new work, from 2017" 😅 I think my colleagues in #NLProc wouldn't even touch such historic sources #academicChatter

  36. Interdisciplinary collaboration can be so eye-opening. I just heard a speaker say off-hand "Oh yeah, this is really new work, from 2017" 😅 I think my colleagues in #NLProc wouldn't even touch such historic sources #academicChatter

  37. Interdisciplinary collaboration can be so eye-opening. I just heard a speaker say off-hand "Oh yeah, this is really new work, from 2017" 😅 I think my colleagues in #NLProc wouldn't even touch such historic sources #academicChatter

  38. Interdisciplinary collaboration can be so eye-opening. I just heard a speaker say off-hand "Oh yeah, this is really new work, from 2017" 😅 I think my colleagues in #NLProc wouldn't even touch such historic sources #academicChatter

  39. Maybe I should’ve posted this photo with the post, but I already took it yesterday…

    #Uppsala #DigitalHumanities #NLProc

  40. Maybe I should’ve posted this photo with the post, but I already took it yesterday…

    #Uppsala #DigitalHumanities #NLProc

  41. Maybe I should’ve posted this photo with the post, but I already took it yesterday…

    #Uppsala #DigitalHumanities #NLProc

  42. Maybe I should’ve posted this photo with the post, but I already took it yesterday…

    #Uppsala #DigitalHumanities #NLProc

  43. I’m pleased to inform you that the candidate successfully defended her thesis today and you may now address her as Dr. Lindqvist.

    #Uppsala #DigitalHumanities #NLProc

  44. I’m pleased to inform you that the candidate successfully defended her thesis today and you may now address her as Dr. Lindqvist.

    #Uppsala #DigitalHumanities #NLProc

  45. I’m pleased to inform you that the candidate successfully defended her thesis today and you may now address her as Dr. Lindqvist.

    #Uppsala #DigitalHumanities #NLProc

  46. I’m pleased to inform you that the candidate successfully defended her thesis today and you may now address her as Dr. Lindqvist.

    #Uppsala #DigitalHumanities #NLProc

  47. My master student Lukáš Eigler just defended his thesis (co-supervised with David Hurych from Valeo.ai) 🎉 Congrats!

    #NLP metric validation needs 🐌💰 human judgment data. Our fix: generate synthetic data for metric validation instead. ✅ Tested on MT, QA, summarization.

    To appear at #ACL2026 Student Research Workshop:
    arxiv.org/abs/2603.09403

    #NLProc #MachineLearning

  48. My master student Lukáš Eigler just defended his thesis (co-supervised with David Hurych from Valeo.ai) 🎉 Congrats!

    #NLP metric validation needs 🐌💰 human judgment data. Our fix: generate synthetic data for metric validation instead. ✅ Tested on MT, QA, summarization.

    To appear at #ACL2026 Student Research Workshop:
    arxiv.org/abs/2603.09403

    #NLProc #MachineLearning

  49. RE: fediscience.org/@tschfflr/1166

    I guess that’s the butt end of “just throw in the data, LLM can write the paper no problem”. I really appreciate the intentions behind the ARR system but it doesn’t work. I first realized just how broken it was several years ago as area chair when some of my reviewers were undergraduate students #nlproc

  50. RE: fediscience.org/@tschfflr/1166

    I guess that’s the butt end of “just throw in the data, LLM can write the paper no problem”. I really appreciate the intentions behind the ARR system but it doesn’t work. I first realized just how broken it was several years ago as area chair when some of my reviewers were undergraduate students #nlproc