How smart is an LLM really?
Let's analyze it! Causally.
1/n
#Causality #CausalNLP #MachineLearning #causalinference #causalAI #CausalSky
Let's analyze it! Causally.
1/n
#Causality #CausalNLP #MachineLearning #causalinference #causalAI #CausalSky
January 31, 2024 at 9:57 AM
How smart is an LLM really?
Let's analyze it! Causally.
1/n
#Causality #CausalNLP #MachineLearning #causalinference #causalAI #CausalSky
Let's analyze it! Causally.
1/n
#Causality #CausalNLP #MachineLearning #causalinference #causalAI #CausalSky
FallacyES is the first dataset dedicated to fallacies in Spanish. It combines prototypical fallacies from the Logic dataset (github.com/causalNLP/lo...) and real-world examples from comments on meneame.net. Available in idus.us.es/items/170398...
Dataset FallacyES
idUS es el repositorio institucional de la Universidad de Sevilla que da acceso a los textos completos de las publicaciones científicas, datos de investigación, códigos y metodologías.
idus.us.es
January 24, 2025 at 8:41 AM
FallacyES is the first dataset dedicated to fallacies in Spanish. It combines prototypical fallacies from the Logic dataset (github.com/causalNLP/lo...) and real-world examples from comments on meneame.net. Available in idus.us.es/items/170398...
We evaluated 19 LLMs:
🚫 Most don’t align with human values
✅ Surprisingly, no strong “language inequality”: low-resource languages show similar alignment with human values as high-resource ones
💻Code: github.com/causalNLP/mu...
🙌 Huge thanks to our amazing co-authors and institutional support!
🚫 Most don’t align with human values
✅ Surprisingly, no strong “language inequality”: low-resource languages show similar alignment with human values as high-resource ones
💻Code: github.com/causalNLP/mu...
🙌 Huge thanks to our amazing co-authors and institutional support!
GitHub - causalNLP/MultiTP
Contribute to causalNLP/MultiTP development by creating an account on GitHub.
github.com
April 14, 2025 at 4:05 PM
We evaluated 19 LLMs:
🚫 Most don’t align with human values
✅ Surprisingly, no strong “language inequality”: low-resource languages show similar alignment with human values as high-resource ones
💻Code: github.com/causalNLP/mu...
🙌 Huge thanks to our amazing co-authors and institutional support!
🚫 Most don’t align with human values
✅ Surprisingly, no strong “language inequality”: low-resource languages show similar alignment with human values as high-resource ones
💻Code: github.com/causalNLP/mu...
🙌 Huge thanks to our amazing co-authors and institutional support!
