The goal is clearly to better understand all available information (images, texts, etc.). Google reminds that it is thanks to transformers.
RESEARCH
-
Nicolas Papernot at SaTML Conference on Trustworthy Machine Learning
By
–
I'll be at SaTML Thursday and Friday (sadly missing Day 1 today). Looking forward to a conference focused entirely on trustworthy ML! Please find me and say hi if you're around. @satml_conf https://
x.com/NicolasPaperno
/NicolasPapernot/status/1613579758792572928
… -
Keynote Speaker at LEAP2023 Conference in Riyadh
By
–
I was honored to give two keynotes at the #LEAP2023 conference in Riyadh, where I joined over 200,000 attendees to explore the latest technologies and advancements… Just giant! https://
lnkd.in/gqSSk2AP @deepfestai #DeepFest2023 @LEAPandInnovate -
ICML 2022 Paper on Robustness in Machine Learning Research
By
–
This seems like cool work! I'm a big fan of problems related to robustness. Have you seen this paper by #ICML2022 paper by @andrew_ilyas @smsampark @logan_engstrom @gpoleclerc @aleks_madry
? Covers a similar problem. https://
arxiv.org/abs/2202.00622 -
Language Models Near AGI: Missing Autonomy and Multimodal Integration
By
–
Jazykovy model je uz take male AGI. Dobre rozumie nasmu svetu, vie riesit problemy atd. Co mu chyba (da sa doplnit) je autonomia, schopnost sa ucit zo skusenosti, dalsie modality (video, real fyzika), schopnost pouzivat externe nastroje, komunikacia s inymi jazykovymi modelmi.
-
DAGGER vs RL: Feedback Methods for LLM Training
By
–
While DAGGER is a great idea to enable Feedback for LLMs (eg chat) it is not a replacement for RL because RL opens up room for different forms of feedback (eg preferences). However, as a teacher I would advise careful measurement of the contribution of each to the final metric.
-
DAGGER Counterfactual Teaching Method for LLM Training
By
–
DAGGER is a form of counterfactual teaching as explained in https://
arxiv.org/abs/2110.10819 – Note that it is the student who always acts. The teacher only provides corrections, which are used to minimise the LLM loss directly. Note however that this imitation IS NOT supervised learning. -
DAGGER: Imitation Learning Alternative to Reinforcement Learning
By
–
People are asking if there are alternatives to RL in RLHF. Yes, imitation with DAGGER (tutorial: https://
ri.cmu.edu/publications/a
n-invitation-to-imitation/
… ). The user provides feedback with corrections, e.g. when the agent says “that” the user tells the agent that instead of saying “that”, it should say “this”. -
ChatGPT future predictions confirmed one week later insights
By
–
En el directo de la semana pasada nos preguntamos cómo sería el futuro de herramientas como ChatGPT y ya 7 días después algunas cosas se han demostrado ciertas. Hay info super relevante para entender lo que va a venir
-
Moûsai: Cascading Latent Diffusion for High-Quality Text-to-Music
By
–
A new AI research presents a cascading latent diffusion approach called Moûsai that simplifies high-quality text-to-music at 48kHz. To read how this method can generate minutes of high-quality music in real-time, visit: https://
bit.ly/3JPoyt6 @jeevprabnivash @_DigitalIndia