Aravind Putrevu
Devrel
ONTDEKKEN, VERBINDEN, GROEIEN
Klaar om aan de slag te gaan met AI? Microsoft Reactor biedt evenementen, training en community-resources om start-ups, ondernemers en ontwikkelaars te helpen hun volgende bedrijf op basis van AI-technologie uit te bouwen. Doe mee!
ONTDEKKEN, VERBINDEN, GROEIEN
Klaar om aan de slag te gaan met AI? Microsoft Reactor biedt evenementen, training en community-resources om start-ups, ondernemers en ontwikkelaars te helpen hun volgende bedrijf op basis van AI-technologie uit te bouwen. Doe mee!
26 februari, 2024 | 12:00 p.m. - 1:00 p.m. (UTC) Coordinated Universal Time
Onderwerp: Datawetenschap & Machine Learning
Language: Engels
Labeled data powers AI/ML in the enterprise, but real-world datasets have been found to contain between 7-50% annotation errors. Imperfectly labelled text data hampers ML models' training (and evaluation) across tasks like intent recognition, entity recognition, and sequence generation. Although pretrained LLMs are equipped with a lot of world knowledge, their performance is adversely affected by noisy training data (as noted by OpenAI).
In this talk, we illustrate data-centric techniques to mitigate the effect of label noise without changing any code related to model architecture, hyperparameters, or training. These data quality improvement techniques should thus remain applicable even for future advanced LLMs like GPT-10.
Sprekers
Voor vragen kunt u contact met ons opnemen op reactor@microsoft.com