Aravind Putrevu
Devrel
LEARN, CONNECT, BUILD
Microsoft Reactor
Partecipa a Microsoft Reactor e interagisci con gli sviluppatori live
Sei pronto per iniziare a usare l''intelligenza artificiale e le tecnologie più recenti? Microsoft Reactor fornisce eventi, formazione e risorse della community per aiutare sviluppatori, imprenditori e startup a sviluppare la tecnologia di intelligenza artificiale e altro ancora. Unisciti a noi.
LEARN, CONNECT, BUILD
Microsoft Reactor
Partecipa a Microsoft Reactor e interagisci con gli sviluppatori live
Sei pronto per iniziare a usare l''intelligenza artificiale e le tecnologie più recenti? Microsoft Reactor fornisce eventi, formazione e risorse della community per aiutare sviluppatori, imprenditori e startup a sviluppare la tecnologia di intelligenza artificiale e altro ancora. Unisciti a noi.
Přejít zpět
Improving Large Language Model by Systematically Improving its Data
26 února, 2024 | 12:00 odp. - 1:00 odp. (UTC) Coordinated Universal Time
- Formát:
Live Stream
Téma: Data Science & Machine Learning
Jazyk: italiano
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.
Přednášející
Související události
Níže uvedené události vás mohou také zajímat. Nezapomeňte navštívit naše Domovská stránka reaktoru zobrazení všech dostupných událostí.
-
Building Context and Reasoning for Agentic Systems with Howso
Building Context and Reasoning for Agentic Systems with Howso
Formát:
Live Stream
Téma: Data Science & Machine Learning
Jazyk: italiano