Published May 26, 2020
Rethinking Model Size: Train Large, Then Compress with Joseph Gonzalez - #378
Joseph Gonzalez from UC Berkeley delves into the complexities of training and compressing large AI models, exploring strategies to enhance efficiency, interpretability, and transparency. This episode covers the evolution of research on transformers, touching on the balance between model and batch size, and the economic impacts of model scaling.
Topics covered
Popular Clips
Episode Highlights
Related Episodes
Deep Learning, Transformers, and the Consequences of Scale with Oriol Vinyals - #546
Answers 383 questions
Systems and Software for Machine Learning at Scale with Jeff Dean - #124
Answers 383 questions
Scaling TensorFlow at LinkedIn with Jonathan Hung - #314
Answers 383 questions
Vector Quantization for NN Compression with Julieta Martinez - #498
Answers 383 questions
Generating Labeled Training Data for Your ML/AI Models with Angie Hugeback - #6
Answers 383 questions
Scaling AI for the Enterprise with Mazin Gilbert - #78
Answers 383 questions
The Case for Hardware-ML Model Co-design with Diana Marculescu - #391
Answers 383 questions
Towards Improved Transfer Learning with Hugo Larochelle - 631
Answers 383 questions
Real-Time Machine Learning in the Database with Nikita Shamgunov - #84
Answers 383 questions
Trends in Machine Learning & Deep Learning with Zack Lipton - #334
Answers 383 questions
Scaling Multi-Modal Generative AI with Luke Zettlemoyer - 650
Answers 383 questions
Scaling Enterprise ML in 2020: Still Hard! with Sushil Thomas - #429
Answers 383 questions
