Pre-Training Corpus Analysis
Sameer and Yasaman discuss the impact of pre-training corpus on model performance, emphasizing the need for transparency and understanding of training data sources. They delve into the potential risks and benefits of model memorization, highlighting the importance of designing diverse corpora to enhance model generalization and guard against data poisoning.In this clip
From this podcast
Machine Learning Street Talk (MLST)
#73 - YASAMAN RAZEGHI & Prof. SAMEER SINGH - NLP benchmarks
Related Questions