About the talk
We will be talking about a comparative study of GPT and BERT, two transformative models in the field of Natural Language Processing (NLP). You will gain insight regarding why GPT became a dominant model in the field of NLP and LLM.
This talk will cover
- A brief introduction to GPT and BERT and their unique language understanding mechanisms.
- An in-depth comparison between GPT and BERT, highlighting their architectural differences and training strategies.
- Discussion on GPT's next-word prediction, efficient training with causal self-attention, and zero-shot/few-shot learning capabilities.
- Examination of the contrasting training requirements and transfer learning abilities of BERT and GPT models.
Testing & Development Tools
About the speaker
Artem has a PhD in Physics. Currently working as a Quant Researcher at Deutsche Bank with prior experience at Morgan Stanley. Besides his primary profession, he has a keen interest in deep learning, specifically Transformer-based Language Models.