Return guests Kanjun and Jonathan discuss Imbue and Databricks' 70B LLM, outperforming GPT-4o. Topics include infrastructure needs, cost-aware hyperparameter optimizer CARBs, and challenges in training large models. They delve into MFU monitoring, bug tracing, and optimizing infrastructure for deep learning. The podcast explores evaluation metrics and AI model practical applications, emphasizing the importance of creating useful tools for customers.

Shawn Wang discusses the AI Engineer role, team composition, trends in AI research, and advice for product creators with a focus on AI Engineer World Fair.

Guests from Elicit share insights on hiring AI engineers, blending conventional skills with ML knowledge. Topics include fault tolerance in coding, model shadowing, sourcing engineers, and challenges in AI engineering.

Guest Mike Conover from Brightwave discusses using large language models in finance, challenges in long context windows, and the evolution of AI research. Topics include implications of polarizing data, value of employee-led datasets, complexities of data temporality in finance, AI in financial services, AI hedge funds, and the evolution of large language models.

Expert guests Graham Neubig and Aman Sanger discuss AI topics like Code Edits, Sandboxes, Academia vs Industry. They delve into Benchmarks like SWEBench, Dataset Contamination Detection, and GAIA Benchmark. The conversation also touches on Reasoning - Self-RAG, Let's Verify Step By Step, and developments in multi-agent systems with MetaGPT.

Mark Huang of Gradient.ai discusses training a 1 million context LLM, highlighting challenges like memory scaling and the need for techniques such as curriculum learning. The podcast explores strategies for training large language models, evaluating ring attention implementations in PyTorch, and enhancing value in AI technology through early fusion models and wise resource investment.

Christian Szegedy, Ilya Sutskever, and Durk Kingma discuss the most notable topics from ICLR 2024, including expansion of deep learning models, latent variable models, generative models, unsupervised learning, adversarial machine learning, attention maps in vision transformers, efficient model training strategies, and optimization in large GPU clusters.

Jesse Silver discusses the rise of NSFW AI chatbots in adult entertainment, exploring challenges and potential in automating interactions. The conversation covers maximizing creator profit, emulating human interactions realistically, and managing creator income streams. It also delves into technical aspects like context, long memory, and personality profiling in generating interactions aligned with creators' brands.

Guests Joscha Bach, Karan Malhotra, and Rob Haisfield discuss the evolution of generative AI, GANs, GPT-2, and simulative AI. They explore the potential of AI for game experiences, chat interactions, and world simulations, highlighting advancements in text generation and creative applications of AI technology.

Jason Liu, Co-founder of Instructor, discusses the evolution of recommendation frameworks with GPT-3 embeddings at Stitch Fix, the importance of function calling for structured outputs, optimizing workflows for model fine-tuning, the role of prompts in AI models, and exploring hiring practices and innovation in AI engineering.