Speculative Decoding and Efficient LLM Inference with Chris Lott - #717

Speculative Decoding and Efficient LLM Inference with Chris Lott - #717

Author: Sam Charrington February 4, 2025 Duration: 1:16:30
Today, we're joined by Chris Lott, senior director of engineering at Qualcomm AI Research to discuss accelerating large language model inference. We explore the challenges presented by the LLM encoding and decoding (aka generation) and how these interact with various hardware constraints such as FLOPS, memory footprint and memory bandwidth to limit key inference metrics such as time-to-first-token, tokens per second, and tokens per joule. We then dig into a variety of techniques that can be used to accelerate inference such as KV compression, quantization, pruning, speculative decoding, and leveraging small language models (SLMs). We also discuss future directions for enabling on-device agentic experiences such as parallel generation and software tools like Qualcomm AI Orchestrator. The complete show notes for this episode can be found at https://twimlai.com/go/717.

Hosted by industry analyst and commentator Sam Charrington, The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence) serves as a vital conduit between cutting-edge research and its real-world implications. This isn't just a series of technical lectures; it's a series of conversations that unpack how AI and machine learning are actively reshaping industries and societal structures. Each episode connects you directly with leading researchers, engineers, and innovative thinkers who are defining the frontiers of the field. The discussions go beyond abstract theory to explore the practical challenges, ethical considerations, and business transformations driven by these technologies. Whether you're a data scientist deep in the code, a tech-savvy leader strategizing implementation, or simply fascinated by the future of intelligent systems, this podcast provides the context and depth needed to stay informed. By focusing on the people behind the algorithms and the ideas powering the platforms, Sam creates a resource that is both intellectually substantive and genuinely engaging, building a thoughtful community around one of the most significant technological shifts of our time.
Author: Language: English Episodes: 100

The TWIML AI Podcast (formerly This Week in Machine Learning & Artificial Intelligence)
Podcast Episodes
Building an AI Mathematician with Carina Hong - #754 [not-audio_url] [/not-audio_url]

Duration: 55:52
In this episode, Carina Hong, founder and CEO of Axiom, joins us to discuss her work building an "AI Mathematician." Carina explains why this is a pivotal moment for AI in mathematics, citing a convergence of three key a…
Vibe Coding's Uncanny Valley with Alexandre Pesant - #752 [not-audio_url] [/not-audio_url]

Duration: 1:12:36
Today, we're joined by Alexandre Pesant, AI lead at Lovable, who joins us to discuss the evolution and practice of vibe coding. Alex shares his take on how AI is enabling a shift in software development from typing chara…
Dataflow Computing for AI Inference with Kunle Olukotun - #751 [not-audio_url] [/not-audio_url]

Duration: 57:37
In this episode, we're joined by Kunle Olukotun, professor of electrical engineering and computer science at Stanford University and co-founder and chief technologist at Sambanova Systems, to discuss reconfigurable dataf…
The Decentralized Future of Private AI with Illia Polosukhin - #749 [not-audio_url] [/not-audio_url]

Duration: 1:05:03
In this episode, Illia Polosukhin, a co-author of the seminal "Attention Is All You Need" paper and co-founder of Near AI, joins us to discuss his vision for building private, decentralized, and user-owned AI. Illia shar…
Is It Time to Rethink LLM Pre-Training? with Aditi Raghunathan - #747 [not-audio_url] [/not-audio_url]

Duration: 58:26
Today, we're joined by Aditi Raghunathan, assistant professor at Carnegie Mellon University, to discuss the limitations of LLMs and how we can build more adaptable and creative models. We dig into her ICML 2025 Outstandi…