Inference is more than GPUs. Baseten delivers the infrastructure, tooling, and expertise needed to bring great AI products to market. Get performant model runtimes, cross-cloud high availability, and seamless developer workflows. All powered by the Baseten Inference Stack
Continual learning in large language models requires balancing weight updates with parametric memory to avoid catastrophic forgetting and maintain general capabilities. While current frontier models rely heavily on KV cache searching, integrating knowledge directly into model weights remains a significant challenge due...
Applying LLMs to complex legal workflows requires moving beyond general-purpose frontier models toward specialized, agentic systems capable of exhaustive document analysis. Legal data rooms often exceed current context windows, necessitating advanced compaction strategies like neural KV-cache compression to maintain in...
The rapid concentration of artificial intelligence development within a few dominant firms threatens to create an oligopoly of intelligence, necessitating a shift toward open-weight models to ensure broader access. Poolside addresses this by prioritizing the industrialization of model building, as evidenced by their ne...
The shift from academic research to industry-led AI development hinges on the ability of individuals to leverage specialized models rather than relying solely on closed-source frontier labs. Charlie O'Neill, co-founder of Parse and team member at Baseten, argues that researchers should adopt a "top-down" learning appro...
The Baseten Frontier Gateway addresses the limitations of generalist AI models by facilitating the deployment of specialized models that smooth out the "jagged frontier" of performance. While the industry previously prioritized single, all-encompassing models, the current landscape features a "Cambrian explosion" of sp...
Medical professionals face a critical information overload, with 10,000 to 20,000 new papers published daily, making it impossible to stay current through traditional tools like PubMed during brief patient encounters. OpenEvidence addresses this gap by using complex machine learning queries and model orchestration to p...
Inference infrastructure presents significant, often underestimated complexity as AI applications scale, moving beyond simple commodity status. While early-stage startups may initially attempt to build internal infrastructure, offloading this to specialized providers like Baseten allows teams to focus on core product d...
Machine learning infrastructure is undergoing a fundamental shift as enterprises move from experimental AI projects to mission-critical applications. The primary challenge has evolved from achieving high-quality model output to ensuring the speed, reliability, and cost-efficiency of inference at scale. As organizations...
AI inference represents the next major market opportunity, shifting focus from model training to scalable, production-ready infrastructure. While enterprises often struggle to transition AI initiatives from experimental "skunkworks" to full-scale production, they increasingly seek long-term partners capable of managing...
Follow this podcast in Podwise
Sign in to get AI summaries, transcripts and mind maps for any episode, including new ones.
