Kanwaljit Khurmi

Off-the-shelf models work for prototyping, but production forces a choice: which technique and which compute to use. In this chalk talk, we work through the decision framework live, covering when to start fine-tuning after you've pushed the limits of in-context learning and which fine-tuning technique to use.

Learn to customize OpenAI GPT-OSS models for your specific business requirements using Amazon SageMaker AI through Amazon SageMaker HyperPod recipes and open-source Hugging Face libraries.

Join us to learn a practical and cost-effective way to customize AI models using spectrum-aware fine-tuning and fully managed training experience provided by Amazon SageMaker training jobs.

In this builders’ session, get step-by-step implementation experience fine-tuning Llama 3 8B using the QLoRA strategy. Learn how you can fine-tune Llama 3 models at scale using the memory-efficient PyTorch Fully Sharded Data Parallel (FSDP2) library with access to built-in debugging and GPU utilization metrics.

Organizations that are running machine learning (ML) systems and generative AI applications on their local laptops/servers want to take advantage of the scalability and performance of the AWS Cloud.

Dealing with I/O Bottlenecks in High Throughput Model Training - Jan Sellner, German Cancer Research Center Have you ever faced the problem of low GPU utilization due to data loading bottlenecks? I/O heavy workloads require optimized data loading strategies as otherwise the GPU usage is suboptimal...

AWS is accelerating contributions to open-source technologies like Kubernetes and Kubeflow to streamline machine learning training and inference tasks. In this chalk talk, learn how ML developers and data scientists can use AWS open-source contributions, Amazon SageMaker Operators for Kubernetes, and SageMaker components for Kubeflow to build secure, scalable, and highly available machine...