engblogs

This blogpost discusses model management for LoRA fine-tuned models using Llama2 and Amazon SageMaker, including best practices and two methods for managing the models.

Learn how to use large language models to detect fake news in this blog post, with a focus on Chain-of-Thought and ReAct prompt techniques.

GraphCast is a state-of-the-art AI model for faster and more accurate global weather forecasting, offering improved predictions and earlier warnings of extreme weather events.

DoorDash Drive implements privacy engineering by using geomasking to protect user privacy while maintaining local analytic capabilities.

Mastering app size is crucial for user experience and environmental impact.

Exploring the concept of being a 'No Sabo' kid and the unique situation of heritage speakers in language and culture.

Discover why Dropbox's latest data centers are using 400G networking, combining AI-powered tools, sustainability goals, and advanced technologies to meet bandwidth requirements and improve efficiency.

Introducing asymmetric certified robustness and feature-convex neural networks for fast and accurate certification against adversarial attacks.

Detecting speech and music in audio content

Overview of MARRS, a Multimodal Reference Resolution System, for handling conversational, visual and background context in a Natural Language Understanding system.

Extending the RoPE: A line of research that improves context length of RoPE for easy adaptation of pretrained language models to challenging tasks.

Accelerating generative AI worldwide by offering data residency guarantees in various countries.