PinnedIntuitive Thinker — Introducing Guided Mental Models to Enhance Small LLM ReasoningAt a high level, Large Language Models (LLMs) function as statistical models that predict the most likely next word(s) in a sequence based…Sep 27Sep 27
PinnedVision Language Models for Radiologists: Detecting Regions of Interest in Brain CT ImagesRecent advancements in Vision Language Models have significantly propelled the realization of Smart Visual Assistants for radiologist image…Sep 7Sep 7
PinnedAgentic Workflow ImplementationsWhen building AI applications involving multiple tasks implemented by different components (chains or pipelines), complexities can quickly…Aug 2Aug 2
PinnedAutomatic Knowledge RAG with R2RThis is part two of Demystify Knowledge RAG frameworks. In our previous writing, we explored Knowledge RAG concepts, Microsoft GraphRAG…Jul 164Jul 164
PinnedBuilding High Performance Computing Clusters (GPU) for AI at ScaleThe use of GPU accelerated machine learning workloads. Multi GPUs and Multi-Nodes enable distributed training at scale. Supercomputers rely…Jul 25Jul 25
Multi-GPU Training for Llama 3.2 using DeepSpeed and Redundancy Optimizer (ZeRO)For inference tasks, it’s preferable to load entire model onto one GPU, containing all necessary parameters, to carry out these tasks…Oct 1Oct 1
PaliGemma: Receipt & Invoice JSON v2In previous work, I created multiple experimental Large Language Model (LLM) architectures to convert receipt images into JSON or XML…Sep 7Sep 7
The Rise of CPU Inference: Powerful and Affordable LLMsGPUs are the default choice for LLM inference due to their speed advantage, but I believe we’re approaching a wall in terms of cost and…Aug 10Aug 10
Invest in Strategies to Save LLM CostAs we all know today, not all language models (LLMs) are created equal; some models are more resource-intensive than others. Factors such…Aug 6Aug 6
Improving Software Reliability and Security through Memory-Safe LanguagesMemory safety issues refer to vulnerabilities that arise from incorrect or unexpected behavior when dealing with dynamic memory management…Jul 23Jul 23
