Category: Blog
-
Latest Research on Medical LLM Models
Models like BioBERT, ClinicalBERT, and PubMedGPT are making waves. They’re fine-tuned on medical texts, making them sharp tools for healthcare tasks. Next, we have multimodal models. These beauties integrate text with images and structured data, boosting diagnostic accuracy and personalized treatment plans. Making these models understandable is key. Researchers are on it. Benchmarks Datasets Medical…
-
How LLMs can get better with Self Reflection
SELF-RAG is a method that can make your LLMs more accurate by using self reflection. You can read all about it in this white-paper 馃憠 Self-RAG: Learning to Retrieve, Generate, and Critique through Self-Reflection TL;DR The paper addresses the challenge of factual inaccuracies in large language models (LLMs) due to their reliance solely on parametric…
-
Hacking LLMs
What not to do and how to do it. This post aims to give you an idea of how critical it is to continuously improve guardrails around your LLM powered solutions. Prompt engineering can be used to make your LLM based chat bots produce harmful content and even reveal sensitive data. People are not going…
-
Use private data securely with an LLM
Large language models (LLMs) are amazing but one shortcoming is that they wont be able to answer questions related to your private data. This is where RAG or retrieval-augmented generation comes in. RAG is a technique that lets you use LLMs with your private data. Dont roll your own RAG RAG architecture has quite a…
-
Power up LLM bots with RAG + Hybrid search
Retrieval-Augmented Generation or RAG is important because it helps LLMs give more accurate answers by also including content (or context) from your private data. See this post on how you can implement RAG quickly and securely. With RAG, your private data is stored as an embedding in a vector database. Depending on the user query,…
-
Build highly scalable AI platforms like the Deutsche Bahn
Deutsche Bahn has developed an AI platform that is scalable, secure, and efficient. They have a dedicated AI platform team and their focus was on rapid project initiation, reduced maintenance costs, and effective management of AI and machine learning tools across diverse data analytics teams. AWS Sagemaker has helped them achieve their goals. For an…
-
To Microservice or not
The first law of distributed systems is, don鈥檛 distribute! M. Fowler If you still feel the need to consider moving to a distributed architecture then microservices is a good alternative to a monolith. Robust and Stable Microservices encourages a design where components are highly cohesive and have low coupling. This is an implicitly stable design.…
-
The monolith architecture
Think of a monolith architecture as a system that needs to be changed and deployed as a single entity. Development teams work on different modules of this entity, Testing teams run tests on the system and then it is eventually deployed as a single entity to users. Contrary to what many think, a monolith is…