I want to share my experience in the cloud, AI, and agents development areas.
The blog post introduces InstructLab, a project by IBM and Red Hat, outlining the fine-tuning process of the model "MODELS/MERLINITE-7B-LAB-Q4_K_M.GGUF." This involves data preparation, model training, testing, and conversion, finally serving the model to verify its accuracy, by using a personal musician example.
This blog post provides a step-by-step guide to setting up InstructLab CLI on an Apple Laptop with an Apple M3 chip, including an overview of InstructLab and its benefits. It also mentions supported models and detailed setup instructions. Additionally, it refers to a Red Hat YouTube demonstration and highlights the project's potential impact.
The popularity of llama.cpp and optimized GGUF format for models is growing. This post outlines steps to run "Phi-3-Small-128K-Instruct" in GGUF format with llama.cpp on an IBM Cloud VSI with GPUs and Ubuntu 22.04. It covers VSI setup, CUDA toolkit, compilation, Python environment, model usage, and additional resources.
This blog post focuses on the importance of Prompt Engineering in AI models, particularly Large Language Models (LLMs), for reducing manual effort and automating validation processes. It emphasizes the need for automation to handle increasing test data and variable combinations, and discusses the use of the Watsonx.ai Prompt Lab for manual and initial automation processes. The post also highlights the significance of integrating automation with version control for consistency and reproducibility.
This blog post delves into the initial fine-tuning process for large language models (LLMs) for multi-turn conversations and their deployment on Text Generation Inference (TGI) servers. It covers topics such as use cases, data formats, training data preparation, server setup, and evaluation frameworks. The goal is to guide readers through the process of fine-tuning and deploying LLMs.
This blog post outlines a bash automation for setting up and testing Text Generation Inference (TGI) using a container. It provides instructions for creating a Python test client, starting the TGI server, and troubleshooting common issues. The post emphasizes the benefits of using containers and references the Hugging Face and Nvidia technologies.
This cheat sheet provides a two-step guide for adding users to your watsonx project in IBM Cloud.
Last time, I wrote a blog post about "IBM Watsonx.ai and a simple question-answering pipeline using Python and FastAPI", and I had an exchange with my family about an OpenAI sample for a FastAPI application, so I created a small FastAPI server to access OpenAI with Python.
The objective of this project is to deploy the Reranker to a Kubernetes cluster in a VPC on IBM Cloud and access the REST API of the Reranker. The Reranker is a component of PrimeQA.
My last blog post was about Watson Speech to Text language model customization and this blog post is about IBM Cloud Watson Text to Speech (TTS) custom voice model configuration. Because, now it's time to have some fun with the Watson TTS service. I created a fun customisation of the service that the German pronunciation sounds a little bit like the palatinate dialect. Here are the differences with two wav file I created with a custom Watson to Text to Speech voice model.
Thomas Suedbroecker's Blog 