Sean RyanPreparing a combined toxic/harmful-prompt/neutral dataset for classification trainingIn order to train a classifier to flag user prompts as toxic or harmful (for example, handling if user asks an AI to recommend how to hack…Jun 13Jun 13
Sean RyanEfficiently flag toxic or harmful texts and prompts: create a simple text based classifier via the…Identifying harmful or toxic prompts is obviously important to avoid an AI service such as an LLM generating problematic content. Whilst…Jun 9Jun 9
Sean RyanBatch summarize and translate docs via OpenAI or a local LLM for privacy: Introducing yet another…An obvious application of LLMs is to summarize and translate text — so already we have quite a few command line tools (such as…Jun 6Jun 6
Sean RyanEPE (Extract-Predict-Evaluate): a pipeline architecture for evaluating AI ServicesEPE (pronounced ‘E.P.’) is a simple pipeline architecture for objectively evaluating AI performance across model versions and…May 29May 29
Sean RyanLLM Reliability: JSON vs YAMLRecently there have been several articles pointing out that as an LLM output format, YAML is both cheaper and faster to generate. This…May 24May 24
Sean RyanReduce LLM costs and increase speed: consider switching to YAML instead of JSONIn recent years, JSON (JavaScript Object Notation) has become a go-to data format for transferring data, for example when calling a web…May 21May 21
Sean RyanLLMs and genai: Is the LLM emperor in need of more clothing? Enhancements and MitigationsThe promise: Knowledge workers made more productive, or even replaced, by genaiApr 28Apr 28
Sean RyanBare bones LLM integration via Function CallingLarger LLMs can be integrated into an existing system (for example a Home Automation system) via Function Calling - via a bare bones…Apr 14Apr 14
Sean RyanUsing an LLM to document a Python code base: a scripted approachDocumenting software is a chore often left undone or incomplete, since it is not a functional requirement, and adds overhead to…Mar 17Mar 17
Sean RyanGenerating test data using Mistral-7B LLM: a hybrid approachTest data is important for software development, to be able to test the software without exposing customer or IP sensitive data to too many…Mar 9Mar 9