28 Mar 2024
This article aims to demonstrate how Ollama Grid Search can streamline the process of comparing and selecting Large Language Models (LLMs) for various tasks and provide answers to common questions such as:
What is the best model for for story telling?
I want to use LLMs to generate RPG scripts, which model should I use?
Read Full Post
21 Mar 2024
Sometimes you have a list of several items that need to be processed by an API, like a list of documents to be summarized by a Large Language Model, or a selection of terms that trigger scraping jobs.
If these tasks are expensive (i.e.: they demand a lot from the server), or if there are rate limits, you may need to use a Limited Concurrency pattern to stop the server from being flooded with simultaneous requests.
This article explains how this was solved in React, using the...
Read Full Post
27 Dec 2023
This article explores the process of optimizing Large Language Models (LLMs) for specific applications through grid search.
Our goal is to streamline parameter tuning for enhanced inference efficiency, complementing prompt engineering efforts, using models hosted in an Ollama instance and an interface built in Rust.
[01-MAR-2024] Note, this article was written before we released a Desktop version of Ollama Grid Search!
Read Full Post
01 Nov 2023
Coming across “Top 15 Rust Projects To Elevate Your Skills”, I decided to build a Todo CLI application (quite the road less traveled, eh?) and defined a few goals/requirements after checking some of their suggested crates and examples:
- Use a database to persist todos (I used SQLite).
- Minimize the use of
unwrap()
and expect()
calls. - Include relevant working test cases, even if they are not 100% necessary.
This article is a walk through the process of...
Read Full Post
15 Oct 2023
After a quick attempt to build a desktop application using Tauri, I decided to learn Rust and started by using the excellent Rustlings resource.
Read Full Post
09 Sep 2021
In this article we explore a way to replicate a backend’s authorization system on a React SPA, so that developers can restrict access to features, pages or components based on group memberships or user permissions.
Although this example is based on a Django backend, it’s easily applicable to any backend framework that can return a user
object by REST or GraphQL API calls.
IMPORTANT
This is not an end all security solution, as any content or data sent to the browser might...
Read Full Post
16 May 2020
This article explores the issue of running “simultaneous” searches for all entries in a list of terms (in reality, we are running searches asynchronously).
I’m also trying to improve on Azure’s Search API example a bit.
Read Full Post