DEV Community: Dalmas Chituyi

Single Page Applications (SPAs) vs. Multi-Page Applications (MPAs). Navigating the Web App Landscape

Dalmas Chituyi — Fri, 29 Mar 2024 13:23:57 +0000

Yes, you can create webpages that hide the internal structure of your website's URLs from users. This technique is called a Single Page Application (SPA) or client-side routing. Here's how it works.

The entire website acts as a single HTML page.
JavaScript dynamically loads content and updates the page based on user interaction (e.g., clicking menus).
The URL bar might only show the domain name (www.example.com) regardless of the specific page the user is on.

Security Considerations.
Reconnaissance. Hiding URLs might make it slightly harder for attackers to understand a website's structure, but a determined attacker can still discover it through various techniques.

Plotly Dash is a framework for building analytical web applications and it natively builds single-page applications (SPAs). With Dash, you create your UI using Python, but under the hood, it’s generating a web-based SPA. This means that user interactions in the application can update the visuals on the page without requiring a full page reload. This makes Dash applications very responsive and efficient. However, it’s also possible to build multi-page applications with Dash by defining different “routes” or URLs for each page, but the core concept remains around SPAs.

[https://www.linkedin.com/pulse/single-page-applications-spas-vs-multi-page-mpas-web-app-chituyi-icknf/?trackingId=S9rFiiZlRQG90q8FgeSunw%3D%3D]

Single Page Applications (SPAs) vs. Multi-Page Applications (MPAs). Navigating the Web App Landscape

Dalmas Chituyi — Fri, 29 Mar 2024 13:23:53 +0000

Yes, you can create webpages that hide the internal structure of your website's URLs from users. This technique is called a Single Page Application (SPA) or client-side routing. Here's how it works.

1. The entire website acts as a single HTML page.
1. JavaScript dynamically loads content and updates the page based on user interaction (e.g., clicking menus).
1. The URL bar might only show the domain name (www.example.com) regardless of the specific page the user is on.

[https://www.linkedin.com/pulse/single-page-applications-spas-vs-multi-page-mpas-web-app-chituyi-icknf/?trackingId=S9rFiiZlRQG90q8FgeSunw%3D%3D]

RAG vs SELF-RAG

Dalmas Chituyi — Sat, 11 Nov 2023 05:41:42 +0000

What RAG does is, btw RAG is rigid in how it considers retrieving the information in order to generate context using some sort of document similarity algorithm. So, 😊 RAG will retrieve information even if it doesn’t have to do it. It immediately assumes the LLM lacks learnt embeddings to make sense of a prompt by the user and that’s why it first generates features from a domain specific corpus and adds it as context to the LLM prompt without considering the LLM might have the most relevant context in the first place (Obviously, this is an implementation detail that as Self-RAG attempts to address). How do we ask the LLM this question before tuning its response with an external source? Inasmuch as RAG is a good way to add relevance to LLMs especially using enterprise data, augmenting RAG with the intelligence to know when to retrieve or not will reduce the impact of verbiage output. Unlike a widely adopted Retrieval-Augmented Generation approach, Self-RAG retrieves on demand and criticize its own generation.

Multimodal multilingual LLM

Dalmas Chituyi — Sat, 04 Nov 2023 12:21:43 +0000

SeamlessM4T is a single multilingual and multimodal model that can multitask to translate and transcribe with multiple input and Output languages.

🗒 Some of the tasks SeamlessM4T model can do:

☑️Speech to Speech Translation
☑️Speech to Text translation
☑️Text to Speech translation
☑️Text to Text translation
☑️Automatic Speech recognition.

↗ This is a significant improvement over previous machine translation models, which could only translate speech to text in a handful of languages with limited output languages. 💡 SeamlessM4T is also able to implicitly recognize the source language, without the need for a separate language identification model.

Built from the work done and the understanding of some of this models :

🔎 No Language Left Behind (NLLB). A text-to-text machine translation model that supports 200 languages.

🔎 Massively Multilingual Speech. Provides automatic speech recognition, language identification, and speech synthesis technology across more than 1,100 languages.

🔎 Universal Speech Translator. Model unwritten language through speech to speech translations.

🔎 Speech Matrix. Large-scale Mined Corpus of Multilingual Speech-to-Speech Translations.

Why I like Dash for ML and DS projects?

Dalmas Chituyi — Mon, 25 Sep 2023 17:52:38 +0000

Interactive data visualization. This is essential for ML and DS projects, as it allows you to explore and understand your data in a visual way. For example, you could create a dashboard that allows you to visualize the performance of your ML model on different datasets, or to explore the relationships between different variables in your data.
Dash is easy to learn and use, even for beginners. It has a simple API and a variety of pre-built components, which makes it easy to get started. It is possible to create a basic dashboard with just a few lines of code.
Flexible and scalable. This makes it suitable for a wide range of ML and DS projects. You can use it to build simple dashboards for exploratory data analysis, or complex dashboards for production use i.e., building a dashboard that allows users to interact with your ML model to make predictions.
Dash is collaborative; therefore, it is easy to share your dashboards with others. You can deploy your dashboards to the web or share them as standalone applications.
Dash is an open-source framework, which means that it is free to use and modify. This gives you the flexibility to customize the framework to meet your specific needs. You can use dash to add new components or extend the existing components.

Most of these projects here! https://dallo7.github.io/ are developed in Dash.

Old but relevant unpublished materials loading...

Dalmas Chituyi — Sat, 19 Aug 2023 13:01:52 +0000

...