## hinst-website 🌐 I organized the code of my personal website.
I identified these packages: ### Server - **base**: Everything
that's needed in all other packages. - **db_objects**: For
interacting with SQL databases. - **page_data**: For HTML
templates. - **rest_objects**: For objects handled by the REST
API. ### gophers - Functions for Go for every scenario. I placed
this package in a separate Git repository
(https://github.com/hinst/go-gophers) and generated a cover
image, as shown on the screenshot. Last week, I had little free
time for coding, so I only logged 2 hours and 30 minutes in my
tracker. I've been thinking about whether to add ORM
(Object-Relational Mapping) and REST frameworks to my project. I
haven't found a clear answer yet. Here's an idea: 1. Run a load
test to see how many requests per second my website can handle.
2. Implement a REST framework and repeat the test. Then compare:
did it get faster or slower? The idea is intriguing, but finding
time for it is a challenge. Currently, I only have time for
small tasks, and refactoring the project under a new framework
is a lengthy process. ## Experiments with Local LLMs I found two
approaches to settings: - **100% Offload**: Ideal for long
conversations. Generation is fast, but context switching is very
slow. You can't jump between chats or change prompts on the fly.
- **50% Offload**: Better for context switching. Generation is
slower, but context switching is faster. You can easily jump
between chats and modify prompts in real time. It's interesting;
it seems like human thinking: you can switch between tasks, but
work will be slow. Or you can focus on one task and work quickly
without distractions (I struggle with that 🙃).
