Languages that try to disallow idiocy become themselves idiotic. — Rob Pike

For the first time I paid to have a work of art professionally framed. For django-cast, the work on separating the data fetching from rendering the content is almost done, so I moved on to the next interesting problem. As the saying goes: To make an easy change, first make the change easy (which was really difficult) and then forget about the easy change and do something completely different because it's not challenging anymore. At least I managed to get the changes I made so far into a new release.

Articles

• Efficient finetuning of Llama 3 with FSDP QDoRA | Oh I desperately need a use case for fine-tuning 🤗
• Styling an Django RSS Feed | Interesting idea, dunno if I want to risk breaking my feed, but fair enough!
• Django Developers Survey 2023 | Interestingly the number of people using SPAs goes down, wouldn't have expected that

Weeknotes

• Weeknotes: Llama 3, AI for Data Journalism, llm-evals and datasette-secrets | Simon Willison
• Weeklog for Week 16: April 15 to April 21 | Johannes

Videos

• DJANGO REINHARDT: What every guitarist should know (beginner to pro) | The other Django 🤓
• 10 Key Questions Every Company Should Ask Before & While Using Al Responsible Al | PyConDE Keynote Ricardo Baeza-Yates - very good!

Software

• django-translated-fields - Django model translation without magic-inflicted pain.

Fediverse

• One of the things about charlatans is that we—oops, “they”—are an uninformed person’s idea of a genius 🧵
• Today at @answerdotai we've got something new for you: FSDP/QDoRA. We've tested it with @AIatMeta Llama3 and the results blow away anything we've seen before. I believe that this combination is likely to create better task-specific models than anything else at any cost. 🧵