stam retweetledi
stam
1.3K posts
stam
@p_stamatis
Architect Engineer UP Greece, MSc Architectural Computation at UCL.Current: Senior Automation Developer at ARUP, PHD student at UCL researching spatial Analysis
London Katılım Şubat 2010
699 Takip Edilen370 Takipçiler
If you cannot support your product as you agreed, maybe you should be forced to make it open source?
youtube.com/shorts/U4-pp1u…
YouTube
English
@cmuratori For me it is funny that people are quicker to opt in to a solution that uses 100s of machines before they think really hard on the core algorithm.
English
Hope nobody was expecting next week to be calm and quiet...
English
@davefarley77 The larger the model and more data less loss but not necessarily higher utility. I admit this could mean higher insecurity, but also results like bing chat going off the rails.
They talk about it in this computerphile video:youtu.be/viJt_DXTfwA at 27:35
YouTube
English
For the people that think that I was being alarmist and too gullible in my video on AI, I watched this excellent video yesterday. Amongst other things it points to the gap between research into AI safety and progress.
youtu.be/qOoe3ZpciI0
YouTube
English
A key debate, not often discussed is "who did it"? When researchers create an AI to detect, faces, or to create poetry, or to retrieve and collate in a coherent manner text (LLMs) then I would argue the researchers did it! AI is a tool, like a brush to a painter.
Dave Farley@davefarley77
I bet you've come across a lot more AI generated content recently... something's going on. It has me a little worried. 🎥 Here's why I think we need to stop artificial intelligence research ➡️ youtu.be/FC01lHMtq2U
English
@AsafAgami92 @davefarley77 I think this is one of the trickiest parts, where you need way more effort to assure everyone is pushing passing tests, onboarding hurdles with git hooks etc...perhaps @davefarley77 has some obvious fixes that we are missing?
English
@davefarley77 When do automated tests run with trunk based development? With PRs, the tests run as a requirement before the PR is approved
English
Excellent blog post on Trunk Based Development in preference to Feature Branching here:
"When you have a team, or teams, however, working on a project together, feature branching creates problems"
infobip.com/developers/blo…
English
@alexbunardzic Why not the same syntax on how you declare variables? Because people like to be different and opinionated! :)
English
My long standing pet peeve is different characters that denote code comments. I get frustrated any time I work in a different programming language & want to comment out some line. Is it // or /* or # or...?
Why no standards? Why is it so hard to agree on that?
English
Neat...but have the number of tokens even changed? The number of characters is irrelevant for LLMs, plus I guess you could automatically remove dead words like "the", "a" and still compress the paragraph without introducing obscure tokens like emojis.
Taelin@VictorTaelin
Pro tip: you can greatly increase GPT-4's effective context size by asking it to compress your prompts using its own abbreviations. #GPT4
English
@AlanZucconi A thread about the quirks maybe? I think it would be interesting to know a few more details as I think any piece of software has quirks and it is hard to avoid it.
English
As someone who pretty much teaches Unity for a living, this is a real issue. ⛔
And the many memes my students are making of me losing it while fighting Unity's poor design choices are a testament to that. 🥲
Dr Kate Compton@GalaxyKate
I'm going to say it, Unity is becoming a net negative for education. Teachers on the ground can report on how much time is spent on teaching Unity quirks, not design or even programming
English
@unclebobmartin It feels like you need to debate over complete pieces of code.The Q: virtual methods vs switch statements -> readability over the two hierarchies? @cmuratori original example in both styles was readable, how does Bob's example with the payroll be written in both styles?
English
Yet another go around, though I think this one might be hinting towards a resolution.
github.com/unclebob/cmura…
English
I was happy to see a civil exchange. The discussion certainly made me think harder about clean code, as I often have to write, in generative design, as fast as possible code but I also need to maintain it over time and add more and more features. There needs to be a balance.
Uncle Bob Martin@unclebobmartin
The current state of the @unclebobmartin / @cmuratori discussion: github.com/unclebob/cmura…
English
@AlanZucconi Visual scripting is scripting, visual coding is coding :)
English
@allenholub @Google Because then they will not be able to upsell you speakers! They used to have a Google audio, which was making any speaker smart...
English
Hey @Google. I don't want your stinking so-called high-quality speakers. I have real speakers. Why don't any Nest devices have a digital audio out? Bluetooth doesn't make it.
English
@p_stamatis On the teams I work with, we usually mark specifications (tests) for new code as "Under Development" or similar while the feature is being built, so that they don't fail the build.
English
I confess that I am not a big fan of "The Test Pyramid" so here's a #Thread on what I think is a better focus for your automated testing strategy...
1/14
English
@davefarley77 Does this mean that all the while you work on a new feature, these tests fail on the pipeline, since your feature is not implemented yet?
English
I start new features by creating user-focussed "Executable Specifications" that define the outcome that I am trying to achieve for the user. (BDD style specifications)
9/14
English
@alexbunardzic I don't like Jira, but can't you use any type of card, and just say to the team this board has the user stories? I did that on GitHub boards with some mixed success.
English
Why are teams who are using Jira attempting to deliver user stories when there is no possibility to craft a user story in Jira? Either abandon Jira or abandon user stories. Stop trying to fit square peg into a round hole.
English