Joe Daws

@Typing_Tech huh, never tried that before, sounds pretty good!

Anyone else use coconut sugar in their coffee? If not give it a try. It adds a different dimension to the flavor. Still sweet but with a touch of coconut. And it's better for you than white sugar. A touch pricey maybe, but I make my own coffee so I'm still saving money 🤑 ☕

Puzzler #8 is live! Submit your solution to puzzler@utk.edu by 6pm Friday, 16 Feb. Submit an individual solution, or you work with others and submit a solution as a group. The most complete solution(s) will be recognized in the next issue. math.utk.edu @ArtsSciencesUT

@simonsarris Agree about N. Descriptive statistics are useful, however treating the average or variance as a pillar of truth that all decisions should be based off may not be the best strategy. These qualities describe the group and not the individual. Many cases you want plan for individuals

I actually think the opposite. In things like parenting studies, people chasing large n values and aggregating everything leads to 1. looking at simplistic vars bc that's all that's clear in the dataset 2. washing out any truly exceptional and interesting differences

favorite editor and then port them over using my tool. Check out the atool project on my github if you want to see more details. 4/4

words will fit into that particular bucket and then starts filling the next one. At the end we update the index with the appropriate denominator (which might trigger words to be moved between buckets if the new index contains more characters). Now I can draft tweets in my 3/4

For long tweets that are broken up into multiple posts, you may want an index for each tweet e.g. (3/10). Computing the required number of pieces is trivial when you allow yourself to chop up individual words arbitrarily. When keeping words whole, the number of pieces depends 1/4

@ferrari955 @htmx_org Nice! Are you using a web server too? Like Flask or FastAPI

building a new dashboard to follow my api status using @htmx_org , is real. 10/10 developer experience Python + Jinja + HTMX + sqlite

@iliaaamiri I have felt a similar feeling as you I think. To combat it I’m trying to be more patient with how long it takes me to learning something. Maybe for me, some things will take years. So far this has made me feel better and I’m happier with my small progresses.

@Neoriceisgood @ghouleebones These are great 👍 I like your mola mola! Please consider a coelacanth

Time to call it a night! Here's full progress on the fish suggestions I've been taking! Will probably continue tomorrow, so feel free to suggest more. #pixelart

@zachwritescode @ThePrimeagen lol 😂 thanks this is new for me

@ThePrimeagen stackoverflow.com/questions/1732…

give me the most unhinged stack overflow questions / answers I NEED THEM NOW

@ryanwinchester @SledgeDev I like this approach! For some users it might be easier to get them interested with a small web app.

@SledgeDev Real MVPs are CLI first, and then Web. Maybe a mobile app for n the off chance you score some users

If you are building an mvp are you going with local storage or firebase to get something out quickly?

@SledgeDev Appreciate your sacrifice

I got snow blindness trying to take this photo for you

@ScottPlusPlus Appreciate the feedback, will check out text tools

@JoeDawsJr honestly my advice would be don't shop around a solution in search of a problem. BUT. chunking out text to fit within an LLM's limit comes to mind. Also check out onlinetexttools.com for some thoughts / inspiration

Building in public with less than 1,000 followers? Share your project below and let me know what type of feedback I can give you. Let's help each other grow! 🚀

@kiyov09 It’s a modal transmission not a manual transmission

If you think of the clutch as the car’s escape key, driving with a stick is pretty much the same as using vim. Although is a lot easier to get out of it.

For splitting a long paragraph into tweet sized chunks, users would like

@Westoncb I’ve seen some work with adding “memory” to deep learning models. They store some part of training set directly and reference it during solution of the task. openaccess.thecvf.com/content_ECCV_2…

Is there anything like bi-directional RAG where the LLM write into a stored document at will? Would be interesting to see it maintain a "world model" that way..

So true! It also came in handy for temporal segmentation for building a training set for an ml model

@Typing_Tech In general I’ve had luck learning (still a bunch left for me to do) by running web technologies locally. When learning new tech I always stumble with the first local setup but eventually it seems worth it even if it’s a pain to get it running.

@Typing_Tech For MySQL learning I’ve found it helpful to run a small database locally on your machine. I’ve found mariaDB to be a good way to get running locally. Then build some tables and put some data into them related to the job you want. Users table, items for sale table etc.

Ok folks, there is a fullstack job I'm eyeing up, but I need to learn more before I'm comfortable applying to it. They want candidates to know Svelte, TypeScript, MySQL, Node.js, CloudFlare workers, and AWS SQS along with the "usual set of web technologies". I welcome any advice.