spott

10 ms·

91.
by spott 1y ago
Do you have a source? I’ve always heard that handwriting improves retention vs typing, but I’m interested in hearing about any research that counters that.
92.
by spott 1y ago
Most of numpysane seems to be doing a loop in Python. It isn’t true vectorization.
93.
by spott 1y ago
Pretty sure all those studies were comparing laptop (typing) to handwriting. Here is the first one I found: https://journals.sagepub.com/doi/10.1177/0956797614524581
94.
by spott 1y ago
> Too often, there’s a gap in introductory material because the vast majority of users of a programming language are not at an introductory level. Not for Python! There is way more “intro” to Python stuff out there than anything else…
95.
by spott 1y ago
There are 150M taxpayers in the US… so $50M is about 33 cents per American… Unless you know something I don’t.
96.
by spott 1y ago
A container can very rarely be reproduced by a dockerfile. I imagine with a lot of discipline (no apt update, no “latest” tag, no internet access) you can make a reproducible docker file…. But it is far from normal.
97.
by spott 2y ago
A question about erlang: Haskell taught me a lot about programming, things that I still use now, even though I only write Python. Does learning erlang teach you a new way of thinking? Or does it just make you wish you had erlang language f
98.
by spott 2y ago
On an ecological scale, it isn’t the heat that is a problem, it is the speed it is increasing. In the past, temperature changes have been slow enough for evolution and ecological systems to adapt, but now it is happening fast enough that th
99.
by spott 2y ago
You realize that the entire executive branch excluding defense is like 10% of the federal budget. There isn’t enough money to be saved to give you back anything.
100.
by spott 2y ago
Deepseek isn’t a distilled model (and neither is llama405), both are pre trained foundation models. Deepseek has distilled deepseek R1 into a couple of smaller open source models, but neither R1 or v3 are distilled themselves.
101.
by spott 2y ago
I wonder where they got 50… llama405 cost like 60M, which puts deepseek at closer to 10x…
102.
by spott 2y ago
Interesting, it looks like you are right, it is usbc: https://www.sonos.com/en-us/shop/sonos-line-in-adapter
103.
by spott 2y ago
What model do you have? The five and era 100 both say they have a line in on the website.
104.
by spott 2y ago
Through an audio jack.
105.
by spott 2y ago
The newer era speakers, and the five have audio in.
106.
by spott 2y ago
I always find it odd when a new benchmark comes out and uses models that are all 6 months - a year old. I’m pretty sure all these models are >1 year old. This looks super interesting and useful, but without Claude or o1 data points, i
107.
by spott 2y ago
I mean, think of LLM output as unfiltered thinking. If you were to make that determination would you make it before you had thought it through?
108.
by spott 2y ago
That is kinda cutting the line a little fine I think. How does one move from a monogamous relationship to a poly relationship except through an open relationship?
109.
by spott 2y ago
I’m honestly not sure why Bert-based-uncased is so popular… the model isn’t that useful on its own. From their huggingface page: > You can use the raw model for either masked language modeling or next sentence prediction, but it's
110.
by spott 2y ago
ModernBERT-Base is larger than BERT-Base by 39M parameters.
111.
by spott 2y ago
Embedding models are frequently based on Bert style models, but Bert models can be finetuned to do a lot more than just embeddings. So an embedding focused finetune of modern Bert should be compared to something like voyageai, but not moder
112.
by spott 2y ago
That is the training gpu… the inference gpu can be much smaller.
113.
by spott 2y ago
I can’t imagine going back to poetry after using uv. uv is like 10x faster than poetry for installs and dependency resolution.
114.
by spott 2y ago
Just. It is a less heavyweight make. Similar syntax and behavior, no .PHONY, a couple of helper functions and behaviors. It is designed as a task runner rather than a build system.
115.
by spott 2y ago
Just curious: what do you use make for in python projects?
116.
by spott 2y ago
Just to be clear, you meant MB of RAM, right?
117.
by spott 2y ago
The FAQ was not part of the challenge statement. It was part of the newsgroup I believe.
118.
by spott 2y ago
You don’t need to loose weight before taking up lifting. Just adjust your expectations.
119.
by spott 2y ago
> most people don’t plan like this. This is a pretty naive take. It is a $85k per year cost. If you can shift some money around and avoid $85k per year, you would absolutely do that. > you could also emancipate your 17 year old. T
120.
by spott 2y ago
He was playing full games, not single moves.

More ›