Yoshimasa Niwa

なるほど！

これ日本語にすると見る人が変わるのかな？それとも、単なるタグ？

Morning!

All stores are close today. Obviously.

Apple seems going to deliver a device a week prior to the estimated date…

Rainy Monday morning.

It’s raining and seems like this for a week.

Freezing!

M1 Max MacBook Pro was ~$4,000 but M4 Max MacBook Pro will be likely ~$6,000, thus $2,000 for inflation of last 3 years! Well, actually it might be true at some degree but, doubled memory, storage and speed.

For training tasks… watt cost can exceed by renting GPUs somewhere in other locations on the earth for sure!

Hahaha, per watt aka per PG&E bill, yeah.

YouTube video by Luke Miani DON'T buy an M4 Max MacBook Pro...

Based on this review, the degradation of the 14 in. MacBook Pro is very limited, slightly slower than 16 inches yet considering about the mobility I would consider to buy 14 in. Any thoughts or experiences you had about for example, thermal throttling?
m.youtube.com/watch?v=SdJv...

M4 Max seems having about x2 performance to M1 Max in terms of TOPS, which is significantly slower than RTX4090 for example at about 40% of it, which is similar to T4, yet I think it’s good for local inference and for development of applied applications.

Morning!

Use `var` in `struct` in Swift Use `var` in `struct` in Swift. GitHub Gist: instantly share code, notes, and snippets.

I wonder how Bluesky can reach now to my friends compare to the others so I am posting same things I posted on the other network lately —
Don’t use “let”, use “var” in struct in Swift (mostly) always!
Curious why? read this. Can’t agree? let’s discuss!
gist.github.com/niw/f41b83aa...

Hello!

Meow

Meow

Meow

These two facts makes operating Twitter or similar platforms much difficult than before. If the platforms make it more easy to access, it would make it easy to scrape, and the motivation to do so is seriously higher than before.

Therefore it requires to implement sophisticated anti-scraping logic in the system to identify if it’s monetizable requests or not by using multiple signals always for any requests, which would costs some.

Second, it’s very difficult to identify which traffic is from valid monetizable users or not. Technically it’s always possible to create unidentifiable scraping software from the original software. Any API tokens can’t be hidden because it’s needed to be in the request for example.

It is also true for data miners. If platform has such value, as same rate, motivation to scrape data is increased.

There is the unique contents therefore users is coming and conversation evolved. Thus they can show ads or relevant contents to the users to get revenue from it.

First, in recent ML hype and ongoing trend, we all know that the content is king. Therefore Twitter or many other user generated content platform is a gold mine, which is true for even one. And the service and business goals are on that value.

Meow

Meow

Meow meow meow

Meow

Meow