void *

2.3K posts

void * banner
void *

void *

@jamesmcn

Chicago maximalist, Japan enthusiast.

Chicago Katılım Ocak 2009
1.2K Takip Edilen775 Takipçiler
Airwreck333
Airwreck333@airwreck333·
@jamesmcn @kendrictonn That's also the other thing that sends me down those rabbitholes. you see something in a historical fiction anime, and aside from all the obviously silly things, it's more educational and historically accurate than most educational shows.
English
2
0
1
17
Kendric Tonn
Kendric Tonn@kendrictonn·
Every time I go even slightly out of the way in this country, I always get stuck on "how did anyone ever unify this place? I swear there should still be uncontacted valley tribes in the year of Our Lord 2026."
Kendric Tonn tweet media
English
7
0
73
1.9K
void *
void *@jamesmcn·
@airwreck333 @kendrictonn But that's not so weird because there is another anime that covers the entirety of Japanese history with anthropomorphic cats, and it's actually a decently accurate summary.
English
1
0
1
14
void *
void *@jamesmcn·
@airwreck333 @kendrictonn And then you find out there is an anime where all the warlords are dogs owned by modern young women and they address that specific issue repeatedly.
English
1
0
1
23
void *
void *@jamesmcn·
@headcrablove 私なりの解釈を述べるならば、AE86とは何よりもまず、極めて「平凡」な存在である。そして、まさにこの「平凡さ」こそが、拓海と圭一の内に秘められた非凡な本質を際立たせているのである。
日本語
0
0
1
203
アタマガニ
アタマガニ@headcrablove·
AE86を持ち上げる記事で1番びっくりしたのはCR-X(バラード)はAE86の影に隠れてたというヤツ 当時を知る人に見せたら予想通り失笑してました 曰く1.5でもAE86は余裕で1.3でやっといい勝負 ZCならAE86なんて敵じゃなかったって教えられました AE86を速い車と勘違いするのと別ベクトルのヤバさ…
アタマガニ tweet media
日本語
133
251
2K
125.4K
void *
void *@jamesmcn·
@bryancsk That's probably how they did it back in the Tang dynasty when they first invented fruity matcha.
English
0
0
0
23
void *
void *@jamesmcn·
@cxy I wouldn't say I love it, but the fact that it is recessed allows me to determine the orientation of the phone in my hand (usually) without landing fingerprints on the lenses. Usually when I would use the CCB I end up using my Fuji instead.
English
0
0
0
6
cxy.eth, music nerd
cxy.eth, music nerd@cxy·
Is there a single iPhone user who loves the Camera Control Button
English
2
0
1
103
The Mind Scourge
The Mind Scourge@TheMindScourge·
The fact that DJs are bigger than ever, yet their job can easily be automated tells you a great deal about how AI will work in practice in the real world
English
73
62
1.1K
71.2K
社長
社長@SyAcYoU_JAPAN·
マリブ、見てごらん。綺麗だろう マリブ、死んで天国の花畑を見てるわけじゃない。君が来たここは“美しい国、日本”だ。今の時期は美しい花畑がそこら中にある。 ところで君の生まれ故郷の美しい物ってなんだ？マクドナルド？
社長 tweet media
日本語
45
309
3.2K
102.1K
void *
void *@jamesmcn·
@JapanTripFriend Yes, Sapporo TV Tower. Sometimes the towers hide - Goryokaku Tower
void * tweet media
English
0
0
1
9
KODAI GO🇯🇵
KODAI GO🇯🇵@JapanTripFriend·
@jamesmcn That’s so true! That’s Sapporo TV Tower, right! Thanks for the photo!
English
1
0
1
24
KODAI GO🇯🇵
KODAI GO🇯🇵@JapanTripFriend·
Tokyo Tower or Skytree, which one would you choose? Tokyo Tower: You can actually climb it not only by elevator, but also by stairs. I’ve done both. lol The elevator is definitely easier, but if you want a challenge, give the stairs a try. Inside, you can even play a shuriken throwing game, and the souvenir shops are pretty great too. When I went last year, there were tons of koinobori decorations and it looked amazing. I added a video from that day too! Tokyo Skytree: I’ve been up twice. What I like about Skytree is that you can walk around Tokyo Solamachi after coming down. There’s a teppanyaki restaurant called “Matenro” inside, and the view, food, and atmosphere were all incredible. It’s also the tallest tower in Japan. 634 meters. If you ask which one I recommend… honestly, both are great. They’re in completely different areas, so I’d choose based on what else you want to do. If you’re exploring central Tokyo like Roppongi, go with Tokyo Tower. If you’re visiting Asakusa or Oshiage, go with Skytree. If you have time, you should try both. lol Which one would you pick?
English
13
4
44
2.8K
void *
void *@jamesmcn·
@JapanTripFriend I’m impressed by the number of towers and other viewpoints scattered across Japan
void * tweet media
English
1
0
1
16
KODAI GO🇯🇵
KODAI GO🇯🇵@JapanTripFriend·
@jamesmcn Nice! Glad you had a great time at Tokyo Tower! Thanks for the lovely photos!
English
1
0
1
31
void *
void *@jamesmcn·
@brconaway @Fremond_ Part of why Siam / Thailand was never colonized is they knew what was up, so when the British showed up and asked for books they offered them a glimpse of their Buddhist Libraries and the Brits' eyes kind of bugged out.
English
0
0
0
65
Brian Conaway
Brian Conaway@brconaway·
@jamesmcn @Fremond_ Most of it was probably not yet available in English translation at the time, which is the most charitable take I can give...
English
1
0
1
79
void *
void *@jamesmcn·
@sudoingX I'm happy at the end of the line. Too busy building out the software side to worry about specific models just yet.
English
0
0
0
46
Sudo su
Sudo su@sudoingX·
total vram across all your machines. whatever tier gets the most votes gets the full benchmark breakdown . every tier gets covered but you decide what comes first.
English
39
2
57
8.5K
void *
void *@jamesmcn·
@Msamalam Actual Japanese history is much stranger and more interesting than the vast majority of western conspiracy theories.
English
0
0
1
35
Rokko🇯🇵🌐💻 ✈︎
Rokko🇯🇵🌐💻 ✈︎@Msamalam·
Cool ad Quick note: contrary to popular belief,⚾️ was popularized far before post-WWII US occupation. Japanese have been ball-loving ameriboos since the 1890s.
BaseballHistoryNut@nut_history

Major League Baseball is aired in the morning for Japan. So technically they eat breakfast with it being on television. Here’s their #openingday commercial. No hyperbole, when I say this, it might be greater than any US MLB commercial I’ve seen. Well done and worth the watch for any baseball fan.

English
4
0
20
1.1K
Sudo su
Sudo su@sudoingX·
i'll stay within my confidentiality and tell you this. stop renting H100s by the hour before you know what you're solving. if you throw compute at a data problem you will burn through funding before product discovery. if you throw data at a calculation problem the end result is same. knowing which one you're facing is the difference between a $200 test on a 4x RTX 3090 and a $20K cloud bill that teaches you nothing. i have tens of cases in my DMs. companies running sota models on enterprise clusters for workloads that could have been tested and validated on 4x RTX 3090s first. the answer was always upstream. wrong evaluation, wrong data mix, wrong base model. not wrong hardware. think sir the order matters more than the budget.
Sudo su@sudoingX

companies are blindly burning thousands $ on fine-tuning before they even know if the base model can do the job. i've seen it firsthand. someone finetunes a large model on enterprise GPUs and the result matches a base model they could have tested for free. the smarter path should be test on a single consumer GPU first. run your actual workload on a 3090-5090 with the base model. find where it breaks. find where it's already good enough. then finetune only the gaps. then scale compute. if your finetuning is the bottleneck, throwing more hardware at it won't fix it. the problem is usually upstream. a $900 GPU and 2 hours of testing would have told you that before you burned $10K in compute. i test models on consumer hardware every day. the number of times a well chosen base model at the right quant outperforms a fine tuned model on 10x the hardware would surprise you.

English
10
5
99
7.6K

Keşfet

@airwreck333 @kendrictonn @egeberkina @headcrablove @bryancsk @cxy @Benmsx @jedpolglase