固定されたツイート
Vangsguard
2.7K posts

Vangsguard
@Vangsguard
19-year-old programmer and game developer from Denmark
Infinity Castle 参加日 Mayıs 2020
432 フォロー中273 フォロワー

@Vangsguard Love your trading insights, curious, which market are you focusing on right now?
English

@AmasPFT Moved from CFD to futures recently, what is a trade copier you can recommend?
English

@zeddotdev @theCTO @cursor_ai Took me over a week to get a response from your billing team about double payment
English

@theCTO @cursor_ai Happy to engage we have a very active discord and email support. Not sure what this is about specifically but happy to take a look
English

honestly, I'm thinking about going from @zeddotdev to @cursor_ai .
I can't use a tool where I don't have access to the team.
If your team isn't reachable - it's not a good look.
English
Vangsguard がリツイート

its 2027.
after 7 months of grinding, little timmy has finally reached grandmaster tier at 5000 LP. he has hopes to once achieve challenger, which is 10k LP
his mom walks in, sees the screen, and says:
“wow sweetie, 5000 points? you must be like… top 10 in the world!”
timmy gulps.
he tabs over to the ladder.
rank 1: 47230 LP
rank 10: 31000 LP
rank 300: 10000 LP
challenger cutoff: 10001 LP
timmy checks his last game:
+15 LP for a win.
-26 LP for a loss.
he does the math.
“so if i just…win 90% of my games before the season resets, i might be able to do it…”
he queues up.
his jungler is autofilled.
his mid is on a 17-game loss streak but still somehow 20k LP.
little timmy contemplates whether he wants to continue to queue up this season after this game or call it quits.
English

@Cooter @Sanchovies no excuse not to watch reels on stream anymore
English

absolutely cracked lol
Victor_@AJR_despiser
@Cooter I will do you one better. Install the blindfold plugin so it only shows green agility obstacles and everything else is black. Set runelite opacity under regular runelite setting to something like 40 and you can just overlay game on full screen video
English
Vangsguard がリツイート

Everybody who had a friend addicted to weed knows this is not true btw. Evidence is your own eyes
Curiosity@CuriosityonX
BREAKING🚨: Research finds NO evidence of long term cognitive decline linked to cannabis
English
Vangsguard がリツイート

i’ve met so many people who smoke weed to know that this is the biggest lie ever told
Curiosity@CuriosityonX
BREAKING🚨: Research finds NO evidence of long term cognitive decline linked to cannabis
English
Vangsguard がリツイート

@kevinfxtrading @astucefx Again, this is a question. I would love to hear if my thought process have faulty logic
English

@kevinfxtrading @astucefx But that's my point. The only reason you would choose a faster payout is the risk of losing said profits before you can withdraw. In which case should you even be trading if you don't trust yourself & risk management enough to wait 2 more weeks?
English
Vangsguard がリツイート

PewDiePie didn’t “train his own LLM.” He fine-tuned an existing open-source model on coding benchmarks. His model started at 8%, crawled to 16% after format fixes, and one run hit 19.6% that briefly passed GPT-4o on a single benchmark before he couldn’t consistently reproduce it.
The tweet makes it sound like a YouTuber casually built a frontier lab in his bedroom. What actually happened is more interesting: a guy with a $41,000 home rig of 10 GPUs and 424GB of VRAM spent months failing, retraining, and iterating on dataset quality until he squeezed marginal gains out of a fine-tune.
This is the part worth paying attention to. The entire arc from October 2025 to now tells you where AI tooling has actually landed. PewDiePie went from building his first PC to running Qwen 235B locally, vibe-coding a custom chat UI, orchestrating multi-agent voting systems, and now fine-tuning models on custom datasets. He did most of this through AI-assisted coding itself.
The video is literally called “I wish I never did this project.” He’s documenting how painful and tedious the process was. That honesty is the signal. The hype accounts strip that away and replace it with “what the f*ck, YouTuber beats DeepSeek.”
The real takeaway: fine-tuning on specific benchmarks with curated data can let anyone temporarily spike a score past models that cost hundreds of millions to train. That tells you everything about how narrow benchmark gaming has become, and nothing about general capability. PewDiePie knows this. The people quote-tweeting him with shock emojis do not.
sui ☄️@birdabo
pewdiepie just trained his own LLM. his model outperformed deepseek v2.5, LLAMA-4 and GPT-4o in coding benchmark. what the f*ck.
English
Vangsguard がリツイート

@turborepo is closing in on 10,000,000 hours of compute saved.
That is literally over a millennium of engineering time not waiting for builds. 🤯

English

@wangjuani If we all agree to int anyone who is toxic in voice then the system will finally allow player controlled punishment
I love it
English

@creepydotorg This is if your friends are nice to you. Most often you will get "kidnapped" and tied to a tree 😭
English

@Vangsguard It looks dark. It films bright. The world is about to get more exciting with the all new Galaxy. #GalaxyAI
Register below for Unpacked event for upcoming teasers, updates, and access to exclusive benefits ahead of #GalaxyUnpacked ✨
Reply #stop to opt-out.
English












