Independent Web3 Creator. Watching projects, people & the next wave.
The Web3 Frontier
Joined September 2021
- Tweets2K
- Following1.7K
- Followers774
- Likes8.1K
The King 👑 retweeted
GM, Onsen Fam! ♨️👻
Starting the day with a spooky new 1/1! Inspired by @BooXRPL bringing some ghostly vibes to the Onsen kitchen. 🕸️🔥
Different day, different inspiration. We never stop cooking!
The King 👑 retweeted
Việc whale trên Hyperliquid đảo từ long sang short BTC sau bảy tháng là tín hiệu đáng theo dõi, nhưng chưa đủ để kết luận xu hướng giảm đã hình thành.
Cần phân biệt vị thế đầu cơ với hoạt động phòng hộ. Nếu short tiếp tục tăng cùng áp lực bán giao ngay, tín hiệu sẽ đáng lo hơn.
Theo Glassnode, các whale trên Hyperliquid đã chuyển từ long sang short BTC sau bảy tháng.
Lượng short hiện ở mức cao nhất kể từ tháng 10/2025. Đây là sự thay đổi đáng chú ý sau thời gian dài giữ vị thế long.
Việc đảo chiều ở quy mô này thường phản ánh thay đổi trong kỳ vọng ngắn hạn của nhóm lớn. teapawt.net/Cointelegraph/status/2…
The King 👑 retweeted
gm☕️ do you still support ordinals?
LETS TALK ABOUT IT 🗣️
Drop your take 👇 (••)
Attitude set to gratitude
Gm ☕️ Gmw, Guki, Izza gmdoe(••), bj, GMB, !RIBBIT, bm📐
Do you still support Ordinals?
Let’s talk about it 🗣️
Personally I support real communities. It’s @ftw_collective or the ecosystem is already doomed.
Weirdos called out extractions, got laughed at for it. Built when everyone left. Show up EVERY DAY. And other communities still fud us.
I get that others feel threatened by our direct and real attitude which cause push back. Now adays it’s the standard to lay down and don’t speak up when you see / feel somethings wrong.
Weirdos are comfortable being uncomfortable. Weirdos are used to the pressure and used to standing out.
A weirdo created bitcoin, a peer to peer electronic cash system. Did you study? What don’t you understand.
People don’t like change. So they settle. And as of now ordinals have settled in the world of corruption and lies.
The problem with Ordinals is the current standard.
#raisethatbar
Weirdos are building different!
#itsnotthesamerace
Saying FTW as it is and creating one to thrive 🌎
For the Weirdos
The King 👑 retweeted
My lovely AI artists 💪💀
If you want to be recognised as an artist, keep experimenting, improving & developing your own vision maybe u can add little masala into it 💀
Please don't settle for the first result, a little experiment, refine, edit & some of your own ideas, build something people can recognise as your style.
No hate but please don't let AI do all the thinking for you 💪💀
The King 👑 retweeted
Sunday CT debate 🐱🐶
What gives a meme staying power after the launch hype fades?
😂 Actually funny
🎨 Recognisable character
🫂 Community that keeps creating
Pick one. No ticker shills, defend your answer 👇
The King 👑 retweeted
That’s so spectacular congrats on another Grail
One day I’ll upgrade to an Ape
This Ape is a reminder for all of us, no matter what, to always stay classy ❤️
Special thanks to @Cymbalda for entrusting me with this amazing Trippy grail. Appreciate you brother……
The King 👑 retweeted
free mint WL check
• @JPEGZart
• @themochieffect
• @giraffies_nft
• @rarelistio
• @FeedTheFireRH
• @Jpegsaredead
• @webhoodr
• @slymzcrew
• @Oxustrail
• @egglorh
0/10 😭, how bout you?
free mint WL check
• @JPEGZart
• @themochieffect
• @giraffies_nft
• @rarelistio
• @FeedTheFireRH
• @Jpegsaredead
• @webhoodr
• @slymzcrew
• @Oxustrail
• @egglorh
0/10 😭, how bout you?
The King 👑 retweeted
Lmao, all that money and GOOBR still needs a condo and a T-shirt 😭😂
#goobrgamdom10k @TheGoobr @gamdom @stevewilldoit @kingadukes
🤖 Made with AI
$10,000. 24 hours. funniest GOOBR meme wins.
promote GOOBR + gamdom with an indian or egyptian theme.
the storyline: nobody knows what race GOOBR is. we thought his balance was fake. turns out he’s the most savage gambler ever. now he needs to buy a condo and a fucking t-shirt with his winnings.
post your memes using #goobrgamdom10k. you have 24 hours.
@TheGoobr @gamdom
The King 👑 retweeted
Before you raise your LoRA rank, check which layers LoRA trains.
Leave target_modules unset in Hugging Face PEFT, and on Llama or Qwen3 it adapts only q_proj and v_proj.
That's 2 of the 7 linear layers in each block. The MLP layers get no adapter at all.
The QLoRA paper found the number of adapted layers mattered most. In their LLaMA tests, matching full fine-tuning needed LoRA on all linear layers. Rank r made little difference.
Try:
target_modules="all-linear"
It trains more parameters, so watch memory. Run print_trainable_parameters() to confirm the change.
