@MarcelioDA

Problem exist between keyboard and chair.

Indonesia
Joined August 2011
The first security updates for BlankOn Linux Sinambung. πŸ˜‡ blankonlinux.id/en/news/firs…
6
14
496
I try to open it, why are you doing this? 😑😑😑
I got early access, and I built a complete invoicing software with a single prompt using Gemini 4 Argon. πŸ˜‚ I'd show you, but it's running locally on "localhost:3000" right now.
1
2
22
Jadi inget pernah lempar API mati ke AI (udah generate baru, yang lama jadi placeholder) trus AI-nya bawel ngingetin berkali-kali kalo API-nya terekspos. Iya tau, itu sengaja, bawel amat, tapi makasih udah ngingetin, baik amat.
WOILAH ADA2 AJA INI ORANG UPLOAD API USERS KE CLAUDE CHAT BUKAN KE ENV OH MY GOD 😭😭😭😭😭😭😭😭😭😭😭😭😭😭😭😭😭😭
20
Nonton Teletubbies loh ya πŸ˜‚
🚨 Ditanya soal banyaknya komentar yang menuduhnya berpihak ke salah satu pembalap di Mercedes, Toto Wolff sebagai CEO & Tim Prinsipal coba berikan tanggapan. πŸ—£οΈ: "Siapa pun yang berpikir seperti itu, mereka tidak punya otak. Tidak masuk akal, apakah mereka mengira saya punya tombol yang bertuliskan 'rusak mesin George sekarang' atau saya sengaja merugikan satu pembalap padahal kami sedang berjuang untuk gelar konstruktor? "Orang-orang ini benar-benar menghibur saya ketika mereka menghina tim atau saya sendiri. Katanya, saya itu setan. Mereka sebaiknya daripada menonton F1, mending tonton Teletubbies atau semacam Mickey Mouse. "Kami selalu punya pola pikir untuk menilai semua orang dengan material yang setara. Tapi saya kira, beginilah memang media sosial sekarang. Nol pemahaman dan nol rasa hormat bagi orang-orang yang mengarang hal-hal seperti ini."
197
Marcelio retweeted
Weird that Ben Affleck uses python. I always saw him as more of a .bat man (if you are under 40 you probably won't understand this anyway so sorry about this)
Ben Affleck do you use uv?
167
239
29
4,110
71,146
Using Nano Banana 2.1 on @FlowbyGoogle
πŸ€– Made with AI
introducing @NanoBanana 2.1 🍌 our latest image generation model outperforms our previous models across the board: - better visual design - mask-based editing - subject consistency - more natural-looking images try it out today in ai.studio
85
I thought it was a new model update πŸ™‚ @antigravity
28
baru juga sekali prompt.
Kita coba kali ya πŸ˜‹
2
35
Kita coba kali ya πŸ˜‹
1
2
66
😌😌😌
Gabakal. 1. Masalah teknis. Harus bisa FIA Grade 1 dulu utk bisa host F1. Dan itu prosesnya LAMA & MAHAL. 2. Jadi host F1 = multi-year contract. Ada yg mau bayarin ngga? Mandalika tiap tahun aja ribut duit mulu. 3. Kalo ada yg bayarin, pas off-race secara bisnis sustain ngga?
1
65
If you hate something, please at least learn about it and try to understand it. Sometimes hate comes from not knowing something. If you still hate it, at least it's informed hate rather than just blind hate.
2
1
2
52
While hate can be subjective or a matter of taste, I don't like misinformation, exaggeration, or simple-minded ways of thinking when tackling things.
10
Plenger ya emang lu Gemini! 😑
27
Spent the day messing around with LLM fine-tuning just to see how it works. Tried turning Llama 3.2 3B into a mini RPG combat engine. Resolving melee actions with narrative text while updating character HP/stamina in structured JSON.
1
1
140
Trained for 3 epochs (~4 mins), loss went from 3.11 down to 1.08. Tested it on an unseen action (spear feint vs goblin warrior). It didn't break formatting, tracked realistic damage, burned stamina, and gave a bleeding status effect.
1
1
18
Just a hobby learning project to understand the tooling, but fun to see a small 3B model handle dual-output so cleanly without overfitting. Pushed the LoRA weights + details to Hugging Face: huggingface.co/marceliooo/rp…
1
18
Because Google was never about the app. It's the ecosystem, the infrastructure, the trust, and the years of advantage you can't prompt your way into.
If AI can build any app, why hasn't anyone vibe coded the next Google?
21
I have a sudden urge to learn to fine-tune a model.
1
16
Jaman kuliah wkwkw
12