i stopp doing these things, too. I use browser-harness to do all of these things. Always trying to keep myself out-ouf-the-loop.

Impressive work, indeed. vermiceli the GOAT.

For AI usage: It was all done by Claude/Codex, but it wasn't that straight forward. First: i started in April and the models weren't good enough and would make errors. You could see those immediately in glitches. Then, i spent some time in building a harness and collect telemetry data in Mesen. Once, i had that: i just played the game in the emulator and asked Claude Opus 5.5 to use this as verification. The rest was smooth.

I ported it in two steps. First to C: the game logic is translated routine by routine from the 6502 assembly, and checked frame by frame against the original in the Mesen emulator. Then the C code to WebAssembly so it runs in the browser.

The cherry on top: online multiplayer, so two players can play co-op over the internet.

Source: https://github.com/epoyraz/nes-contra-us

Comments URL: https://news.ycombinator.com/item?id=50319169

Points: 90

# Comments: 80

Comments URL: https://news.ycombinator.com/item?id=49247362

Points: 57

# Comments: 2

Comments URL: https://news.ycombinator.com/item?id=48894640

Points: 10

# Comments: 2

I was able to reproduce the results of the original gpt-1 paper with my gaming PC. I don't even have alot of VRAM. My NVIDIA GeForce RTX 2060 SUPER was able to reproduce most of the results with just 1 hour of training. I would totally recommend to do the same, if you are interested in pre-training LLMs.

The code is here: https://github.com/epoyraz/modded-gpt-1 But, you can also just ask Claude 4.8 or Codex 5.5

This is one of the reasons people love using xona: no subscriptions :)

We've been working on this project for the past several months and launched our beta in April. Since then, we have attracted over 10,000 users, and I couldn't be happier with the results.

Xona.ai makes it incredibly easy to generate beautiful redesigns of your home within seconds. Our focus is on usability and design. We collaborate with experienced interior designers, and our customers love the handcrafted design styles we offer. We've invested countless hours in prompt engineering to get the most out of our models. Of course, you can also create your own prompts.

At Xona, we are aware that AI-generated images are not always perfect. To address this, we've developed a set of tools to improve the results. Our features include:

Magic Eraser: Erase small mistakes (similar to cleanup.pictures). Creative Upscaling: Enhance generated images by upscaling them (thanks to clirityai.co). Find and Replace: Easily change materials or colors of items, like flooring or couches. Google Lens Integration: Love the generated sofa? Find similar furniture with Google Lens. Feeling overwhelmed by the options? Don't worry, we provide short tutorials to help you get the most out of Xona.

And we're not stopping here. We are currently experimenting with Virtual Staging. After evaluating competitors' offerings, we believe we have a solution that will surpass everything currently available on the market.

One last note: we dislike subscriptions as much as you do. You can get 4 credits for free, and a single purchase unlocks all the extra features (Magic Eraser, Find and Replace, Upscaling) for a lifetime.

Thank you for your attention, and we hope to see you at https://xona.ai!

Comments URL: https://news.ycombinator.com/item?id=45593636

Points: 7

# Comments: 2

Comments URL: https://news.ycombinator.com/item?id=44046441

Points: 3

# Comments: 5