hackmaxim

Launchyard is hosting a contest with $1000 in payouts for whoever makes the most viral website by April 24.

Instructions: 1. Give Launchyard your prompt, have it build you something (takes five minutes) 2. Market your website on socials (reddit, hackernews, etc) 3. Top 3 websites by visitor count with $600, $300, and $100 respectively!

See the attached link for more detailed rules. We already have some good submissions: 1. https://humanorclanker.launchyard.app (249 visitors) 2. https://replaced.launchyard.app (109 visitors) 3. https://admitit.launchyard.app (22 visitors)

I'm Maxim. Launchyard is a tool that I built that enables you to vibe-code an entire business. Launchyard will create an app, deploy it, purchase a custom domain, SEO monetize, add advertisements + stripe, etc. I moved my online business to launchyard and now get ~200k clicks/mo on Google. No engineering or technical work needed!

It's crazy how quickly you can deploy end to end apps today. With launchyard it takes hours, and the only bottleneck is token inference. Quality is genuinely on par with some of the best landing pages. Try it out and let me know what you think!

Something like this is reasonable. To clarify, we can't support arbitrary tasks yet on our submission platform; the tasks have to belong to a set of supported repositories. But tasks of this difficulty level generally tend to be hard enough.

You can't know for sure what is going to be the right difficulty in advance; however, you can definitely develop intuition for it as you make more tasks and better understand the model's strengths and weaknesses!

Nothing is free, of course, since we transparently pay per task (and as previously stated, the pay is very generous). There are certainly guidelines for what kinds of tasks you can submit to our platform, but beyond that, you can make any verifiable coding task in any of our qualifying open-source repositories.

We also definitely do not require you to write new code into the repo. As I said, you can adapt existing merged commits into coding tasks.

Yes, to be perfectly transparent, this is for a contract work position! However, many of our contributors appreciate the flexibility and transparent payment structure. If you are exclusively looking for full-time roles, this is not for you.

Yes, of course!

Tasks must be coding problems to complete in a set of open-source repositories. We have eight repositories now and will add more in the future. You can source tasks by identifying merged commits into these repos, and making the following: - A well-defined task description (what should the AI do?) - A golden solution (can be sourced by the implementation in the merge commit) - A test patch (can be sourced from the merge commit, this is a testing suite that verifies whether the AI's solution is correct).

If you can make a task that is hard enough for our AI, you will get paid a fixed amount. That's it!

Once you are approved onto the site, there will be a more detailed tech spec + payment info + a tutorial.

Habitat Inc - We pay $100-300/hr, remote.

To apply: https://code.habitat.inc/signup. Our company: https://habitat.inc, LinkedIn: https://www.linkedin.com/company/habitat-ai-inc

If you can make a coding task that our AI cannot solve, you will get paid. It is that simple. Our best contributors make more than $200/hr, but the pay depends on how effectively you can find and make these tasks!

Our company sources coding data that reveals weaknesses in the current generation of models so that the next generation of models are trained to not make these mistakes.

We will be paying out more than hundreds of thousands of dollars in bounties over the next few months.

To apply: please sign up for an account on https://code.habitat.inc/signup - and mention that you came from HackerNews!

Looks like you can now use LLMs to translate all languages, better than Google.

I'm one of the authors on the paper. Actually, sentence-by-sentence translation is important in a machine translation system because in many cases users will only provide single sentences. We also test document-level translation in Section 5, and find large improvements (but it isn't the focus of our paper).

Although, the paper argues that we can create artificial datasets using the LLMs which would improve the special-case translation models without having to pay for the inference and latency time of the large models.

Yes, but the translation model in the “Attention is All You Need” was trained on parallel sentences (input: English sentence, output: French). LLMs such as Claude are not trained on any labeled data. Yet, they still surpass the special-case models.