qwerty3344

Godot is also getting really good

what's lacking about Godot C# support?

fwiw Google did the same trick with their servers

monte carlo or some kind of ant colony simulation probably

I think it's still significantly behind GPT 3.5/4, both of which can get 67% on HumanEval, and 88% with Reflexion

OOOH I've got one - I made a short script that lets me play any video on any site at arbitrary speed (speed controlled by arrow keys).

//@ts-check window.kVideoSpeed = 1; window.initialVolume = 1;

NUMPAD_3 = "99" NUMPAD_2 = "98" NUMPAD_6 = "102" NUMPAD_5 = "101"

UP_ARROW = "38" DOWN_ARROW = "40"

let timeoutId = null; document.onkeydown = (e) => { if (!document.querySelectorAll("video").length) { // no-op for pages with no videos return; } window.initialVolume = document.querySelector('video')?.volume || 1

    let myDiv = getOrMakeDiv();
    e = e || window.event;
    let KEYCODE = e.keyCode;

    // NOTE: can't use left/right b/c those go forward/back 10s on some sites
    if ((KEYCODE == UP_ARROW || KEYCODE == NUMPAD_6) && window.kVideoSpeed < 4) {
        // up arrow
        window.kVideoSpeed += 0.5;
        myDiv.textContent = window.kVideoSpeed;
    } else if ((KEYCODE == DOWN_ARROW || KEYCODE == NUMPAD_5) && window.kVideoSpeed > 1) {
        // down arrow
        window.kVideoSpeed -= 0.5;
        myDiv.textContent = window.kVideoSpeed;
    }
    for (let el of document.querySelectorAll("video")) {
        el.playbackRate = window.kVideoSpeed;

        // prevent volume from changing
        setTimeout(() => {
            el.volume = window.initialVolume
        }, 100)
    }
    if (timeoutId) {
        clearTimeout(timeoutId);
    }
    // set timeout to remove div
    timeoutId = setTimeout(() => { myDiv.remove() }, 1000);

}; function getOrMakeDiv() { if (!document.getElementById("myDiv")) {

        let div = document.createElement("div");
        div.id = "myDiv";
        // add styles to div
        div.style.padding = "8px";
        div.style.textAlign = "center";
        div.style.fontSize = "16px";
        div.style.position = "fixed";
        div.style.fontWeight = "600";
        div.style.border = "1px solid yellow";
        div.style.color = "white";
        div.style.backgroundColor = "black";
        div.style.zIndex = "999999";
        // insert div at the top of the body
        if (document.fullscreenElement) {
            document.fullscreenElement.prepend(div, document.fullscreenElement.firstChild)
        }
        else {
            document.body.insertBefore(div, document.body.firstChild);
        }
    }
    return document.getElementById("myDiv");
}

Yep this is what most data structures and algo questions optimize for.

> last updated 2023-03-19

Yep, anything that old is ancient in the AI world

This isn't true, many open source companies make money in ways that don't diminish the core open source product, Kafka/ElasticSearch/MongoDB/Supabase/Vercel/Grafana/Red Hat/etc

Can someone explain how LangChain compares to Llama Index [1]?

[1] https://github.com/jerryjliu/llama_index

additionally it gets a lot better via iteration, eg. feeding it back any error messages and asking it to try again.

> it's not got a single coding question I've asked it right

Have you used common prompting techniques like CoT and self-reflection?

just use the API which has no limit

using chatGPT to summarize:

This section summarizes the issue with the ChatGPT service which caused some users to be able to see another user's chat history and payment information. OpenAI took the ChatGPT service offline and published technical details of the problem. The bug was found in the Redis client open-source library and has been patched. OpenAI has reached out to affected users and improved their systems to prevent this from happening again. They have also expressed their commitment to protecting user privacy and their dedication to the open-source community.

Would love to see a side by side comparison of the completions from this with completions from Copilot

There is huge variation in how well people can prompt LLMs, prompt-engineering has many tricks that aren't obvious, that's why there's prompt engineers,

There are entire sectors of the economy that LLMs can't touch - hospitality, manufacturing, caregivers, religious sectors, live-action entertainment, etc. Sure some of these will be replaced by robots but there will always be new jobs too.

gaming wise seems the same as Ryzen 5 7600X which is <1/2 the cost?

would humans not make the same mistake?

What are the cons of 0xFast vs Alchemy?

what are the other core modules near the top?

In the simulation you can see very common helper deps do receive quite a disproportionate amount of funding relative to the effort to maintain though. For example, `rimraf` (~400 LOC), `del` (~100 LOC), `cross-env` (this one is archived) all receive a ton of funding, whereas major projects like `electron` get a small amount since it's not used as often even though it would be much more impactful if it were to fall into disrepair.

https://news.ycombinator.com/item?id=32841229 touched on this also, auto allocation would lead to disproportionate allocation for basic helper deps

> to clearly see how the charities use the funds

https://www.charitynavigator.org/ is a start

doesn't include equity though which is usually a large chunk of comp

rope - https://en.wikipedia.org/wiki/Rope_%28data_structure%29 skip list - https://en.wikipedia.org/wiki/Skip_list

I think it'd be nice to have a way to sort/filter by locations and also by salaries

Not really surprising given the state of crypto in general.