examplary_cable

I'm currently using fly.io and I have no complains so far(most of the problems I had were my fault like a DNS/IP/Certification problem).

So they're saving them from a fate worse than prison ... being a java developer.

I had to say it sorry (҂◡_◡)

It's part of the new modern WebDev semantical lexicon:

!React = HTML/CSS/JS

¯\_(ツ)_/¯

> The column layout is basically impossible for me to meaningfully navigate

Same. Somebody said it was done this way because of mobile. I'm not sure.

It's just weird to see "items" horizontally and properties vertically. I'm usually familiar with:

prop | prop | prop

item | value| value

item2| value| value

Awesome list!

I love that the cognitive overhead for web apps now is so great many people are "returning to monke" and coding in simple HTML/CSS/JS haha.

Question: Shouldn't the items be in the row so I can see all of them at once? Why are the items in the columns? I'll open a issue (:

I have made a bash script(using rofi) to use chatGPT if anyone is interested.

https://github.com/ilse-langnar/bashGPT

Interesting Thanks!

I'm still working in it, but that's going to be the link.

Trying to make universal components is hard, like, I wanted to make all components available trough a "filter" where you could re-use React.js, Vue.js, Svelt, Solid components with each other. When you think about it, components are just I/O with maybe some libraries. I'm thinking this is field is right for some standardization.

Thanks

I'm currently working with Alpine.js in order to try to build "Universal Components" or "Hypermedia Components". Where you just hit a URL and get your component like /components/infinite-canvas or /components/svg-2-base64 or /components/lib-somelib.

[1] ilse.ink/components(It's not ready yet)

My example here was silly and I admit. But the point was that this simple task cab become more "nuanced"(Aside from ChatRWVK-raven, no other model quite "works" like Vicuna or "tuned LLama"), it can, given the correct prompt act as someone in a fictional work which might help you learn the language better by increase conversational time(most important metric, I'm talking comprehensible input here) by the virtue of being more enjoyable.

Overall I like the progress: LLama releases -> LLama fine turned on larger models gets similar performance to ChatGPT on lower parameters(more efficient) -> People can replicate LLama's model without anything special, effectively making LLMs a "Commodity" -> You are Here.

(I'm not an expert)

> How can someone get into using these models

You can use gradio(online) or download(git will not download, it's too big, do it manually) the weights at https://huggingface.co/lmsys/vicuna-13b-delta-v1.1/tree/main and then load the model in pytourch and try inference(text generation). But you'll need either a lot of RAM(16GB,32GB+) or VRAM(Card).

> How might I go about using these models for doing things like say summarizing news articles or video transcriptions Again, you might try online or setup a python/bash/powershell script to load the model for you so you can use it. If you can pay I would recommend runpod for the shared GPUs.

> When someone tunes a model for a task, what exactly are they doing and how does this ‘change’ the model? From my view ... not much ... "fine-tuning" means training(tuning) on a specific dataset(fine, as in fine-grained). As I believe(I'm not sure) they just run more epochs on the model with the new data you have provided it until they reach a good loss(the model works), that's why quality data is important.

You might try https://github.com/oobabooga/text-generation-webui they have a pretty easy setup config. Again, you'll need a lot of RAM and a good CPU for inference on CPU or a GPU.

https://huggingface.co/lmsys/vicuna-13b-delta-v1.1/tree/main

You can ultra fine tune those models ... look at vicune 13B, if you know how to prompt it well, you can get it to work as """"well"""" as ChatGPT. Running on local hardware .... I just got vicune 13b on gradio[1] to act as japanese kanji personal trainer, and I've only used a simple prompt: "I want you to act as a Japanese Kanji quiz machine. Each time I ask you for the next question, you are to provide one random Japanese kanji from JLPT N5 kanji list and ask for its meaning. You will generate four options, one correct, three wrong. The options will be labeled from A to D. I will reply to you with one letter, corresponding to one of these labels. You will evaluate my each answer based on your last question and tell me if I chose the right option. If I chose the right label, you will congratulate me. Otherwise you will tell me the right answer. Then you will ask me the next question. Avoid simple kanjis, let's go."

[1] https://chat.lmsys.org/

GPT-GB

Sure, but I think when compared 1-to-1 some people are generally perceived to be better writs ... I'm not sure if GPT-4 would be significantly better than GPT-3 ... might depend on prompt and other issues.

Would you rather have it be a plain text website? No attack, just wondering.

Interesting ... Organizations can also have and inherit psychological traits inherited from the humans that compose them.

I wonder what immortality would do to fertility and demographics ... people dying often "gives" a place so someone new can fill the old place. But if everybody is immortal then the number of people on earth would just keep adding up, even with the demographic stagnation in developed countries.

Probably a simple pokemon-like 2.5D(Isometric) game engine.

The issue you're having is due to the fact that Anki "Accumulates" cards if you skip one day, which can build up and create such a large amount of scheduled cards for a day that you end up dropping the thing. I'm working on a spaced repetition algorithm that solves this issue by letting you review when you have time and letting you skip the days you can't do the reviews.

Can't resolve? Anyone know the IP so we can try it directly?

Parrots can't use sign language, but some great apes can(koko)[0]!

Even thought monkeys don't have the vocal complexity to generate all the vowels(a,e,i,o,u) I think "great ape language" communication is more complex and nuanced than a fricking parrot repeating some phonemes after some form of stimulation.

Fascinating.

[0]: https://en.wikipedia.org/wiki/Koko_(gorilla)

As a loyal cultist, I'll drop here one of the most transformative "educational" or reading technique I have even encoutered: Incremental Reading[0].

I would recommend it to anyone to give it a try, it absolutely changed how I study.

[0] https://supermemo.guru/wiki/Incremental_reading

If this is true, then I believe it will unfortunately get worse. Given the curve of technological progress and social media trends and the overall turn humanity is going right now I believe the future is going to be increasing polarized in terms of happiness. Half of the popular will be extremely happy while the other half will be extremely depressed. Not exactly like this but the psychological well-being bell curve is going to get away from a normal distribution and become more distorted.

I believe the good news is that now they have a way of sustaining the product while giving up a free? service. Does anyone has the link about that article about the marginal costs of inference of LLMs? I wonder how much profitable it is.

Overall interesting times to come. I just hope llama, Alpaca etc will soon be improved, fine-tuned and commoditized.

What about AI collaborating with AI?

Prompt engineers: huh?

[I'm not an expert] but I believe .ckpt and .safetensors.

The problem with .ckpt is that it executes arbitrary code in your machine(very unsafe). While .safetensors was made by huggingface in order to have a safe format to store the weights. I've also seen people load up the llama 7B via a .bin file.