hacking 1000 articles.

[crnack]

1000 articles should make your $$$. Good luck...

Ain't gonna do it without well programmed ML. It looks like not many on this forum are interested in ML in general.

 

[M4DM4X]

Yep you still have to push some buttons manually if you use the scrapebox addon to generate unique content, it's not 100% automated.
There is no better thing than coding your own bot if know how to do it.
Concerning the garbage spinned content, it's not really a problem since google loves trash

 

[crnack]

Yep you still have to push some buttons manually if you use the scrapebox addon to generate unique content, it's not 100% automated.
There is no better thing than coding your own bot if know how to do it.
Concerning the garbage spinned content, it's not really a problem since google loves trash

He is already on machine learning, so I don't think scrapebox can beat that.
But I also have hard time figuring out how he wants to beat up the GPT-3 which is trained like a monster.

The best thing I found is this https://huggingface.co/EleutherAI
And it's not very accurate when it comes to writing long text.
Unless you train it yourself somehow.

It tends to repeat itself and is almost retarded. There is not much difference between this and conversion.ai. Of course there is some, but both go stupid when you ask them to produce more than 3 ideas.
Put any technical subject and they will just write science fiction.

 

[crnack]

@tazarbm you still laughing with your hidden set of technologies. And we still do our homework. Damnit.

 

[tazarbm]

@tazarbm you still laughing with your hidden set of technologies. And we still do our homework. Damnit.

I'm not getting your points. Can you clue me in?

 

[crnack]

I'm not getting your points. Can you clue me in?

This was my specific sense of humour. I know you use AI assisted content and gave me a laughing emoji on my post about garbage content. I concluded you must have something better, so I replied in this specific way.

 

[tazarbm]

This was my specific sense of humour. I know you use AI assisted content and gave me a laughing emoji on my post about garbage content. I concluded you must have something better, so I replied in this specific way.

oh, I see.... I gave you a smiley emoji because your last paragraph gave me a chuckle

And no, I don't have anything better. If I did, everyone on this forum would find out about cause I'm not very good at holding things to myself.

 

[crnack]

oh, I see.... I gave you a smiley emoji because your last paragraph gave me a chuckle

And no, I don't have anything better. If I did, everyone on this forum would find out about cause I'm not very good at holding things to myself.

Ahhh, and I was hopeful to produce that million characters of human like content in 2 days. Hope is for naive. Let's just wait for gpt-4 and less computational power required to use it.

 

[tazarbm]

Ahhh, and I was hopeful to produce that million characters of human like content in 2 days. Hope is for naive. Let's just wait for gpt-4 and less computational power required to use it.

I think you are mistaking me for someone else cause I never said anything about 1 million characters. And if I did, I honestly, genuinely don't remember....

But I do want GPT-4 to be a thing

 

[crnack]

I think you are mistaking me for someone else cause I never said anything about 1 million characters. And if I did, I honestly, genuinely don't remember....

But I do want GPT-4 to be a thing

No, I don't mean to shot you. I have no weapon in hands. I meant to say I could conclude there is possibility to do it due to what I read recently. Time to take a rest.

 

[feralAddict]

He is already on machine learning, so I don't think scrapebox can beat that.
But I also have hard time figuring out how he wants to beat up the GPT-3 which is trained like a monster.

The best thing I found is this https://huggingface.co/EleutherAI
And it's not very accurate when it comes to writing long text.
Unless you train it yourself somehow.

It tends to repeat itself and is almost retarded. There is not much difference between this and conversion.ai. Of course there is some, but both go stupid when you ask them to produce more than 3 ideas.
Put any technical subject and they will just write science fiction.

I've found T5 to be a very flexible all-purpose model. Also, BART is very good for paraphrasing.

 

[feralAddict]

Though i agree with the 80/20 approach, i disagree on optimization. A 1% improvement compounded daily really adds up. Thats how giants are born, and as the saying goes "big things have small beginnings".

ceteris paribus. And that's the flaw in your argument. Let me unbox my thinking a little on this.
All things being equal, 1% with daily compounding does add up to a lot. However, all things are not equal; they aren't held constant.
Incremental optimization is generally a lot easier to perform than magnitudinal leaps in gains. Therefore, it's rational to assume your competition is also pursuing incremental improvement. So you end up burning through capital in some inefficient little fight with your competition to see who can pull a little ahead of the other.
The better plan is to outright squish your competition and establish a monopoly. This is only achievable through big jumps. Google, FB, BTC, Tesla, Apple, AWS, MSN, OpenAI > all big jumps.
1% compounds ok. 1000% changes the world.

 

[feralAddict]

Hey BHW. Just want to thank you all for following this thread. Once I put this algo in production, I will be happy to shout you all an article.

 

[WorkingNow]

ceteris paribus. And that's the flaw in your argument. Let me unbox my thinking a little on this.
All things being equal, 1% with daily compounding does add up to a lot. However, all things are not equal; they aren't held constant.
Incremental optimization is generally a lot easier to perform than magnitudinal leaps in gains. Therefore, it's rational to assume your competition is also pursuing incremental improvement. So you end up burning through capital in some inefficient little fight with your competition to see who can pull a little ahead of the other.
The better plan is to outright squish your competition and establish a monopoly. This is only achievable through big jumps. Google, FB, BTC, Tesla, Apple, AWS, MSN, OpenAI > all big jumps.
1% compounds ok. 1000% changes the world.

You're using survivorship bias, sure google and others changed the game but they stay relevant through incremental improvement (and incremental failure, see google+, google hangouts, soon to be stadia ect).

You cant innovate without throwing lots of stuff at the board and seeing what sticks. Imagine if google poured all their resources into g+ only to have it fail, they would be out of luck and out of cash fast.

Its better to pivot quickly and iterate based on data, the best way to do this is to not release a 10X / 1000% product, but to release a MVP (10% product) and allow the users and data to steer your iterations. This is how a 10X happens, through data driven iteration. Failures, espcially in tech typically happen because they do what you are suggesting and just press on in the dark on useless features or services no on wants or asked for.

There are edge cases, but the above is tech entrepreneurship 101. Smart people learn from the (well documented) mistakes of others, dont try to reinvent the wheel on SOP's focus your energy elsewhere. If you truly believe in executing on the 80-20 rule, then doing the above is a none choice.

 

[blogchat]

I've found T5 to be a very flexible all-purpose model. Also, BART is very good for paraphrasing.

brother, WHERE are you pulling these names from, and all this jargon. it's insane how many random ass tools you know. Please elaborate thank you

 

[crnack]

brother, WHERE are you pulling these names from, and all this jargon. it's insane how many random ass tools you know. Please elaborate thank you

https://huggingface.co/models
Not too hard to find them.

 

[blogchat]

Won't google be able to detect my quillbot spun content, even if I hand-edit it and do paragraph by paragraph mixing multiple researched articles?

 

[4dbrainarray]

Hey BHW. Just want to thank you all for following this thread. Once I put this algo in production, I will be happy to shout you all an article.

So, I can order content from you? What information you need to start?

 

[feralAddict]

So, I can order content from you? What information you need to start?

Hi I would love to but Im not allowed to sell on BHW. I will ask the admins if Im allowed to give out other contact details or give some content away for free to people who have been following this thread.

 

[feralAddict]

brother, WHERE are you pulling these names from, and all this jargon. it's insane how many random ass tools you know. Please elaborate thank you

I have 2 decades worth of experience in ML and AI development. While Linkedin may think all Im good for is being a Dominos pizza driver, I feel I have a little bit more knowledge to share