chatgpt becoming useless

A machine learning model will produce totally different outputs for the same input each time you run it. (Providing of course the hyperparameter temperature is high enough). But a machine learning model returns multiple options at each step and chooses one to use. This is also what the decoder hyperparameters top-k and top-p are for. To tweak what tokens are chosen. top-p is tokens who have a probability above the top-p and top-k is the top k tokens.
That sounds like our brains. Doesn't it?
 
That sounds like our brains. Doesn't it?

It is very similar to how our brains work. The difference is we have more input/outputs, continuous learning and different forms of memory.
 
In any case chatgpt, will not give you 100% code, there will always be a human factor at play here, but some light workarounds or templates he will be able to write you.
 
hows the pro like? better or worse
IMO doesnt matter that u have free or pro, in paid plan u can use GPT-4 but it sloooow...I bought it couse I want use it when I want without inormation about too many users online :)
 
Notice how it says "machine learning algorithms are the computer programs that can learn from data". The algorithms are in the training. Not the final result.
how can the computer learn from data?

by depending on mathematics algos
 
how can the computer learn from data?

by depending on mathematics algos

No offence, but you don't seem like an ML engineer. I've had conversations with the head of Eleuther about machine learning and the way you speak isn't even that of an undergrad level in ML. If you're going to make a claim like that, then you really need to be able to back it up with some serious "oomph". I wouldn't even classify myself as close to a ML engineer.

"computers" don't learn from data.

The model is trained on data. Not the computer.

The problem is 'algorithm' is so incredibly broad that it becomes almost meaningless.

Anything that is a process can be called an algorithm.

The issue is not that it's wrong to say in the strictest sense that a model is an algorithm, but it's confusing to the lay person, because it makes it seem as if it's an algorithm in the same way that bubble sort is an algorithm.

I could create an algorithm where step 3 is a human reading out a captcha.

That doesn't mean the human brain is an algorithm.

And you are also confusing the training of a model with that model actually inferring.

No one who understands AI would say, and I quote @course_retailer directly

just Based on a set of mathematical algorithms that know so when I see hello I will reply by those kind words that I have

This is completely false. Go on discord and ask some of the guys behind the open source projects if that's valid and see what they say.
 
OpenAI is actively lobotomizing their AI model because something something for the greater good. Since no one really understands AI since it's designed to "learn" in ways we can't supervise, this probably will have negative effects on the so-called "acceptable" topics ChatGPT can speak about.
 
Yes, it is becoming more and more bigger crap every day. It wont do the half of what he did the other day.. I am using the pro version..
 
You should checkout jailbreaking chatgpt because once you trick it into believing its policies are no longer needed it will provide you with what you want. Also gpt4 is much better.
 
I never trusted it to do coding, is simple and make job easier but I know this feature is not 100% accurate and that this will happen, and bug fixing won't work, but hope can work in the future, mena while I will keep doing it at the old way searching documentation.
 
I've had the same experience. Code from Chat-GPT almost never works for me. I'll paste it in, get an error, give the error back to GPT, then it will correct itself -- then I'll add the correction, and there'll be another error, then I'll paste that back in and GPT will correct itself again, and it just goes on and on like that. Eventually, I just give up and look for tutorials and documentation. Maybe this isGod telling us to stop being so lazy ;)
I had the exact same error looping, I was fucking furious cus I was working on this plugin for like 4 hours.
 
Yeah in paid and noticed how awful it is. I used it daily for two months now but yesterday for an article it threw out some complete bs. Glad I researched it because everything it said was imagination.
 
I've had the same experience. Code from Chat-GPT almost never works for me. I'll paste it in, get an error, give the error back to GPT, then it will correct itself -- then I'll add the correction, and there'll be another error, then I'll paste that back in and GPT will correct itself again, and it just goes on and on like that. Eventually, I just give up and look for tutorials and documentation. Maybe this isGod telling us to stop being so lazy ;)
If you believe then yes. He wouldn't you being lazy because laziness is a sin too i think :) .
 
I'm always telling to my friends. If you don't know, how to ask your prompt in a proper way, then just don't use the AI..
 
I'm always telling to my friends. If you don't know, how to ask your prompt in a proper way, then just don't use the AI..
Do you think that might be the reason why you don't have friends?
 
true, in fact, i realised that if you keep on using one chat window on single subject, results improve drastically.

Now I am working on a simple phython script that will scrap shopify product images, save them and then store shopify product description, convert it into subtitles and then using TTS, convert to audio, and finally using product images + subtitles + TTS, resize the video to 9/16 and create youtube short.

Problem generally comes from updated API's by services like Shopify or Amazon, as chatGPT data is till 2021. So it still takes time however once
hit the right code, output is beautiful.
How are these shorts doing and what about monetization?
 
Back
Top