Cam101's AI Tweet Generator: Master Your Social Media Game

[Official_Lovense]

Are you struggling to keep your social media content fresh and engaging? Look no further than Cam101's AI Tweet Generator!
Still in its beta phase, this powerful tool can generate custom tweets tailored to your audience. Say goodbye to social media stress and hello to effortless, on-brand tweets that will boost your online presence and keep your fans coming back for more. With Cam101's Premium Tweet Generator, you can generate up to 10 tweets day, so you'll have more time to focus on creating killer content and connecting with your audience.

 

[KingMarti]

Are you struggling to keep your social media content fresh and engaging? Look no further than Cam101's AI Tweet Generator!
Still in its beta phase, this powerful tool can generate custom tweets tailored to your audience. Say goodbye to social media stress and hello to effortless, on-brand tweets that will boost your online presence and keep your fans coming back for more. With Cam101's Premium Tweet Generator, you can generate up to 10 tweets day, so you'll have more time to focus on creating killer content and connecting with your audience.

View attachment 96181

You know you can turn the NSFW filter off in the api right (at least on the text-davinci-003 model)? (or did you need to turn it on... I dont remember, I know it does NSFW though because I have had it write a few video scripts and video descriptions for me) and given that we are all in the adult industry it seems baffling to me that you have it on.

Other than that, its basically just a proof of concept, it has limited use without having a custom model trained on a persons tweet history so that it can tweet with the same personality and mannerisms as the person its pretending to be. Without custom training the tweets you get will be bland and generic.
Like sure you can say something like:

(that looks too long to be a tweet to me, so I think you need to do some refinement on how the prompts get wrapped)
or even something like:

But the issue is when you try and add some personality into these tweets you end up with something thats almost where you want it but never quiet there 100% which is why custom models are a thing.

Given that you dont have it set up as an auto tweet though maybe you are expecting the people using it to re-write what the ai generates?

ETA: Not meant to be a post shitting on it but more just, it's a lot more limited in its use than you may have thought, until it's trained on custom data sets.

 

[Antichristrix]

Tried it out already and I liked it! Will be using in the future, since I'm terrible at coming up with tweets myself

 

[Official_Lovense]

Thank you for the feedback

 

[MarieElise]

View attachment 96185
You know you can turn the NSFW filter off in the api right (at least on the text-davinci-003 model)? (or did you need to turn it on... I dont remember, I know it does NSFW though because I have had it write a few video scripts and video descriptions for me) and given that we are all in the adult industry it seems baffling to me that you have it on.

Other than that, its basically just a proof of concept, it has limited use without having a custom model trained on a persons tweet history so that it can tweet with the same personality and mannerisms as the person its pretending to be. Without custom training the tweets you get will be bland and generic.
Like sure you can say something like:
View attachment 96186
(that looks too long to be a tweet to me, so I think you need to do some refinement on how the prompts get wrapped)
or even something like:
View attachment 96187
But the issue is when you try and add some personality into these tweets you end up with something thats almost where you want it but never quiet there 100% which is why custom models are a thing.

Given that you dont have it set up as an auto tweet though maybe you are expecting the people using it to re-write what the ai generates?

ETA: Not meant to be a post shitting on it but more just, it's a lot more limited in its use than you may have thought, until it's trained on custom data sets.

Lol I love your ideas there Marti hahaha.

"Gonna light up a blunt and vibe to these sick treats"
This is a good idea for a service though, to save time. Honestly it looks kind of fun, just for the entertainment of seeing what it would come up w next.

 

[KingMarti]

Lol I love your ideas there Marti hahaha.

"Gonna light up a blunt and vibe to these sick treats"
This is a good idea for a service though, to save time. Honestly it looks kind of fun, just for the entertainment of seeing what it would come up w next.

you'll be able to get better results just using the gpt playground. You do have a limited credit amount to use the service before you have to start paying for requests but it's a lot of requests you get for the $18 credit they give you.

You'll get better results because you have more control over the prompt (if you just prompt it with "give me a good morning tweet" it will still be a super generic tweet with no personality, but it will do NSFW content so its more useable in that respect). Still not as good as building a custom model but thats a lot more complicated and you need a lot of processing power to be able to get through it (basically need a crypto mining farm worth of processing power). But gpt playground gives you the option of using a couple of different models they have (they all cost differently so some will burn through the credit a little faster than others) the most balanced one I have found was text-davinci-003

https://platform.openai.com/playground

Its defiantly useful but the lovense implementation is more restrictive than the gpt api it's using so may as well just use the api imo (playground is like the test site for testing prompts to use in the api). You also have the option of setting it up with your twitter api keys (if you write the script, I might put a script for it on my github) so once you have it giving you the style of tweet that you want you can just set it on a task scheduler so that every day at say 10am it will generate a good morning tweet and send it without you actually having to go and copy and paste it.

It actually has a lot of information on different fetishes too so if you need to learn how to do a new kind of fetish it can give you information on it, help generate new video ideas and even video scripts and descriptions, but for the twitter thing you can tell it for example "you are a femdom that specializes in small penis humiliations content. using that personality write a tweet that will generate engagement", which is something you cant do on the lovense implementation because you will hit their nsfw filter.

 

[Slapstick]

ChatGPT should have been trained on Twitter data is it was trained before Elon fucked up API access. At least it does a reasonable job imitating my boss. Does this sound like something you would Tweet?

Get ready for some serious action tonight! Tonights "Balls to the Walls" cam show is going to be the craziest one yet! Cum join me and let's have some fun! #NSFW #BallsToTheWalls #CamShow #AdultEntertainment #LetsGetWild

I used the prompt: "Generate a tweet about tonights "Balls to the Walls" cam show in the style of @KingMartiCam".

 

[Slapstick]

you'll be able to get better results just using the gpt playground. You do have a limited credit amount to use the service before you have to start paying for requests but it's a lot of requests you get for the $18 credit they give you.

Interesting. I have been using https://chat.openai.com which is their ChatGPT3.5 with unlimited free usage minus all the bells and whistles available in the playground. No idea how long it will remain free and sometime the service is unavailable due to demand. Rather than paying $18 for the playground I would (and probably will) subscribe to the premium ChatGPT ($20/month) that gives access to the latest ChatGPT4. A co-worker says it is noticeably better.

 

[KingMarti]

ChatGPT should have been trained on Twitter data is it was trained before Elon fucked up API access. At least it does a reasonable job imitating by boss. Does this sound like something you would Tweet?


I used the prompt: "Generate a tweet about tonights "Balls to the Walls" cam show in the style of @KingMartiCam".

GPT dosnt know who I am, you can ask it who is x on twitter, it only seems to have data on well known celebrates and people with a high follow count. I wouldnt say its a bad tweet but it's not in the style of @kingmarticam, for that you would need to train a custom model.

Interesting. I have been using https://chat.openai.com which is their ChatGPT3.5 with unlimited free usage minus all the bells and whistles available in the playground. No idea how long it will remain free and sometime the service is unavailable due to demand. Rather than paying $18 for the playground I would (and probably will) subscribe to the premium ChatGPT ($20/month) that gives access to the latest ChatGPT4. A co-worker says it is noticeably better.

Yeah, chatgpt and the gpt api are different. chat-gpt is a model that is built ontop of the gpt-api, with additional filters, and convocation threading added.

 

[Slapstick]

GPT dosnt know who I am, you can ask it who is x on twitter, it only seems to have data on well known celebrates and people with a high follow count. I wouldnt say its a bad tweet but it's not in the style of @kingmarticam, for that you would need to train a custom model.


Yeah, chatgpt and the gpt api are different. chat-gpt is a model that is built ontop of the gpt-api, with additional filters, and convocation threading added.

Yeah, I think they must have cut waaay back on what it will say about people ever since it falsely accused a law professor of sexually harassing a student. Now if you ask about Jonathan Turley ChatGPT errors rather than just making shit up.

US law professor claims ChatGPT falsely accused him of sexual assault, says 'cited article was never written'

Professor Turley revealed that a lawyer had reportedly asked the AI chatbot to generate a list of legal scholars who had committed sexual harassment as part of a study

www.businesstoday.in