[3.2][3.3][RC] AI Labs with Midjourney, FaceSwap, PixVerse, Pika, Claude, ChatGPT, Gemini, DALL-E, Stable Diffusion

[nou nou]

Seems very interesting. Got to do allot of reading to see how this all works. Anyone got video demos of them using it?

Videos would be pretty boring. It works, and it works well. You post something, you wait, a response comes.

[nou nou]

Oh and while you wait you can just go do something else on the forum. It's neat, really.

[salvocortesiano]

Greetings,
I understand. I use phpBB hosting server on Linux platform. Could you point me to the user who solved this problem?

Thank you
Salvo

It's John Rambo, just above.

Thanks

[privet.fun]

I wonder if I should add support for recently released Llama 3.1, which is now leading in multiple tests/categories?

Here's API vendor with 450B support I found so far https://llama-api.com, there's probably more.
Price table from https://console.llama-api.com/account/usage, pricing is based on the number of parameters of the model:
<7b = $ 0.0004 / 1K Tokens
between 7b and 34b = $ 0.0016 / 1K Tokens
>34b = $ 0.0032 / 1K Tokens (I assuming this will be covering 450b model)

Which makes is a lot cheaper than Chat-GPT 4o https://openai.com/api/pricing/
gpt-4o
$5.00 / 1M input tokens ($0.5 / 1K Tokens)
$15.00 / 1M output tokens ($1.5 / 1K Tokens)

​

[nou nou]

I wonder if I should add support for recently released Llama 3.1, which is now leading in multiple tests/categories?

I literally just logged on to come ask you if you would consider supporting Llama. Yes please!

Also, open source so this is going to go places, people will optimize/specialise it for all sorts of purposes.

[privet.fun]

I wonder if I should add support for recently released Llama 3.1, which is now leading in multiple tests/categories?

I literally just logged on to come ask you if you would consider supporting Llama. Yes please!

Also, open source so this is going to go places, people will optimize/specialise it for all sorts of purposes.

Good point, should be relatively easy to add. Just need to find good API provider, there appears to be a lot of them but pricing vary wildly.

[nou nou]

Maybe check in with the useapi folks?

[nou nou]

Question for you - I'm trying to get answers that are more technically consistent with phpBB so I tried adding this as a prefix:

Code: Select all

Your answers must be clear, organized and concise. You MUST write any Lua, Python or other programming code inside [code] BBCode tags.

But both ChatGPT and Gemini insist on code samples to be written inside markdown ```

When I add that prefix inside a prompt though, they come back with neatly BBCode tagged code, so I'm not entirely getting it.

Do you have any insights there?

[nou nou]

Iiiiiinteresting... for ChatGPT it seems to be solved by writing

Your answers must be clear, organized and concise. You MUST write any Lua, Python or other programming code inside [code][/code] BBCode tags.

i.e. adding BOTH tags.

Gemini not so...

[nou nou]

I may have asked this before - apologies if I did - but is there a coming-soon kinda plan to support attachments for ChatGPT? So I can upload a text file to it to read for example?

[privet.fun]

Maybe check in with the useapi folks?

I don't think they have that, will have to shop around to see what's available.

[privet.fun]

Question for you - I'm trying to get answers that are more technically consistent with phpBB so I tried adding this as a prefix:

Code: Select all

Your answers must be clear, organized and concise. You MUST write any Lua, Python or other programming code inside [code] BBCode tags.

But both ChatGPT and Gemini insist on code samples to be written inside markdown ```

When I add that prefix inside a prompt though, they come back with neatly BBCode tagged code, so I'm not entirely getting it.

Do you have any insights there?

My attempts to have consistent BBCode tags in responses have so far failed. They work maybe 50% of the time, at best. Hopefully as models progress this will get better.

[privet.fun]

I may have asked this before - apologies if I did - but is there a coming-soon kinda plan to support attachments for ChatGPT? So I can upload a text file to it to read for example?

That was my plan to release full multimodal support for all three (Gemini, ChatGPT and Claude). I was mostly curious to see if I could work with images and audio files and have continuous conversations around them. It turned out to be quite difficult mostly due to some limitations around data management. All three also have slightly different approaches to this problem so I'll probably have to draw a line somewhere and release them one at a time. I have some home improvement projects to complete this summer which take up big chunks of my free time but I do hope to release this before EOM or early in September.

[nou nou]

Nice thank you! Obviously take all the time you need

[nou nou]

PS this extension is a nice addition to working with the bots' response styles:

https://github.com/AlfredoRamos/phpbb-ext-markdown

Testing it on my test server before deploying, I've had some (minor but annoying) conflicts with earlier versions of it.