AI General Thread

[the-pi-guy]

Mind sharing your best vs worst results so far?

Probably not. It's mostly been NSFW stuff.   


I can mess around with it more.

[Legend]

Probably not. It's mostly been NSFW stuff. 


I can mess around with it more.

Have I got the thread for you! 


In other news, I am done with Claude. It is just such a lazy and self righteous llm. It's crazy how bad it has gotten.

Claude: "User, you told me to do A but that isn't correct so I did B" (actually did C which was explicitly forbidden by the user)

[the-pi-guy]

Mind sharing your best vs worst results so far?

13 minutes for 7 seconds isn't bad if most generations are useful. Could let it render overnight

I sent two to your discord.

[the-pi-guy]

https://www.reddit.com/r/StableDiffusion/comments/1viwvuq/better_avoid_saul_2_minimax_h3/

[the-pi-guy]

https://www.reddit.com/r/StableDiffusion/comments/1vjavt5/updated_methods_on_getting_long_videos_in_minimax/

I feel like this is the first time I've seen in the community a real push for long videos. 

[Legend]

https://www.reddit.com/r/StableDiffusion/comments/1vjavt5/updated_methods_on_getting_long_videos_in_minimax/

I feel like this is the first time I've seen in the community a real push for long videos.

What are we talking, over 10 minutes yet?

[the-pi-guy]

LTX-2.5 just dropped.  

Lol, and they have a comparison table to Minimax.  

https://ltx.io/


Most of their Minimax complaints kind of aren't particularly relevant. 

The biggest win is the speed from what it looks like. 

[Legend]

LTX-2.5 just dropped. 

Lol, and they have a comparison table to Minimax. 

https://ltx.io/


Most of their Minimax complaints kind of aren't particularly relevant.

The biggest win is the speed from what it looks like.

16 gb vram 

[the-pi-guy]

16 gb vram 

Im running minimax on 11 GB VRAM despite that table saying I need 115GB. 

[the-pi-guy]

Sounds like Minimax is still in a class of it's own. 

[the-pi-guy]

https://www.reddit.com/r/StableDiffusion/comments/1vr7sr3/followup_to_my_6minute_tng_video_i_changed_the/

[Legend]

https://www.reddit.com/r/StableDiffusion/comments/1vr7sr3/followup_to_my_6minute_tng_video_i_changed_the/

I made it one minute in. Cool for a tech demo but dry!!!!!!

[the-pi-guy]

There is some bizarre difference between the LLM/it's settings that Google uses in the search window vs the Gemini chat.

Like the search window completely reversed my question from what I was asking. 

I was just googling "does amazon price match Barnes and Noble"

And the AI overview gives me a "No. Barnes and Noble does not match Amazon."

Bro,  that's the opposite of what I asked. I've noticed a lot of very similar errors like that, that I haven't really seen with any of the other LLMs. 

[Legend]

There is some bizarre difference between the LLM/it's settings that Google uses in the search window vs the Gemini chat.

Like the search window completely reversed my question from what I was asking. 

I was just googling "does amazon price match Barnes and Noble"

And the AI overview gives me a "No. Barnes and Noble does not match Amazon."

Bro,  that's the opposite of what I asked. I've noticed a lot of very similar errors like that, that I haven't really seen with any of the other LLMs. 

It's bad. I assume they use their worst model for google search.

Also idk if I mentioned it but gemini app silently discards the past convo once you hit 15 messages. Sucks so much, very stupid.

[darkknightkryta]

To be honest, I think Google is the only one that survives.  They've realized larger neural nets become way worse and the hardware costs are astronomical.  Like, I've noticed Google slashed down Gemini's neural net significantly.  Just relying on having it search for info, and then doing a quick and dirty inference on the data to see what's right.  Sounds bad, but when you look at the server costs and how the larger the net gets, the more mistakes it makes, they have the right idea.  "Good enough".