Search Results
7/24/2025, 11:17:20 PM
7/24/2025, 7:15:07 AM
>>511199704
yes, t2v is faster, but it really depends more on the model than anything else
if a model needs 30 steps there's just no way around it
that's why models that give you good quality in 8-10 are paramount
and you can also try Wan with Wan21_CausVid_14B_T2V_lora_rank32, and use 8 steps at 0.6, or 12 at 0.3 and the results are pretty good
yes, t2v is faster, but it really depends more on the model than anything else
if a model needs 30 steps there's just no way around it
that's why models that give you good quality in 8-10 are paramount
and you can also try Wan with Wan21_CausVid_14B_T2V_lora_rank32, and use 8 steps at 0.6, or 12 at 0.3 and the results are pretty good
7/20/2025, 4:10:40 AM
>>510852251
>We're nowhere near that
we are
>You'd still be limited by factors such as power consumption and heat generation.
obviously
>None of this even answered other anons question of why "AI" should be considered people legally. If anything what your saying is a good example of why the industry WOULDN'T want them to be considered people. It's easier to exploit non-people.
because that's not what I meant by "people"
>>510852413
I mean having the full cognitive capacities of a human being
basically a digital human being
>We're nowhere near that
we are
>You'd still be limited by factors such as power consumption and heat generation.
obviously
>None of this even answered other anons question of why "AI" should be considered people legally. If anything what your saying is a good example of why the industry WOULDN'T want them to be considered people. It's easier to exploit non-people.
because that's not what I meant by "people"
>>510852413
I mean having the full cognitive capacities of a human being
basically a digital human being
7/17/2025, 5:38:03 AM
>>510598041
I'm certain cloud AI services use some kind of private hidden context window assigned to the user, but haven't revealed so because it could be seen as unlawful data-collection
>>510598426
I don't remember the metric, anon, sorry
but that's why large context windows make the interaction a lot slower, because the VRAM prioritizes the model, and the context is offloaded to RAM/CPU
I'm certain cloud AI services use some kind of private hidden context window assigned to the user, but haven't revealed so because it could be seen as unlawful data-collection
>>510598426
I don't remember the metric, anon, sorry
but that's why large context windows make the interaction a lot slower, because the VRAM prioritizes the model, and the context is offloaded to RAM/CPU
7/15/2025, 2:35:49 AM
>>510406347
those are technicalities
where there's a will there's a way
we WILL impregnate our robot wives
those are technicalities
where there's a will there's a way
we WILL impregnate our robot wives
7/2/2025, 6:09:43 AM
Page 1