forced in the sense that we are "forced" to use social media today, probably
April 17, 2025 at 3:01 AM
forced in the sense that we are "forced" to use social media today, probably
the question of ethics only arose once the ouputs started to look like the inputs
March 29, 2025 at 9:13 AM
the question of ethics only arose once the ouputs started to look like the inputs
*enough real data from B
March 28, 2025 at 6:02 AM
*enough real data from B
im not contesting that, im simply stating the pinpoint solution for any given model not adequately doing A for B is to add more B to the dataset - and considering, perhaps to your point, real data from B may not exist, it must be generated.
a model outputs what you input
a model outputs what you input
March 28, 2025 at 5:58 AM
im not contesting that, im simply stating the pinpoint solution for any given model not adequately doing A for B is to add more B to the dataset - and considering, perhaps to your point, real data from B may not exist, it must be generated.
a model outputs what you input
a model outputs what you input
everyone takes consciously or not, its an issue to me when it becomes nonreciprocal
March 28, 2025 at 5:27 AM
everyone takes consciously or not, its an issue to me when it becomes nonreciprocal
something like that
but a more detrimental (to those like OAI) and perhaps quicker remedy would be requiring them to open source. theyd likely just bow out
but a more detrimental (to those like OAI) and perhaps quicker remedy would be requiring them to open source. theyd likely just bow out
March 28, 2025 at 5:22 AM
something like that
but a more detrimental (to those like OAI) and perhaps quicker remedy would be requiring them to open source. theyd likely just bow out
but a more detrimental (to those like OAI) and perhaps quicker remedy would be requiring them to open source. theyd likely just bow out
surely in this specific instance one could round out the set with enough synthetic data, no?
March 28, 2025 at 5:14 AM
surely in this specific instance one could round out the set with enough synthetic data, no?
"best" being defined as "what the most amount of people want" - the intrigue stems from being able to prompt what used to take building some ungodly workflow
thats besides the point that none of it will matter anyway until its open source, or they at least publish a paper
thats besides the point that none of it will matter anyway until its open source, or they at least publish a paper
March 28, 2025 at 5:11 AM
"best" being defined as "what the most amount of people want" - the intrigue stems from being able to prompt what used to take building some ungodly workflow
thats besides the point that none of it will matter anyway until its open source, or they at least publish a paper
thats besides the point that none of it will matter anyway until its open source, or they at least publish a paper
generative models do not output on their own, at least not yet
i cant think of an example of an image one might see that doesnt, at some point in some way, involve a human
i cant think of an example of an image one might see that doesnt, at some point in some way, involve a human
March 28, 2025 at 4:42 AM
generative models do not output on their own, at least not yet
i cant think of an example of an image one might see that doesnt, at some point in some way, involve a human
i cant think of an example of an image one might see that doesnt, at some point in some way, involve a human
i came to the realization, perhaps just now, that this only works as a result of natural discovery. intentional subterfuge only hurts.
March 28, 2025 at 4:40 AM
i came to the realization, perhaps just now, that this only works as a result of natural discovery. intentional subterfuge only hurts.
its important to understand the difference between closed and open source tech. open (local) software (models) do NOT refuse requests. it is not "the AI saying yes or no" it is the company providing the service that dictates what can be generated.
local models are paintbrushes.
local models are paintbrushes.
March 28, 2025 at 4:29 AM
its important to understand the difference between closed and open source tech. open (local) software (models) do NOT refuse requests. it is not "the AI saying yes or no" it is the company providing the service that dictates what can be generated.
local models are paintbrushes.
local models are paintbrushes.
what i find interesting is the fact that, for awhile relatively speaking, the most widely used image model was in fact originally created to cater to bronies (and furries) by a brony. it just so happened to be able to output non-fur
March 28, 2025 at 4:17 AM
what i find interesting is the fact that, for awhile relatively speaking, the most widely used image model was in fact originally created to cater to bronies (and furries) by a brony. it just so happened to be able to output non-fur
Using a T-pose is smart. That's not bad at all.
January 26, 2025 at 8:26 AM
Using a T-pose is smart. That's not bad at all.
im sad b-loras didnt seem to take off. i thought they were very impressive
December 1, 2024 at 8:39 AM
im sad b-loras didnt seem to take off. i thought they were very impressive
taking openai / altman seriously is something i haven't seen since it stopped being a non-profit
none feel downstream effects from the "abuse". no one cares. this is besides the fact that their current models are irrelevant
none feel downstream effects from the "abuse". no one cares. this is besides the fact that their current models are irrelevant
November 29, 2024 at 4:54 AM
taking openai / altman seriously is something i haven't seen since it stopped being a non-profit
none feel downstream effects from the "abuse". no one cares. this is besides the fact that their current models are irrelevant
none feel downstream effects from the "abuse". no one cares. this is besides the fact that their current models are irrelevant
aversion to inclusion seems predicated on "it sucks"
November 28, 2024 at 7:19 AM
aversion to inclusion seems predicated on "it sucks"