torchcompiled's profile picture. trying to feel the magic. cofounder at @leonardoai_ directing research at @canva

Ethan

@torchcompiled

trying to feel the magic. cofounder at @leonardoai_ directing research at @canva

مثبتة

personally I feel like the inflection point was early 2022. The sweet spot where clip-guided diffusion was just taking off, forcing unconditional models to be conditional through strange patchwork of CLIP evaluating slices of the canvas at a time. It was like improv, always…

torchcompiled's tweet image. personally I feel like the inflection point was early 2022. The sweet spot where clip-guided diffusion was just taking off, forcing unconditional models to be conditional through strange patchwork of CLIP evaluating slices of the canvas at a time. It was like improv, always…
torchcompiled's tweet image. personally I feel like the inflection point was early 2022. The sweet spot where clip-guided diffusion was just taking off, forcing unconditional models to be conditional through strange patchwork of CLIP evaluating slices of the canvas at a time. It was like improv, always…
torchcompiled's tweet image. personally I feel like the inflection point was early 2022. The sweet spot where clip-guided diffusion was just taking off, forcing unconditional models to be conditional through strange patchwork of CLIP evaluating slices of the canvas at a time. It was like improv, always…
torchcompiled's tweet image. personally I feel like the inflection point was early 2022. The sweet spot where clip-guided diffusion was just taking off, forcing unconditional models to be conditional through strange patchwork of CLIP evaluating slices of the canvas at a time. It was like improv, always…

Image synthesis used to look so good. These are from 2021. I feel like this was an inflection point, and the space has metastasized into something abhorrent today (Grok, etc). Even with no legible representational forms, there was so much possibility in these images.

eprombeats's tweet image. Image synthesis used to look so good. These are from 2021. I feel like this was an inflection point, and the space has metastasized into something abhorrent today (Grok, etc). Even with no legible representational forms, there was so much possibility in these images.
eprombeats's tweet image. Image synthesis used to look so good. These are from 2021. I feel like this was an inflection point, and the space has metastasized into something abhorrent today (Grok, etc). Even with no legible representational forms, there was so much possibility in these images.
eprombeats's tweet image. Image synthesis used to look so good. These are from 2021. I feel like this was an inflection point, and the space has metastasized into something abhorrent today (Grok, etc). Even with no legible representational forms, there was so much possibility in these images.
eprombeats's tweet image. Image synthesis used to look so good. These are from 2021. I feel like this was an inflection point, and the space has metastasized into something abhorrent today (Grok, etc). Even with no legible representational forms, there was so much possibility in these images.


Ethan أعاد

New post! As opposed to building reward models over human ratings and using them for RL, can a model develop its own reward function? Humans seem to develop their own aesthetic preferences through exploration and socializing. How can we mimic this for generative models?

torchcompiled's tweet image. New post! As opposed to building reward models over human ratings and using them for RL, can a model develop its own reward function? 

Humans seem to develop their own aesthetic preferences through exploration and socializing. 

How can we mimic this for generative models?

2 raised solutions here: 1 captures the social aspect while leaving out the difficulty of exploration: That one's taste develops by learning about other's tastes, we could imagine training a generative model trained over a dataset of many reward models, and sample new plausible…

torchcompiled's tweet image. 2 raised solutions here:

1 captures the social aspect while leaving out the difficulty of exploration: That one's taste develops by learning about other's tastes, we could imagine training a generative model trained over a dataset of many reward models, and sample new plausible…
torchcompiled's tweet image. 2 raised solutions here:

1 captures the social aspect while leaving out the difficulty of exploration: That one's taste develops by learning about other's tastes, we could imagine training a generative model trained over a dataset of many reward models, and sample new plausible…

New post! As opposed to building reward models over human ratings and using them for RL, can a model develop its own reward function? Humans seem to develop their own aesthetic preferences through exploration and socializing. How can we mimic this for generative models?

torchcompiled's tweet image. New post! As opposed to building reward models over human ratings and using them for RL, can a model develop its own reward function? 

Humans seem to develop their own aesthetic preferences through exploration and socializing. 

How can we mimic this for generative models?


Ethan أعاد

New post! I believe we can think of ourselves in two different lenses: an exact point of experience and the history of our patterns of behavior. Though the two are deeply interconnected.

torchcompiled's tweet image. New post! I believe we can think of ourselves in two different lenses: an exact point of experience and the history of our patterns of behavior. Though the two are deeply interconnected.

Ethan أعاد

Last one in this series! Can we pinpoint where the self is located in the mind? We know that a number of parts of the brain can be ablated with minimal personality changes while others are more damaging. Not to mention how the body at large plays a role.

torchcompiled's tweet image. Last one in this series! Can we pinpoint where the self is located in the mind? We know that a number of parts of the brain can be ablated with minimal personality changes while others are more damaging. Not to mention how the body at large plays a role.

Ethan أعاد

Yet another post on identity. Similar to the last, our model of the world is shaped by our observations of the world, which in turn precipitates our identities.

torchcompiled's tweet image. Yet another post on identity. Similar to the last, our model of the world is shaped by our observations of the world, which in turn precipitates our identities.

Loading...

Something went wrong.


Something went wrong.