I read the white paper on this that I stumbled across. It showed the different systems and I guess the ethical principles of each model. You did a great job explaining it. It was creepy seeing the sneaky logic some of them tried!
The teams could use SIGMA 6 to really map out all the tasks and labeling the data sets with keywords that help with defining what the parameters are. I think that will be a tedious process but will make a huge impact when implementing the model with the data and guidance needed to bring huge value to it.
Is it true what Sam is saying about o1's success being made pissible by extending its compute time? It seems - like you were explaining- the chain of thought and the internal qualitative judge are important differences compared to past llms
Even though o3 isn’t yet released to the public, we can assume that they already cheapened the next unreleased model and improved its benchmarks. Can’t even imagine what’s next. The leap between o1 and o3 are so impressive that “o4” or whatever it will be called will be insane. Pls update us thank you!
@generichuman_ here's my two cents, its limited availability, high cost, and focus on specific tasks suggest that it is still primarily a proof of concept with significant challenges to overcome before it can be widely adopted and integrated into real-world applications
I read the white paper on this that I stumbled across. It showed the different systems and I guess the ethical principles of each model. You did a great job explaining it. It was creepy seeing the sneaky logic some of them tried!
The teams could use SIGMA 6 to really map out all the tasks and labeling the data sets with keywords that help with defining what the parameters are. I think that will be a tedious process but will make a huge impact when implementing the model with the data and guidance needed to bring huge value to it.
Is it true what Sam is saying about o1's success being made pissible by extending its compute time? It seems - like you were explaining- the chain of thought and the internal qualitative judge are important differences compared to past llms
Even though o3 isn’t yet released to the public, we can assume that they already cheapened the next unreleased model and improved its benchmarks. Can’t even imagine what’s next. The leap between o1 and o3 are so impressive that “o4” or whatever it will be called will be insane. Pls update us thank you!
"they did no release o3" - lucky us 😅🎉🎇🍻
Real
Yes but the reality is that model is too expensive too run period.
it literally just came out...
2025 - "We" are too expensive to run 😱
O3 seems more like a proof of concept than anything else
how so...
@generichuman_ here's my two cents, its limited availability, high cost, and focus on specific tasks suggest that it is still primarily a proof of concept with significant challenges to overcome before it can be widely adopted and integrated into real-world applications