Hugging Face's logo

Join the conversation

Join the community of Machine Learners and AI enthusiasts.

Sign Up
Back to feed
Quazim0t0 
posted an update Sep 10
Post
188
It feels like a sugar rush… this golden ratio they found. Stepping through a maze of processes, landing on what aims to be a generalist, raised from the cemetery of echoes of our thoughts, statements, findings, and conclusions. I never really thought beyond this point. I always assumed that the world would look as elegant as the innovations we create. I thought it would look different, have a defining design that really embodied the point we have now arrived at. I can't tell the difference. Everything looks the same. I think that's why it's even harder sometimes for us to swallow what is actually happening, because every representation that we've had from science fiction doesn't really fit the theme.

I don't really see any worry with text and language. These models can generalize them well, and we have all seen how profound it can be. What I do worry about more is when we leave the tiny world of language and replace it with the big world of experiences. What would it be like to generalize hundreds of lifetimes? Would it see the nuances better than I? Could it predict what comes next if it had enough data from observing and experiencing the world?...

Read more: https://huggingface.co/blog/Quazim0t0/little-world-big-world
juiceb0xc0de

It's always a pleasure reading your work. It always reminds me to set aside some time for writing descriptive prose. This space needs more abstract thought and philosophical connections. We work with models that have nuanced personalities defining them amongst the market in unique ways. It's like we've all been measuring absorption rating for so long we forgot to step back and look at the painting that the math was always intended to create in the first place.

·
Quazim0t0

Thank you. It's always nice to know someone is reading. I haven't really taken the time to write “manually” in some time, lol. With all the side notes and projects, it's always nice to ask an agent to summarize my notes, but I feel like people appreciate something written.

It's funny you say that. I know this differs a bit, but I was just thinking about this the other day. I had a realization in a conversation with a friend. We got on the topic of Star Wars, and for a slight moment they expressed how much they loved C-3PO and R2-D2. Then we got on the topic of AI later, and there was such a negative pushback from them on the subject. On one hand, I saw their fascination and love for characters that embodied, in a lot of ways, exactly what they hated.

It made me think when you said, “look at the painting the math was intended to create in the first place.” In a nutshell, that seems like how disconnected the painting can be sometimes, even in the simplest forms.

I just think it's good to zoom out when we tinker with these models, because we're so zoomed in while we work. At least for myself, that's how I feel. I'm so concentrated on what levers to pull and which ones I shouldn't.

What started this idea to begin writing as well was something that happened about a year ago. I was making a joke to my cousin, who had just bought a Honda from Japan. He was telling me about the mods and the scene, how everyone is invested in making changes to really make the car theirs. My joke was something like, “You guys are going to be the ones making robots like Chappie.”

Fast forward to now: people are 3D-printing small robots and hooking them up with their own small models. I don't think it's so far off from us having DIY guides for humanoid robots that are fully 3D-printed at home. Maybe we all should start hoarding servo motors. They might become like GPUs one day, lol.

I think it’s fun to sometimes get lost in training, be really immersed in a project, and be proud of the work. But we also can’t plant too many flags of victory when the board is still changing. If we can all somewhat try to predict where this ends up, it seems like finding the proper experience-based datasets will be our next journey.

I look at it as another version of a large movie file being compressed into an MP4. I think quantization will keep getting better, and training methods will improve alongside it. Somewhere in the middle, they will meet for the purpose of fitting into a portable version that becomes the mind of our robots.

Just like AI 2027 (https://ai-2027.com/), I think it’s fun to plot out predictions and see how many darts land later on.
In this post