This is like a single prompt AI. The term labels are illegible against the white background. The connections don't make any sense. "Training vs Inference" connects to Large Language Models and GPU. It doesn't connect to RLHF. Machine Learning connects to fewer things than Large Language Models. The whole thing is insane.
Labels were a fair hit — that fix just deployed, should be legible now. On connections: Machine Learning actually has 11 edges vs LLM's 4, it's one of the most connected nodes in the graph. But you're right that the edges are hand-authored per concept, so they're locally sensible and globally uneven. Training vs Inference → RLHF is a fair call.
This was clearly - at least designed with AI. Which is not bad by default. But a common thing I’ve come across with designing with AI is it not understanding that people can’t see light text with light background - which is shown here.
Clearly you spent effort in the knowledge and the amazing web view graph which feels great to use on iPhone. But an equal effort needs to be made on all the small corners off the rest of it
This is like a single prompt AI. The term labels are illegible against the white background. The connections don't make any sense. "Training vs Inference" connects to Large Language Models and GPU. It doesn't connect to RLHF. Machine Learning connects to fewer things than Large Language Models. The whole thing is insane.
Labels were a fair hit — that fix just deployed, should be legible now. On connections: Machine Learning actually has 11 edges vs LLM's 4, it's one of the most connected nodes in the graph. But you're right that the edges are hand-authored per concept, so they're locally sensible and globally uneven. Training vs Inference → RLHF is a fair call.
LLM slop, shame on you
I made something similar for my blog: https://thegustafson.com/map
But better, imo, because of the content behind the visual: https://thegustafson.com/series
wonderfully crafted.
This was clearly - at least designed with AI. Which is not bad by default. But a common thing I’ve come across with designing with AI is it not understanding that people can’t see light text with light background - which is shown here.
Clearly you spent effort in the knowledge and the amazing web view graph which feels great to use on iPhone. But an equal effort needs to be made on all the small corners off the rest of it
How is "Chain-of-Thought" explainability in the strictest sense of the term in regards to interpreting the neural networks decision making process?
anyone else also think better with visual nodes and edges?