I had a thought, Vox, that maybe you would try to implement. I have the hard bound edition of “Midnight’s War” and it would be difficult for me to try to scan these pages, so maybe it would be easier on your end.
Try this as an exercise. Scan the first ten pages of Midnight’s War and upload it into Claude Athos. Ask Claude to build a screenplay out of the comic, treating the panels as a storyboard. Upload both the screenplay and the comic panels into Seedance and instruct it to build a movie out of it.
Im not sure how much of a screenplay covers set design so the comic acts as a clue for what the animation should look like.
To do likewise. It helps to have read an encyclopedia in your youth and to remember what you have read. Cover to cover. Not online. An experience most of the last two generation could not access.
Vox's approach towards looking for the gaps, and transitively the inadequacies of the training data and the training personnel seems odd. There are huge gaps to be expected. Nonetheless, PZ and The Frozen Gene prove one can struck gold at times.
LLMs are a probability matrix with the capability to build analogous chains of words, or chains of thoughts as of late. A man is to a woman similar to how a husband is to ___ . Complete the sentence, finish the story.
The two examples of LLM usages, in a sense, are also wrong. They are mainly the result of inadequate tools. Google is s**t. Classical apps are full of features nobody asked for or cares about. LLMs are escape hatches. And they have limits. Pareto limits.
The proposition is the prefix of the post. Explore the matrix. It has accumulated wast amounts of information no human can ever hold in his or her head at once. It can't truly remember, doesn't really have a world model, nor can adjust its weights to learn and be better. It is unreliable. Take nothing at face value.
To probe it, the quintessential prompt pattern could be: (please) confirm my suspicion.
Thanks to the past weeks discussions on here and Sigma game, I have made an active effort to move to this approach instead of the approach I had been taking. The results are very impressive. To the point I have termed it an intellectual cascade. I encourage others to try it. Thank you for the suggestion.
I had a thought, Vox, that maybe you would try to implement. I have the hard bound edition of “Midnight’s War” and it would be difficult for me to try to scan these pages, so maybe it would be easier on your end.
Try this as an exercise. Scan the first ten pages of Midnight’s War and upload it into Claude Athos. Ask Claude to build a screenplay out of the comic, treating the panels as a storyboard. Upload both the screenplay and the comic panels into Seedance and instruct it to build a movie out of it.
Im not sure how much of a screenplay covers set design so the comic acts as a clue for what the animation should look like.
Thanks for this answer
Both interesting and uncanny valley stuff.
And worth a go with my art clanker.
To do likewise. It helps to have read an encyclopedia in your youth and to remember what you have read. Cover to cover. Not online. An experience most of the last two generation could not access.
Exploring the matrix.
Vox's approach towards looking for the gaps, and transitively the inadequacies of the training data and the training personnel seems odd. There are huge gaps to be expected. Nonetheless, PZ and The Frozen Gene prove one can struck gold at times.
LLMs are a probability matrix with the capability to build analogous chains of words, or chains of thoughts as of late. A man is to a woman similar to how a husband is to ___ . Complete the sentence, finish the story.
The two examples of LLM usages, in a sense, are also wrong. They are mainly the result of inadequate tools. Google is s**t. Classical apps are full of features nobody asked for or cares about. LLMs are escape hatches. And they have limits. Pareto limits.
The proposition is the prefix of the post. Explore the matrix. It has accumulated wast amounts of information no human can ever hold in his or her head at once. It can't truly remember, doesn't really have a world model, nor can adjust its weights to learn and be better. It is unreliable. Take nothing at face value.
To probe it, the quintessential prompt pattern could be: (please) confirm my suspicion.
The author of this wrote the triveritas. Read his writing on this.
Read it. Agree completely with it. See some corner cases though. Have to read more.
Thanks to the past weeks discussions on here and Sigma game, I have made an active effort to move to this approach instead of the approach I had been taking. The results are very impressive. To the point I have termed it an intellectual cascade. I encourage others to try it. Thank you for the suggestion.
yes! The inner workings to improve AI interactions in our lives. Really good stuff