ComputerPerson

Great comment. Wish it was the top one so I didn't have to read through the others to get here.

Do you have thoughts as to how the courts would debate the deletion that you could present on a similar intellectual plane?

I agree, but this is getting lost in the weeds.

The simpler issue is that "mimic" doesn't apply to LLMs in the domain of biology. That would be like saying a digital recording of a bird's flight mimics the bird.

Whether or not the plumage on the bird is red (in both life and the video) is an entirely different question.

Hard disagree on number 4. Open-ended questions that facilitate conversation are ideal. (They are an art form in themselves.) Since we're being analytical, you need a scaffold for the conversation, and questions are a better way to build it.

Maybe I’m misinterpreting "early on." I guess their point holds for the first 500 words or so.

ChatGPT is, indeed, part of the revolution. Analogously to the early 1900s, I see this more as another machine that changed the factory landscape. For example, fine fabrics were mass-produced and turned into dresses.

From there, fine dresses became something affordable to people who never could have dreamed of them. ChatGPT is simply one of these machines, producing "deep research" as a fine dress; just an example.

Chatbot subscriptions will never be ubiquitous household and industry mainstays. I'm asking bigger.

Or maybe they will! I know that if I had something to yap to, that responded quickly, had perfect (better than Whisper) understanding, and remembered everything perfectly, I'd journal with it. This isn't a reality yet. Models are too slow and expensive for this to work with tools like OpenClaw, Hermes, etc.

It's very common. PaddleOCR is enough to get extremely well-done bounding boxes, and it runs very fast on a $150.00 GPU.

There's always room for improvement, though. I suspect a tool will emerge for highly detailed OCR that implements a nested bounding-box-based multi-scale approach, effectively OCRing small sections at a time and then gradually compiling them by expanding the surface area using the bounding boxes.

I've thought a lot about implementing it anyway.

edit: I see you're asking about the block labels. Leaving the comment in case someone finds it interesting.

Are you familiar with the Thesaurus Linguae Graecae? It's a project that may meet your needs. Paywalled, unfortunately, but I assume it would be a one-time expense.

You're the second person to mention handwriting. I think it might be handwriting-specific; perhaps Anthropic has a better corpus for this.

I really wouldn't know, though. Anthropic models barf out copyright issues for my use case, so I'm unable even to benchmark them. It's a common problem when you're scanning public domain books. Mine are reference texts often cited.

My use case isn't in the realm of old-school OCR, so it's not a good comparison, but anyway:

As another user pointed out, it's surprisingly random (task-specific). Llama Scout outperformed Gemini Flash 2.5 on a benchmark I built at the time. I didn't include an OCR models.

Mistral might indeed be the best OCR-specific model for my task, now that you ask. Funny. It's so bad at my work that I didn't register it might be the best in its category. This is just based on vibes from my single scan.

I've got a scan from a book that I OCR with new releases. Ligatures, critical sigla, Fraktur letterforms, subscripts, superscripts, etc.

Nothing special about this model for overly-detailed work like mine.

It's been a while since I last tested (and discontinued my subscription), but the "pro" models from OpenAI dominate. Not surprising, given the price difference, but it would be nice if an OCR-specific model could perform better. It's worth mentioning that even the highest-end models do a pretty poor job with intricate text like mine.

Which is not great for history, right? Seems like a real detractor.

You pretty much need $30/m subscriptions for historical data. Not that it couldn't be worked into this.

I use the CLI, too, but I enjoy watchinng people use the desktop apps. It seems like the desktop apps may be a little bit better suited for human-in-the-loop work. If you're already switching back and forth between a mouse, having an IDE running may make more sense.

The best feeling has to be in the CLI, in my oppinion. A low-stakes project where you can let ultracode runaway from you in the CLI is the most fun.