You've got to be careful with PDFs : We can't see how they are rendered internally for the LLM, so there may be differences in how it's treating the margin/gutters/bleeds that we should account for (and cannot).
"Make this better in a loop" is less powerful than using evolution on a population. While it may seem like evolution is just single steps in a loop, something qualitatively different occurs due to the population dynamics - since you get the opportunity for multiple restarts / interpolation (according to an LLM) between examples / and 'novelty' not being instantly rejected.
I think the "Aha" is that the RL caused it to use an anthropomorphic tone.
One difference from the initial step is that the second time around includes the initial step and the aha comment in the context : It is, after all, just doing LLM token-wise prediction.
OTOH, the RL process means that it has potentially learned the impact of statements that it makes on the success of future generation. This self-direction makes it go somewhat beyond vanilla-LLM pattern mimicry IMHO.
Google Colab gives you $free GPU (usually a 16Gb T4) preloaded with frameworks, ready to run. Later, you might be tempted by the Pro(+) version, but there's plenty of scope to move up the learning curve before spending any money.
If you go to the associated code, you'll see that it needs a 'backbone', 'neck' etc. What is a backbone? Questions that arise directly from the code will lead you towards good blog articles, etc. https://huggingface.co/spaces/nateraw/yolov6/blob/main/yolov...
OTOH, you could go and have a look at (for instance) the Stanford vision courses for a more 'theoretical' approach. But the code itself is often solid guide to what's going on (the frameworks used for Deep Learning map well onto what's being discussed in blogs/lectures/papers).
So, to clarify, does this mean that companies cannot use these models in the course of business, or is it more about selling the translation results directly?
Could be... Except their page (should you choose to believe it, of course) specifically addresses the advantages:
"""
"Advantages over Traditional GANs" : Thus, we observe that our model exhibits _better training stability_ and mode coverage.
"Why is Sampling from Denoising Diffusion Models so Slow?" : After training, we generate novel instances by sampling from noise and iteratively denoising it _in a few steps_ using our denoising diffusion GAN generator.
The search term you're looking for is "Keyword Spotting" (or "Wake Word Detection") - and that's what's implemented locally for ~embedded devices that sit and wait for something relevant to come along so that they know when to start sending data up to the mothership (or even turn on additional higher-power cores locally).
FWIW, for the International audience, the (excellent) UK series "Spooks" was renamed "MI-5" - it was all about the domestic intelligence services in the UK, and (IMHO) worth looking for on NetFlix...
"I wanted to use splines for the baseline survival" - isn't this a modelling step that could have been revisited? It seems a somewhat arbitrary choice (there are other ways to ~interpolate that are much more framework friendly) - and it seems that it forced you down a bit of a rabbit-hole.