I'll add Minimum Description Length to the mix. Under certain definitions and conditions, it equals the Bayesian Information Criterion plus an extra term, which I consider a very interesting result in this "two faces of the same coin" perspective.
Tangentially related: I don't think OCR is the right term and I am generally vocal about that. But seeing this unquestioned here, I am wondering if I am the one who is wrong here. Is it ok to call this OCR? To me ocr means text in the end, not visual tokens.
OCR means optical character recognition. The terms do not require a direct transcription, but that is mostly what OCR meant in the past. If you’re using an LLM’s vision capability to pass in text and the LLM actually understands it, then I would say that it recognized the characters, hence OCR seems okay to use.
But characters only exist when we ask the model about this and it does it best to do this projection if asked. Vision model are richer than that. It "understands" visually a document. If it was only about characters, then there will be no way it beats the traditional pipelines of image->text->extractions or obtain the kind of results we see in this article. Vision models are more than characters recognition and OCR term don't do it justice IMHO.
I used to be critical of almost everyone who used it. Now that I've been learning the ecosystem for a couple years, I'm not as angry about it, and it doesn't affect me as much personally.
I still think almost all of my complaints were valid:
- not just container orchestration, more like a whole AWS solution (its own routing)
- has single-points of failure that would be avoided if you just used aws
- unnecessary complexity for almost all use cases
- ecosystem easily allows you to 2x-8x that complexity (helm+argocd+istio+carpenter+dozens more=hundreds of new failure modes)
- fundamentally moves many teams to a "I don't understand why my service is crashing, so let's just bring up more nodes every time rather than ever learning to debug it" mentality.
- All the feature teams that are supposed to "own their own kubernetes implementation for their own services" never do.
Of course a lot of that frustration is more at working at a place with a poorly-operated SOA where it takes a half-dozen services to send an email to a client or send a text or something silly. It sure is a waste of runway to obsess over a SOA when you aren't profitable yet.
But at this point it's sort of a sunk cost because it's become the industry standard. And AI can help with 90% of the complexity, which is it's own yellow-flag, but here we are.
I’m one of those top commenters who used to be a K8 naysayer. I’m also definitely not young blood.
The reason my position has changed is because:
1. The tooling has gotten better for setting up and managing K8.
2. In two of the last 3 jobs where we opted for a simplified alternative to k8, we came to regret that decision within a couple of years of that decision being made. If you’re core architecture is changing on a timescale of months (not years) then you picked the wrong foundations to build from.
That all said, I still think there is a pragmatic decision that needs to be made. And if I were in the author of this articles position I probably wouldn’t have picked k8s for this task either, despite what I said above. But, and as I said in my comment dismissing this article, they are dealing with low traffic and none of the problems that lend themselves to the benefits of k8. So my criticism of this article is that it’s misleading because their problem is easy but they’re writing as if they’re having to deal
With problems of scale when they’re actually not.
Happy user of recall here. I
rarely need it as I try to keep conversations small and files-focused. But when I do need it, it brings a lot of value. Sometimes there are conversations where I failed to capture some interesting things. Recall is also very helpful to me to audit my system like when I start to suspect some inefficiencies around some tools (skills, mcps, clis). Recall was efficient to retrieve "tranversal context" required for such audit.
So HR and middle management, legal is ... measuring?
They want to focus on builders and sellers but will support them like robots. A great recipe for disaster.
Legal is measuring too? I can't wrap my head around the reasoning process here. Unless cloudflare is going to be a teal enterprise (from the reinventing organisations book), I don't see how it makes sense.
Your comment made me ask myself: "Then why remove it? If it really is just a system prompt, I can't imagine tech debt or maintenance are among the reasons."
My best guess is this is product strategy. A markdown file doesn't require maintenance, but a feature's surface area does. Every exposed mode is another thing to document, support, A/B test, and explain to new users who stumble across it. I'm guessing that someone decided "Study Mode isn't hitting retention metrics", and decided to kill it. As an autodidact, I loved the feature, but as a software engineer I can respect the decision.
What I'm wondering about is whether there's a security angle to this as well. Assuming exposed system prompts are a jailbreak surface, if users can infer the prompt structure, would it make certain prompt injection attacks easier? I'm not well-versed in ML security, and I'd be curious to hear from someone who is.
Honestly, it probably led to long conversations. The tokens/GPU time for one long conversation is more expensive than multiple short conversations. They’re trying to shore up their finances, and they’re moving away from the consumer market and towards enterprise, and students were probably a bad demographic to sell to.
Likely, students are a very expensive demographic to cater for in the age of ai with minimal return for the companies because their use cases are often even more compute intensive than what the average office professionals do at work. So not much to win there if the companies must decide between an enterprise customer with big pockets or some school who will insist to get a free ride because its ed.
People make a hobby out of tricking chat apps to leak their system prompt. But I doubt there’s much gain to be had by using this one vs coming up with a custom prompt.
There used to be a “Custom GPT” feature which basically just creates a prompt wrapper with some extra functionality like being able to call web APIs for more data. Can’t seem to find that menu right now, but it would have easily replicated the study feature. Maybe it was limited to paid accounts only.
Yeah custom gpts are only for paid users. However u can create a new project under "Projects", name it, then when u create it, you can see on the top right the three dots button, click it, open project settings, and there u can place your system prompt under instructions. Every chat you start in that project would send those instructions as a system prompt to the model you are chatting with. so essentially "Study Mode" could be recreated with this approach, or at least it should.
Nor do I know what an "eval" is, or which of the no less than three different deacronymings of "PM" (that I know of, thus far) FB uses or what that role would mean to them.
