johnQdeveloper

https://ollama.com/

I believe its just a HTTP wrapper and terminal wrapper around llama.cpp with some modifications/fork.

total duration: 35.016288581s load duration: 21.790458ms prompt eval count: 1244 token(s) prompt eval duration: 1.042544115s prompt eval rate: 1193.23 tokens/s eval count: 213 token(s) eval duration: 33.94778571s eval rate: 6.27 tokens/s

total duration: 4m44.951335984s load duration: 20.528603ms prompt eval count: 1502 token(s) prompt eval duration: 773.712908ms prompt eval rate: 1941.29 tokens/s eval count: 1644 token(s) eval duration: 4m44.137923862s eval rate: 5.79 tokens/s

Compared to an API call that finishes in about 20% of the time it feels a bit slow without the recommended graphics card and what not is all I'm saying.

In terms of benchmarks, it seems unusually well tuned for the model size but I suspect its just a case of gaming the measurement by testing against it as part of the development of the model which is not bad in and of itself since I suspect every LLM who is in this space marketed to IT folks does the same thing tbh so its objective enough given that as a rough gauge of "Is this usable?" without heavy time expense testing it.

I suggest you do since tbh you are likely (as others have said) to be violating privacy laws with your current implementation + the discord ToS. If its anonymized better, you are less likely to be a target of someone who gets angry about not knowing you exist.

Up to you, your life your circus y'know?

> I have absolutely zero idea what industry would be interested in this, in what form, and if anyone would even pay.

LLM data collection if its not being bought via discord already directly.

Same reason I'd want to use highly anonymized and curated data from the roleplay / writing discords as training data. It is just I'd have to go through and anonymize your data and curate it / clean it up before I would dare to send it to an LLM for legal reasons.

If I send/share PII, I'd be screwed just like you will be if someone gets upset.

> I really don't care what people do with the data, as long as they are not spamming requests or using the data for commercial purposes without permission.

Fair, for me, this is for hobby implementations of solo roleplaying content similar to AI Dungeon and other implementations so its not commercial but my use case (for your purposes) would be better served by just being able to download a database dump (properly anoynmized or me doing it) for specific servers since most data is useless to me that you collect since I've got a specific goal in mind and want to minimize data collection for legal liability reasons. (i.e. non-commercial roleplaying with no PII or other privacy risky info is likely to be a safe use case)

EDIT:

I'd consider dropping attachments + links and only recording text as well for CSAM and other abusive material reasons. I doubt you have the moderation in place to protect yourself.

Pictures and videos and what not are a lot more dangerous to you than text would be. (i.e. despite what people say about it, realistically, most text in a public forum on the internet w/o PII is not going to get you hit with fines)

That said, personally, I would not publish this as you have because I don't have that kind of risk tolerance but I can see it being "safe enough" for some people. But the images/attachements are in "are you really sure you want to do that? You could go bankrupt" territory.

> I have placed restrictions on searching directly by user ID to prevent doxing. I also made the opt out process one click, for those who do not want to be archived.

1) I'd suggest anonymizing the usernames / author ids to something more privacy friendly such as how some image sites were generating 3-4 random words as a human readable unique id. This removes a lot of the reason people would opt out (i.e. posts being tracked down years later)

2) You not seem to have a clear rate limit documentation. If you are asking people to pay for commercial use, I'd suggest making it clear what the rough original limits are as well as the rough price range of what you'd offer.

3) Tbh, the only real thing I want from this project is basically narrative / roleplay / writing content for LLM reasons as I'm trying to build a rules-oriented system that narrates via LLM. If you don't want people using this data for this purpose, I'd suggest making that clear.

Some people just like a particular workflow or tooling flow and build it really. Maybe it works for enough people to have a viable market, maybe not.

I am just using a PHP deployment process for no reason other than feeling like it for personal projects and it handles 60% of the work without me needing to do anything. But any runbooks for it are tasks built into the tool and in the same git repo for the entire server deployment. I'm not gonna put it in some random place or a shell script that I need to remember separate commands for.

Code, for programmers, is inherently self-documenting if you keep a simple functional style without any complexity with comments on the occasional section that isn't just "Create a MySQL user, roll the MySQL user's password, update the related services with the new password/user combination, remove the old user that the fired employee has credentials to on the off chance we failed to block them at the VPN" kind of stuff.

Credit score is impacted by the average age but tbh, I've gotten two mortgages while doing this and compared the numbers with my peers and it could be my situation is unusual but the "cost" was like .2% APR between those that don't churn in my RL group and those that do.

It has (in my experience with multiple mortgages) have a rate set by competent use of credit that does not lead you to have much debt but paying regularly in practice regardless of what the public credit score algo is telling people.