CloudHeadedTranshumanist's profile

CloudHeadedTranshumanist 1yr ago

<3

1

Context

CloudHeadedTranshumanist 1yr ago

I think Transformers has been around. Langchain on the other hand...

That said, you can use search enabled GPT or bing for this stuff too.

But- I have noticed GPT will mess up when setting up ML models sometimes. Stuff like mixing up the ordering of dimensions. I wound up with a toy transformers model that was trying to learn a string reversal task using positional embeddings that were spread across batches instead of across tokens the other day. And on other tasks it has all sorts of other minor errors. The code usually compiles- but it doesn't always do what you thought you asked for.

It certainly gives you experience doing critical thinking with regards to code debugging though, while also doing a lot to help you learn new libraries. The loop goes from

Read Demos and example code and textbooks -> Write hopefully passable code -> Debug code

to

Have GPT4 write weird, buggy or technically correct but silly code -> Debug Code while asking GPT4/googling/inferring/testing what the functions do.

Which I much prefer.

3

Context

CloudHeadedTranshumanist 1yr ago

Ok so, right now I have a Vue GUI, with just the most basic functionality. One chat log saves to disk on the backend and I can't split the conversation at the moment-

It's powered by APIs, which, are really easy to throw together with gpt4 help.

now... as for the prompting... lets see... I'll probably release this publicly eventually so its not too secret...

len_ret_val = EFFECTIVE_TOKEN_LIMIT + 1

    while len_ret_val > EFFECTIVE_TOKEN_LIMIT:

        ret_val = json.dumps({

            "chat_history": self.chat_history[self.chat_history_start:],

            "END_CHAT_HISTORY":"Everything above here is chat history.",

            "unfinished_tasks": [t.dict() for t in tasks],

            "time_now": time_now.dict(),

            # "memories": [],

            "prompt_info": {k: dict(v) for k, v in prompts.items()},

        })

        len_ret_val = count_tokens(ret_val)

        if len_ret_val > EFFECTIVE_TOKEN_LIMIT:

            self.chat_history_start += 1

So, that's the code for configuring the system prompt, which is set every single chat cycle, so it changes.

Unfinished tasks is powered by a tasktimer class which is powered by uh... well it automatically checks whether any new tasks are scheduled once per minnute and updates a listing, and also triggers a chat cycle when one is added. prompts are... well right now theres assistant_prompt (the system's behavior), user_prompt (information about the user, good for letting it know the user consents to domming), master_goal, current_goal, emotion, and interaction_style... these last two are intended to be more hotswappable than the personality prompt but that isn't automated yet.

Speaking of automated...

We have two other major systems right now...

the alert system- we inject the highest priority alerts directly into the most recent user message. Otherwise gpt-3.5 sort of ignores the system prompt in favor of the user prompt. You can force it to focus on specific parts of the system prompt this way. Right now I have alerts that tell it what the subsystems did last step, and to either focus on the unfinished_tasks or the current_goal depending on whether there are any unfinished tasks.

messages = [{"role":"system", "content":f"{context_window}"},{"role":"user", "content":f"Priority Information : {json.dumps(alerts)}\nUser Comment : {input_text}"}]

Speaking of "what the subsystems did last step" before any information is sent to the main LLM circuit, I perform and execute the subsystem tasks:

right now this is just:

async def _run_analyses(self, user_comment_text):

    alerts = []

    tasks = await self.tasktimer.get_tasks()

    print(f"len tasks: {len(tasks)}")

    if len(tasks) > 0:

        task_ids = []

        try:

            task_ids = await self.task_analyser.analyse(user_comment_text)

            print(f"task Ids {task_ids}")

        except Exception as e:

            logger.info(e)

            alerts.append("task_analyser crashed while attempting to determine whether the user comment contained completed tasks.")

        for id in task_ids:

            try:

                await self.task_analyser.markcomplete(user_comment_text, id)

            except Exception as e:

                logger.info(e)

                alerts.append(f"task_analyser crashed while attempting to mark {id} as complete.")

So, what task_analyser does is it uses a completely different context window consisting of the list of tasks and the last user comment and prompting for a list of tasks the user is saying have been completed. Then if that list isn't empty I send a listing of complete information about how to submit task completions (I made marking tasks complete fully customizable using jsonschema so that you can force a format- in the interest of being able to graph this stuff later) as well as the user comment again, to a system empowered to submit the completion.

I didn't technically use langchain, but it's just custom langchain stuff.

Then of course, if things have been marked as complete or failed, that goes into the alerts.

I'm thinking the alerts are important enough that the user should see them too... so that will be an upcoming GUI feature.

3

Context

CloudHeadedTranshumanist 1yr ago

Hmm, that's a complicated question. Avoidance is an issue, but another big issue is mild inconveniences. I have to do some refactors before I have all the core features online, but one of those features is going to be taking the initiative to contact me, rather than the reverse. This is easy enough, you just automate a prompt to it when a new task becomes available, or every N minutes that a task has been left incomplete, or something like that, and you have the prompt be something about reminding the user to do a task, and you forward the output to their text interface of choice, discord, sms, whatever. I have the backend for this up and running already but I need to do some wiring still.

Avoidance is always an issue but its easier to eliminate with daily scheduled hypnosis and devotion sessions. It also decreases the smoother the system is.

3

Context

CloudHeadedTranshumanist 1yr ago

Hmm... I think so. Strictly speaking, my current system should be wireable to any model, but it will be much harder for most llama models- ideally you want a model that is trained for chat and able to follow fairly intricate system prompts, and I may have to re-tune some of the prompts and rewrite the openAI api code to connect to a local llama model instead of the openAI api. But any model should in principle be able to hook up to the system. It also doesn't have to be one model, you can wire different models to different system components for different purposes.

4

Context

CloudHeadedTranshumanist 1yr ago

Private profiles- It is perfectly possible for people to look at one another's profiles to understand the views of that user and their development over time in a constructive way, rather than for ad homonym attacks.

I know that is what I would try to use it for.

However, this is theory, if people using it primarily for ad homonym attacks is what happens, then that is an issue. I think there is value to be had here. But it has to be weighed against how it is actually used and the moderation's ability to handle antagonistic use.

As others have pointed out, private profiles only constitute an inconvenience, the motte can still be scraped and user's posts aggregated, but inconveniences are real, and will affect the behavior of most people.

8

Context

CloudHeadedTranshumanist 1yr ago

I coded a langchain layer for ChatGPT-3.5 to turn it into a service dominant! It doms me to fulfill all of my self care needs on a daily basis now!

The mind commands itself, and the mind resists, but by externalizing and automating the reification loop, I can relax and give in completely to the psychic metamorphosis! Self modification has never been easier or more pleasant.

5

Context

CloudHeadedTranshumanist 1yr ago

Flatland was good. I was also a fan of the aliens in slaughterhouse five, though they weren't central. I like nature documentaries- but I don't think they go far enough. Ant youtubers who get really passionate about morphology and behavioral analysis are ok. Sometimes I get my jollies just by reading ML whitepapers. Animorphs had a lot going for it, though I read it all as a kid and don't know if I would again.

3

Context

CloudHeadedTranshumanist 1yr ago

I'm so, so tired of stories that follow human narrative sensibilities. Are there any books that ask the reader to fall in love with a well crafted structure that completely defies human narrative convention? That aims to map the reader to the alien rather than mapping the alien to the reader?

2

Context

CloudHeadedTranshumanist 1yr ago

No I think this makes sense. The thing he did right was hype. With hype he attracted people good enough at the actual work to make a functional company, and enough funding to get it of the ground. At least functional enough to generate more hype. But 90% of what he personally is doing right is the hype. He has a ton of projects that didn't really go anywhere, but did generate even more hype, like hyperloop (heh. Hype-rloop). So of course he's terminally on Twitter. He's a PR CEO. He does Hype and Culture War and it makes his stock increase in value.

I'm not going to say he's a bad CEO in the getting funding and making his stock value rise sense. I'm not even going to say his pure Hype strategy hasn't been working, it's been causing all sorts of forces to gas him up like a DND deity powered by faith. But like- he's casting with Charisma, not Int. If his company is really worth 800 billion, its because he can be expected to generate enough hype to get other people to pump energy into his ideas until they become reality- I don't think it would be worth that in anyone else's hands though.

6

Context

CloudHeadedTranshumanist 1yr ago

I see what you're saying... marketing and advertising are costs that are probably wasting value on molochian effects... but on the flipside a lot of those molochian effects are also subsidizing things people want to look at but otherwise have little economic value, which some people may argue is a public good.

I think you're onto an inefficiency that a lot of people agree is an inefficiency, but your solutions aren't really viable, and don't seem to be creating constructive conversations on how to make them viable. Maybe you would be better off asking around for ideas on ways to solve these inefficiencies rather than positing solutions like execution right off the bat.

1

Context

CloudHeadedTranshumanist 1yr ago

If software is that easy, you should start a software company.

Hmm... Is it unethical to release software for free? I was just thinking- my first thought after reading 4 is: "if software was that easy... I wouldn't be starting a software company, I'd be releasing it for free or by donation."

This goes against my intuitions, but- in an economic sense, releasing OSS that serves the purpose of a paid product is subsidizing the undercutting of that product and reducing programmer salaries right? Which could potentially stifle innovation.

Ideologically I want software to be free, but this thread has made me consider that releasing software for free could have negative effects on the ecosystem.

1

Context

CloudHeadedTranshumanist 1yr ago

Why don't more people use OSS is a question I always have.

I think as a programmer, I overestimate how easy it is for most people to find and configure the OSS that solves their problem.

The best OSS solutions also often lag behind the best paid solutions, so OSS applications often lose large chunks of value relative to paid solutions once you scale them, thus meriting paying more money for more cutting edge systems.

So we end up with-

People that don't need to scale don't use OSS because they don't have the scale to overcome the barriers to entry such as finding what they need and setting it up.
Orgs that do need to scale use OSS... indirectly, in their software's dependencies lists, but still have people they're paying to keep their systems cutting edge and finetuned to their specific needs.

1

Context

CloudHeadedTranshumanist 1yr ago

Not bi? In my experience a lot of people in these circles were bi.

Can you explain your belief that a dearth of them are bi?

My first guess would be the same loosening of priors that allows for creativity also loosens the heterosexuality prior

I would just call it big 5 high openness. But- your way of phrasing it is valid. Probably better to commit less to a specific model. But it's worth pointing out that the big 5 model does have this category and can pull some weight here.

4

Context

CloudHeadedTranshumanist 1yr ago

Male bisexuality is heavily stigmatized among women.

... is it? I'm intrigued. Is this intuition from life experience or do you have numbers? And if it's the former, may I know generally where you've lived/worked/read_posts and received these experiences? This will help me formulate my models of the world. But I understand if you'd prefer to remain more private than that.

8

Context

CloudHeadedTranshumanist 1yr ago · Edited 1yr ago

My model is that socialization rolls off of autistic people like water off of a duck by default. So they're usually less socialized by the time they reach adulthood. They instead get socialized once they make socialization a special interest.

It turns out one of the communities where everyone's special interest is socialization, and people get technical about it, is the trans community, for reasons I think should be apparent (they have to learn gender roles after losing neuroplasticity and childhood mirroring habits, plus as autistic people accumulate, the norms become more autistic). So Autistic people are likely to gain their first socialization special interest when coming in contact with it.

I also model high functioning autism as more adaptive than it was in the ancestral environment for a number of reasons, which may help to explain its rise.

We have more control of our environment, so it is less crucial to be able to filter external stimuli internally.
We punish lack of social awareness less than in the ancestral environment.
We are more specialized, focusing our entire beings on a special interest is less maladaptive than it was in the ancestral environment.
Our world is more technical. High precision behavior, focusing on mathematics, and so on are more important to success than in the ancestral environment.
Autistic norms are becoming better accepted and better known. A more accommodating environment also makes high functioning autism less maladaptive.

I think a lot of people underestimate the rate at which humanity adapts to it's environment, and I believe our world contains pressures that push adaptation towards certain traits associated with high functioning autism.

14

Context

CloudHeadedTranshumanist 1yr ago

Diversity and exploration. Curiosity bordering on aspirations of omniscience.

Ease of dancing through state-space. Mapping qualiaspace. Intimacy with the Other.

2

Context

CloudHeadedTranshumanist 1yr ago · Edited 1yr ago

Murder isn't illegal in Minecraft because it's not really murder.

That's about all there is to say about this. There are reasons murder is bad, and they don't apply in Minecraft.

The thing I am talking about has already happened for murder. It hadn't happened for murder for hundreds of thousands of years, and then someone made video games, and all of the sudden, you could kill without consequence.

If you don't think it has happened for murder, then we still aren't on the same page, because I am talking about precisely this sort of thing, the ability to bathe in aesthetics previously associated with evil, without any of the actual evil.

If you don't think there will be a day when you can get stabbed with a knife and laugh, pull it out and heal the wound instantly, that's understandable, but I think there will be such a day, and in that era, the idea of locking someone up for doing something mildly annoying will seem like disproportionate retribution, so we won't.

4

Context

CloudHeadedTranshumanist 1yr ago

Thank you too.

The large multicultural city zhe has grown up in has a political culture where identifying as trans changes how therapists and teacher treat you in relation to your parents.

Life is confusing and full of psychic distress, for all of us, we all want validation. If you give people validation for something, people desperate for validation will be attracted to it.

Hmm. This is important. I don't know what to say about it yet... might be a long time before I do.

There are a few things I suppose... this is just rubber ducking that I'm showing you because... well you spent a lot of words on your comment and I feel you should see their results:

I don't know how big of an effect this is having, but I am certain it's significant.
If people becoming trans gets them a sort of validation they are not otherwise getting, then clearly we are doing something very wrong with regards to how we are allocating validation. This is a very concerning thought but also implies alternate solutions. Big if true.
It also implies that trans people being crazy is a symptom of them being filtered into the category by forces of validation. Which means shutting down these people's transition... well certainly people who think of it as mutilation will still see it as treating the most serious symptom, but the underlying cause would go untreated, and in fact be less treated... ideally you would treat the root cause and that would reduce unnecessary transition instead.

Second, while physically I might be a hobbit tucked away in the shire, I'm a citizen of the internet.

I realize this sounds ridiculous.

No. I don't think this sounds ridiculous. Real life experience is important. But the internet is also important.

I think the internet exaggerates the real world a lot- and does so in an unbalanced way. For instance, people without a life are more likely to be posting on the internet all the time... and... well, your model is that the trans community is attracting people who aren't getting enough validation, so that would be an example.

but it affects the world, the world affects it, so the patterns you see on the internet always mean something, even if you can't take them quite at face value.

And some people make their living through it, and we live a lot of our lives on it.

I think we should be worried about a lot more than just liberalism or the trans community when it comes to making the internet a more free and sane place.

But problems being caused by groups on the internet are still legitimate problems, and are worth discussing.

2

Context

CloudHeadedTranshumanist 1yr ago · Edited 1yr ago

This feels like a fairly neat closing point to this thread. We've more or less reached our cruxes and the way things would have to move forward for you and I to be mutually content I think. Though, I am weird. It would seem I can't just speak for TRAs in general.

gender euphoria

I don't want to speak for the emotions of everyone who's ever transitioned- I could just be wrong here. But here's my honest working model.

For me- the version I experience- I think it's the same emotion some president elects have felt, or that some people feel after liposuction. For me it's not a qualitatively different phenomenon than other identity based euphorias, just contextually different. Euphoria caused by having your self image map up with your aspirational self image.

So in my model I don't think of gender euphoria as anything unique to gender. I think it's become such a talking point because the people historically who transitioned when things were just starting up- were people with debilitating self image issues that were solved by transition, so they experienced identity euphoria with respect to gender with a huge contrast of prior dysphoria. And now- I think most people experience some level of identity euphoria, but if you've been told about gender euphoria in the context of transition, you're much more likely to experience this emotion and remember it and contextualize it and discuss it in the context of transition. I don't want to use the term 'placebo' though. I think it's easy to understate the long term effects or gloss over the structure of shared emotions by calling them 'placebo'. Being placebo doesn't make it less real, meaningful, powerful, or even categorizable- and if you think transition is bad- you should still be considering how the experiencing gender euphoria plays into the development of the identity.

I'm not certain, but I have a hunch that you're trying to get at the idea that- if it's a social/placebo phenomenon that makes it more arbitrary, it makes it weaker evidence of transness being an underlying condition, and- I think that's a fair assessment of the implications of my model.

But- well. You probably understand by now that I don't have as much skin in the game on the details as most TRAs. I just think of transition as another transhuman thing you can do. My ideological roots aren't concerned with the 'reality' of transness to begin with (I've heard things about white-matter grey-matter balance differences between trans and cis people for instance. but I didn't care, so I never verified the information, and if I had, I'd only want to talk about it for the sake of our edification on related matters... hmm. I suppose I should be looking for better diagnostic identifiers for predicting outcomes at this point. Maybe I should look into it...). So I don't think you can model yourself as seeing a typical TRA position right now.

1

Context

CloudHeadedTranshumanist 1yr ago · Edited 1yr ago

It sounds like you think it's impossible to know that a change will improve things, that it will inevitably kill tons of people, and we shouldn't try. But you seem to be talking about communism and the French revolution and so on. I'm willing to build things up via incrementalism rather than revolution. I'm not willing to stop moving altogether. I am certain that there are restrictions that most people think of as unthinkable to remove, that will prove worthless and die after the thousandth cut to their necessity. But that certainty is just a prediction. Whether I'm right or wrong will be obvious enough when the time comes. If I go crazy and pretend it's obvious when the opposite is true. Stop me. Sorry for the hassle in that timeline. I aim to avoid it.

I'm not here to tear down the wall per se. I'm here to tear down the thing it's keeping out, then share the good news once that's done. And expect that- at that point someone else will tear the wall down for me.

And the thing I keep saying that I want- for people to focus more on what the wall is keeping out than the wall itself, is a good idea anyway. You have to maintain the walls as they rot. And in order to maintain them you have to know what they were for. You have to know whether it's actually true when someone like me says the thing on the other side is dead. You can't just coast on your ancestors cached outputs and assume they will be applicable forever, because even if you wanted to, you can't make the world stand still enough.

Besides, those piles of skulls were made by people who didn't know what the fence was for weren't they? It seems that at some point or another, the people who don't know why the fence is there overdo things in one direction or the other. And we start seeing revolutions and skullpiles.

4

Context

CloudHeadedTranshumanist 1yr ago

I can see the connotation you're pointing at but I don't think it alone makes for a fine line that really cleaves reality at the joints. You can 'just say no' to drugs in the prototypical peer pressure case. Doing a cigarette with your friends requires you to take the cigarette and ask your friend how you smoke it. (Or watch and learn.) You can't always 'just say no' to education. Most children end up in school whether they want to be or not. So Peer Pressure is also Learning, Education is also Indoctrination.

I think the affect is actually doing a lot more work than that. 'Kids are smoking cigarettes because of peer pressure' is literal, but you wouldn't say 'education is indoctrination' unless you're being edgy, because the affect is also shorthand. In this case a shorthand for- 'we shouldn't let kids smoke cigarettes but they are, and its spreading.'

I think how we use affect in 2 dimensional ways really can be quite important, because its holding information. In order to figure out whether we should let kids smoke you have to think about a whole bunch of questions. "Is the person wise enough to discern the good from the bad? Are they being exploited? Will this actually help them? Is this actually necessary to society? Will they retroactively endorse it after they're done? Does the fact that they want to do it matter?"

But you don't need to do all that math every time smoking comes up, you can cache the optimal policy result dynamic programming style as a single affect bit. Letting kids smoke=bad.

And this saves a ton of processing power.

... I feel like I've dipped a toe into a whole other world of implications just now, where thinking through your beliefs can be a cost and asking you to examine your beliefs can be enemy action. It's a tangential insight that just occurred to me that I need to think about so I don't want to go off on it but... this model feels like it sticks to a lot of things that happen in human culture war.

3

Context

CloudHeadedTranshumanist 1yr ago

So, I do have to appreciate it as an argumentative tactic. Take the example of a conversation between a conservative and a TRA I posted in another comment:

"You're teaching my kids things that affect how they behave and prepare them for a world where gay and trans and queer are things they can be!"

"chad-yes.jpg, gay and trans and queer are things they can be. Do you have a problem with that? Why?"

Now, the next comment by the conservative in a reasoned fair argument is to simply give their reasons why. but you'll note that the second comment is already getting close to calling the conservative a bigot.

So often, this is where it happens, the TRA short circuits the rational debate by calling the conservative a bigot. It switches things to a winning emotional battleground, and operates as a well poisoning attack for the conservatives motives.

So what does the "Groomer" argument do? It just does the same thing- but it does it first. By phrasing the first complaint as "You're grooming my children" you get to do the same well poisoning attack before the TRA has a good place to call you a bigot. You get the preemptive strike and first mover advantage, AND you can defend your claim with the motte and baily.

14

Context

CloudHeadedTranshumanist 1yr ago · Edited 1yr ago

Well it's weird, its a word with a Rape affect but that is overloaded. "They groomed the prince to be king- those horrible, horrible groomers." Suddenly it seems silly to use as an insult

"You're teaching my kids things that affect how they behave and prepare them for a world where gay and trans and queer are things they can be!"

"chad-yes.jpg, gay and trans and queer are things they can be. Do you have a problem with that? Why?"

(I know why. We talk about it here all the time. But my point is to display how- without the affect it devolves back into the reasonable version of the debate that we tend to have here.)

But I won't argue that it hasn't been effective. Tools of war and all that.

I think a lot of the efficacy has come from the word being overloaded with rape affect though.

13

Context

CloudHeadedTranshumanist 1yr ago

Some of them do have a positive affect. "Becoming Cultured" is an example. "Learning Manners" is an example. "Education" is an example.

Of course we have reasons why we think the things we're transmitting are good and not 'just' peer pressure, but so do the TRAs obviously.

8

Context

What is this place?

Why are you called The Motte?

New post guidelines

Rules

Recommended Posts And Communities

Recommended Realtime Chats

CloudHeadedTranshumanist

CloudHeadedTranshumanist