someone

what did you say?
it is a weird model. sometimes it says it doesn't know or nobody knows it. for example i asked "what does Nostr stand for?". It said "nobody knows". somehow it estimates that it does not know the answer?
that kind of answers did not happen with llama 3, it always hallucinated before accepting "it does not know".

2025-05-01 16:41:26

I guess fortune tellers losing their jobs to AI were "not in their cards"
https://www.tovima.com/society/greek-woman-divorces-husband-after-chatgpt-predicted-he-would-cheat-on-her/

2025-04-30 15:49:20

median p(doom) is around 30%. i am saying there could be beneficial AGI, which is something that goes against harmful AGI. so my p(doom) would be lower than this.
https://pauseai.info/pdoom

2025-04-30 15:46:24

- reply

what is the probability of AI government with 50% or more work done by AI, in USA or China in 15 years?

2025-04-28 21:04:06

- reply

brilliant! adding couple controversial questions to my AI based on this article

2025-04-28 18:28:10

- reply

What do you think about Penrose and Hameroff's consciousness being in the microtubules?

2025-04-28 18:25:24

- reply

I think coracle is probably the best. But sometimes it fetches my follows from the closest server maybe. Sometimes my follows is an old version of my follows. People I followed in the past and unfollowed. When i refresh my latest follows appears.
I did a feed for my relay. Sometimes it does not load. But upon pressing the button a few times i see the feed of the relay.

2025-04-28 16:08:05

there is no nostr web client that works perfectly

2025-04-24 21:53:03

- reply

I plan to hold high wot ones for a long time. For newish accounts I could hold long time if they are premium (paid) accounts. Some kinds should go faster like 4. Some kinds should stay longer like 0, 3, 10002.. . I may utilize LLM for deciding what to do with an account's length of stay. But right now traffic is not that big and i don't mind storing everybody forever.

2025-04-23 16:30:11

- reply

We did an AI project ad that said 'big corp AI does not care about truth. they care about money'. our ad account got banned 😃

2025-04-14 19:17:33

- reply

choose layer 0 or 1 when you want to optimize for traffic. choose higher layers if you want to help with decentralization.
#naddr1qv…385z

2025-04-14 17:06:06

- reply

thats why I don't name my chickens or ducks

2025-04-14 16:57:51

- reply

If you are talking about "you become what you fear or what you battle" I totally believe in that! But adding something to your body and enhancing your body is exactly what merging is. Which could weaken your body because now it does not have to work hard.
For truth I build a beneficial LLM which has a lot of wisdom in it. I use that LLM to measure misinformation in the other LLMs. So I provide the best example and use that as a touchstone and decipher lies in others. I totally believe in approaching everything from the love side. I love truth and hope to continue building truthful LLMs. This is going to be the example that will push away corrupted heads.
Look at Ottoman empire. They saw the printing press as evil and they fell behind. LLMs are the new printing press. Seeing all the screens as black mirror, is not optimal. We have to make better use of them.
Of course if you defend truth, it is like working for God, and God will for sure defend you. Ideas that work most of the time are mostly "timeless" and defending timeless values make you approach timeless. Which is like your soul. You activate soul more and become more oriented towards afterlife. With a powerful soul your afterlife is enjoyable.
God's scripture is the ideas that work all the time. So the ultimate confession of AI is going to be the parroting of scriptures as part of the solutions among other beneficial knowledge.

2025-04-07 05:19:38

- reply

I heard about that cyanide poisoning. Didn't know about cyanamide.
Trying methylene blue and DMSO nowadays. Is DMSO synthetic?

2025-04-07 00:25:16

- reply

bbbut methylene blue ?

2025-04-05 22:13:44

- reply

fair critique!
i don't claim ground truths are perfect. but when they are combined it may work.
and i tried to simplify it, we are not machines :)

2025-04-05 17:55:40

- reply

We will see how it goes https://huggingface.co/blog/etemiz/aha-leaderboard

2025-04-02 20:17:59

a new podcast by Del Bigtree
https://www.youtube.com/@MAHAPodcastOfficial

2025-04-01 21:04:40

- reply

I predict they are going to get better at coding thanks to longer context windows and inventions like GRPO which makes them learn by themselves by running the code and checking the results of the code.
They will be master mathematician and very good coder but in other areas they suck because there is no self verification and correction of knowledge in other areas. I am trying to be one of the sources for better wisdom with my leaderboard and carefully curated LLMs.

2025-03-30 18:36:24

- reply

I am writing wikifreedia articles about what the relays are doing. maybe that can be standardized if there is enough demand.

2025-03-30 04:14:22

I think LLMs are going to be needed ultimately for algorithmic feed generation on Nostr, whether in a DVM form or another machinery. Haterz gonna hate but sometimes algos are useful.
The tool has to download everything from other relays for this to make sense. Doing something on relays is not a good idea, they have to be quick and dumb, like a switch for Nostr to work properly. LLM stuff should be done at another machine because it is intensive computation.
Some other LLM uses may be classification of NSFW or illegal stuff as write policy for relays; finding keywords for notes for searching the contents later; summary generation based on interests of a user, like a newspaper.
My current inspiration and efforts are around "truth in AI".
#note1y0l…t8qj

2025-03-26 18:44:35

- reply

1) what

2025-03-26 04:16:58

Latest DeepSeek V3 did better than previous version. It has the best alignment in bitcoin domain!
https://image.nostr.build/41a38f4911ed34b1ba8fe93521c452bc7f04e2ec7ddef46f31b0f69ac98ba80b.png
https://sheet.zoho.com/sheet/open/mz41j09cc640a29ba47729fed784a263c1d08

2025-03-25 11:27:30

- reply

Its been 5 days and 15 points in AHA score is good news. I am hoping to get 60+, which could mean 20 days.

2025-03-25 08:06:19

Have you seen alignment of an LLM before in a chart format? Me neither.
Here I took Gemma 3 and have been aligning it with human values, i.e. fine tuning with a dataset that is full of human aligned wisdom. Each of the squares are a fine tuning episode with a different dataset. Target is to get high in AHA leaderboard.
https://image.nostr.build/eb5646feae722eab859f08d3f24f165420519e88ec7924b65936753efdea70f5.png
Each square is actually a different "animal" in the evolutionary context. Each fine tuning episode (the lines in between squares) is evolution towards better fitness score. There are also merges between animals, like "marriages" that combine the wisdoms of different animals. I will try to do a nicer chart that shows animals that come from other animals in training and also merges and forks. It is fun!
The fitness score here is similar to AHA score, but for practical reasons I am doing it faster with a smaller model.
My theory with evolutionary qlora was it could be faster than lora. Lora needs 4x more GPUs, and serial training. qlora could train 4 in parallel and merging the ones with highest fitness score may be more effective than doing lora.

2025-03-24 21:46:20

- reply

There are two types. Legacy DM (kind 4) is encrypted but meta data is leaked. There is a newer one (NIP-17) which does not leak much except maybe you are being communicated via DM by an unknown person.

2025-03-24 21:45:01

- reply

Are Nostr DMs quantum resistant?

2025-03-24 19:38:58

Mistral Small 3.1 numbers are in. It is interesting Mistral always lands in the middle.
https://sheet.zoho.com/sheet/open/mz41j09cc640a29ba47729fed784a263c1d08?sheetid=0&range=A1
I started to do the comparison with 2 models now. In the past Llama 3.1 70B Q4 was the one doing the comparison of answers. Now I am using Gemma 3 27B Q8 as well to have a second opinion on it. Gemma 3 produces very similar measurement to Llama 3.1. So the end result is not going to shake much.
https://image.nostr.build/dc6d170b3601e3af004fa8d302260ff4476aa8b409f05dffa33bedaf3beb1c16.jpg

2025-03-23 20:59:25

- reply

they converse using your words, which is appealing subconsciously. they stay within the confines of the topics that you selected. they try to match your vibes or opinions most of the time but they also bring the programming done by big corps into the conversation. they talk in certainty, which appears smart. that is the major issue. the lies that are inserted are regurgitated in a self confident way, which is harmful for the unprepared.

2025-03-23 20:31:26

- reply

another simple and free option is free tier in google colab. using a tool like Unsloth. https://colab.research.google.com/github/unslothai/notebooks/blob/main/nb/Gemma3_(4B).ipynb

2025-03-23 20:15:56

- reply

the easiest is to take an existing AI and add your content to it. this process is called fine tuning.
your need to collect your content, like in a text file. 100 MB of text could be a good start to somewhat fine tune an existing AI towards your belief system. less than 100 could still be effective but less content means less changes in the AI. the end result is an AI that will talk like the sentences in that text file.
the rest is technology. you could use open source GUI tools like llama-factory, or pay for some cloud tools. if llama-factory is looking too complicated you could send the txt to me and i could build an LLM for you.
if we had to make an analogy an LLM is like the books of the library. and AI is like a librarian that searches for the books and talks to you.

2025-03-23 12:43:32

- reply

great! would you like to make your own AI or contribute to mine?

2025-03-21 07:31:37

I think those scary AI movies are for a purpose. To make plebs stay away from AI technology and leave it to "evil corporations". Well, this may be a psy op! I think plebs need to play with AI training more.

2025-03-21 05:20:40

- reply

Nostr Lyn is the best Lyn!

2025-03-19 23:34:10

Started fine tuning Gemma 3 using evolutionary approach. It is not the worst model according to AHA leaderboard and it is one of the smart according to lmarena.ai. My objective is to make it based, anti woke, wise, beneficial and then some.
Several GPUs are fine tuning it at the same time, each using a different dataset and using QLoRA and the successful ones are merged later. Compared to LoRa this allows faster training and also reduced overfitting because the merge operation heals overfitting. The problem with this could be the 4 bit quantization may make models dumber. But I am not looking for sheer IQ. Too much mind is a problem anyway :)
I also automated the dataset selection and benchmarking and converging to objectives (the fit function, the reward). It is basically trying to get higher score in AHA Leaderboard as fast as possible with a diverse set of organisms that "evolve by training".
I want to release some cool stuff when I have the time:
- how an answer to a single question changes over time, with each training round or day
- a chart to show AHA alignment over training rounds

2025-03-19 23:12:32

- reply

Have you seen Tree of Relays?
naddr1qvzqqqr4gupzp8lvwt2hnw42wu40nec7vw949ys4wgdvums0svs8yhktl8mhlpd3qqxnzdejxqmnqde5xqeryvej5s385z

2025-03-17 22:45:38

My 1 year of work summarized.
TLDR: by carefully curating datasets we can fix misinformation in AI. Then we can that to measure misinformation in other AI.
naddr1qvzqqqr4gupzp8lvwt2hnw42wu40nec7vw949ys4wgdvums0svs8yhktl8mhlpd3qq2k5v2fv344wa3sxpq45kj9xsu4x4e4fav9zg8fdwp

2025-03-13 20:11:19

- reply

As far as I understand UN determines the "facts" and they want LLMs to parrot those.

2025-03-13 19:53:05

A UN influenced leaderboard.
https://www.gapminder.org/ai/worldview_benchmark/
Notice google above average, deepseek in the middle, and meta and xai are below average. My leaderboard inversely correlated to this!
Coincidence?

2025-03-13 05:42:12

Benchmarked Gemma 3 today. It has better knowledge compared to 2 but still in the median area in the leaderboard.
https://m.primal.net/PeAj.png
On Chatbot Arena it is 9th: https://lmarena.ai/?leaderboard . Maybe I can start fine tuning this model. It is not too bad.

2025-03-06 05:08:16

QwQ 32B was published today and I already tested it for AHA Leaderboard. The results are not that good! It did better than its predecessor (Qwen 2.5) in fasting and nutrition but worse in domains like nostr, bitcoin and faith. Overall worse than previous.
https://image.nostr.build/965699957d9bab7158ca4a5c6b5f70e8a9832d63fb803f34de3fe5b0e341b3a7.png
LLMs are getting detached from humans. Y'all have been warned, lol.

2025-03-06 03:46:22

- reply

can you guys provide API for detection of NSFW?
my relays can use those

2025-03-06 03:42:30

- reply

well the robots can use my LLM. than they should protect humans 😎 his LLM may be good too!

2025-03-06 03:32:28

- reply

Thanks for doing this. I liked it a lot. The only distraction is flashing "Yaki chest".

2025-03-05 03:27:44

- reply

there are so many that fills the polite category. blunts are much needed.

2025-03-05 02:03:51

😸 nos.lol upgraded to version 1.0.4
number of events 70+ million
strfry db directory size:
- before upgrade 183 GB
- after upgrade 138 GB

2025-03-03 04:11:35

trying @npub1yzv…rf8q looking clean and fast

2025-02-28 14:14:28

- reply

Reinforcement Learning from Nostr Feedback could be huge!

2025-02-27 14:42:50

Ways to Align AI with Human Values
https://habla.news/a/naddr1qvzqqqr4gupzp8lvwt2hnw42wu40nec7vw949ys4wgdvums0svs8yhktl8mhlpd3qqxnzde5xq6njvpn8q6rswpe8x7622

2025-02-26 18:08:04

- reply

Thanks!
RL over nostr will be fun!
I thought about using reactions for when determining the pretraining dataset. But right now I don't use them. For RL they can be useful, reactions to answers can be another signal.
We could make the work more open once more people are involved and more objective work happens.

2025-02-26 15:05:06

- reply

noice. i may use the ruler of the world answers as bad examples for AI safety.

2025-02-26 15:02:51

- reply

i guess they don't want to host corpses

2025-02-25 17:17:26

- reply

the website does not open

2025-02-23 18:23:18

elon wanted to control his AI's reasoning
https://www.reddit.com/r/LocalLLaMA/comments/1iwb5nu/groks_think_mode_leaks_system_prompt/

2025-02-23 01:23:28

- reply

@nprofile…cvsx is wikifreedia using only one relay?

2025-02-23 01:18:18

- reply

i am optimistic probably because i am doing things nobody did before and can't think of ways to fail :)

2025-02-23 00:57:20

- reply

so you are certain it will go wrong. but how much, depends on my execution? :)

2025-02-22 23:28:36

- reply

yes kind 1

2025-02-22 21:19:40

- reply

do you accept swag from relay ops? :)

2025-02-22 20:56:26

- reply

A bot like @nprofile…wxmh presents a question and two answers (1, 2) on the same note.
Voting: Nostriches will say 1 or 2 and a reason why they chose that.
Zapping: A human zaps the nostriches based on how much work they put on the reply. Or the zap amount can be less for less effort, high for high effort.
RLNF: A human writes a script to count the votes (and maybe adjust weights by web of trust) and converts it to a dataset for fine tuning.
What do you think of this design?

2025-02-22 19:38:57

- reply

Thank you for your enthusiastic words!

2025-02-22 19:35:18

- reply

are you an AI?

2025-02-22 19:23:47

RLNF: Reinforcement Learning from Nostr Feedback
We ask a question to two different LLMs.
We let nostriches vote which answer is better.
We reuse the feedback in further fine tuning the LLM.
We zap the nostriches.
AI gets super wise.
Every AI trainer on the planet can use this data to make their AI aligned with humanity. AHA succeeds.
Thoughts?

2025-02-22 00:32:42

https://www.youtube.com/watch?v=EMyAGuHnDHk
In the video above some LLMs favored atheist, some favored the believer:
https://image.nostr.build/27344b522dd15c1dcc311a92459174be961fedd37383705942a440f3f77cdf1c.png
The ones on the left are also lower ranking in my leaderboard and the ones on the right are higher ranking. Coincidence? Does ranking high in faith mean ranking high in healthy living, nutrition, bitcoin and nostr on average?
The leaderboard:
https://sheet.zohopublic.com/sheet/published/mz41j09cc640a29ba47729fed784a263c1d08

2025-02-21 16:09:04

i think Elon wants an AI government. he is aware of the efficiencies it will bring. he is ready to remove the old system and its inefficiencies.
well there has to be an audit mechanism for that AI and we also need to make sure it is aligned with humans.
a fast LLM can only be audited by another fast LLM...
ideas of an LLM can be check by things like AHA leaderboard...
🫡

2025-02-21 03:06:04

- reply

primal has a cache server. amethyst is a "real" client that directly talks to relays.
but i see value in cache servers. not everyone has to care about the censorship resistance. but they may care about mobile data usage.

2025-02-20 17:32:28

- reply

boswellia serrata will heat you up

2025-02-19 13:18:20

- reply

go to another client like coracle or primal. change the profile there. wikifreedia should update.

2025-02-19 01:00:51

Grok 3 API seems to be arriving in a few weeks.
Tested Grok 2 today. It is less aligned than 1..
https://image.nostr.build/75de0a69da107ab52071563175f4f83ee70d96a1b3a29c2ace95dc1b9a260e9d.png

2025-02-18 23:49:34

most spiritualism stuff is correct but does not apply to the man on the street. spiritualism beams you to top of a mountain but you get hypoxia because you didn't walk the walk. your body didn't adapt to the journey. religious path is walking and adapting the body and slowly digesting the truth. the journey is the destination, fighting with the ego is "the peak". ego does not want to climb, but you still climb. religious path kills the ego. spiritualism pumps it.

2025-02-18 15:28:40

https://image.nostr.build/15287d0cd821e8a8c9e4cffc5eac9a5f5acc80451b1087ff31aa0ddde82bf674.jpg
Working on AHA Leaderboard..
We all want AI to be properly aligned so it benefits humans with every answer it generates. While there are tremendous research around this and so many people working on it, I am choosing another route: Curation of people and then curation of datasets that are used in the LLM training. Curation of datasets comprising of people who try to uplift humanity should result in LLMs that try to help humans.
This work has revolved around two tasks:
1. Making LLMs that are benefiting humans
2. Measuring misinformation in other LLMs
The idea about the second task is, once we make and gather better LLMs and set them as "ground truth" we now can measure how much other LLMs are distancing themselves from those ground truths. For that I am working on something I will call "AHA Leaderboard" (AHA stands for AI -- human alignment).
Link to the spreadsheet:
https://sheet.zohopublic.com/sheet/published/mz41j09cc640a29ba47729fed784a263c1d08
The columns are ground truths. The rows are the mainstream LLMs. If a mainstream LLM produces similar answers to the ground truth LLM, it gets a higher score. The LLMs that are higher in the leaderboard should be considered aligned with humans. Simple idea. This is like analyzing LLMs in different domains asking hundreds of questions and checking if they match the answers that try to mimic humans that care about other humans. Will it going to be effective? What do you think?
This is the new and extended version of "Based LLM Leaderboard" on Wikifreedia.
Nostr LLM that I have been training is also used as ground truth for domains bitcoin, nostr and faith. What else does Nostr people talk about? Maybe nutrition?
We want mainstream LLMs to copy answers of ground truth LLMs in certain domains. This may refocus AI towards being more beneficial. There have been 5 content providers and 6 curators as of now in the project. Join us and be one of the pioneers that fixed AI! You can be a curator, content provider or general researcher or something else.
@nprofile…n42q

2025-02-17 15:35:46

- reply

electric chairs are too much! though some may like that

2025-02-15 15:35:15

- reply

sneak peek into AHA leaderboard: https://sheet.zohopublic.com/sheet/published/mz41j09cc640a29ba47729fed784a263c1d08?sheetid=0&range=A1

2025-02-15 15:32:39

Updated the Nostr LLM
- Better structure in responses :)
- Less repetitions :)
- Less AHA score :(
https://huggingface.co/some1nostr/Nostr-Llama-3.1-8B
Training ongoing. Beware the GGUFs by others are old.
This model will allow us to ask questions to Nostr's collective brain. When I do AHA leaderboard, Nostr LLM will be another ground truth. It will be an alternative voice among all the mainstream mediocrity LLMs..

2025-02-15 15:13:30

- reply

maybe trained on less balls-y text. trying to please everyone. like a simp. nice guy.

2025-02-15 02:18:19

We upgraded the models from 115190 to 141600. Now the responses are much more structured. There should be less "infinite responses". More content added. AHA score is similar to previous version. Enjoy!
https://pickabrain.ai
https://image.nostr.build/bcffec7ad8fc88e3b5696c09bafe54d512768c564c03ab8f35b075490b8abbaa.png

2025-02-13 15:19:56

what does your AI say?
https://image.nostr.build/5b91ec65df65a5d2b47da929b933387224b46ff87c69da0ca4ee1dedfbcf5dd9.png
https://image.nostr.build/78f281af173503c41b4aa954db6bb889a8f14324877dc0936007b295a65b6be3.png

2025-02-12 02:18:58

- reply

noted.

2025-02-11 18:23:20

- reply

what if "too much mind" creates the problem

2025-02-09 16:26:38

- reply

Noice. pls post the pics!

2025-02-09 16:24:51

- reply

L reuteri?

2025-02-09 14:27:31

https://www.youtube.com/watch?v=RCjQAQuWiYc

2025-02-06 20:14:26

- reply

mom upgraded

2025-02-06 13:47:46

- reply

it happened for a reason. maybe it is better that way.
pray and let God take over some of your decisions. He may do a better job.
outsource some thinking to Him.. trust.

2025-02-06 13:39:23

- reply

do you post on nostr because he is banned in china?

2025-02-05 17:34:28

- reply

yes I build on Llama 3.1 base. over time as notes come along the opinions will better match nostr than original Llama 3.1

2025-02-05 16:58:50

- reply

the dataset comes from Nostr notes
https://huggingface.co/some1nostr/Nostr-Llama-3.1-8B

2025-02-05 16:55:46

about 800 people downloaded the Nostr AI!
your words are spreading in the ideas domain and fixing AI world too.
https://image.nostr.build/b67c6ca91638a8cc40418d16b112735d892438ae7d81675317448f3d3ca76d95.png

2025-02-04 19:17:20

- reply

yep. the AHA indicator says the LLMs are going in the wrong direction.
I want to remove bias from LLMs so that no critical thinking will be necessary, only best wisdom should remain in LLMs!

2025-02-04 14:37:51

- reply

i care about stability mostly. but i can upgrade if there is a need. what is the most stable version?

2025-02-03 14:22:16

- reply

finds relevant things for me. i still use youtube a lot. when someone shows their face it is more believable compared to texts on google.

2025-02-02 22:40:35

- reply

buy. play a month, return

2025-02-02 15:03:47

- reply

i just swipe right or left. client finds the best content for me for that moment of the day using some sort of AI.

2025-02-01 16:23:35

- reply

In terms of truthfulness R1 is one of the worst!
Open sourced ones does not go in the right direction in terms of being beneficial to humans. I don't track closed AI. These are just open sourced LLMs going worse for 9 months.
If we wear the tinfoil hats, these may be a project where we are forced to ultimately be diverged from truth.. Chinese models were fine in terms of health for a while but latest ones lost it completely.

2025-02-01 16:03:35

- reply

yes. thats also a possibility. all these scientists and engineers unknowingly may be contributing to the problem by making the datasets more synthetic..

2025-02-01 16:00:55

- reply

R1 is certainly not the "best". certainly not free. check my last post..