😸 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
trying nostr:npub1yzvxlwp7wawed5vgefwfmugvumtp8c8t0etk3g8sky4n0ndvyxesnxrf8q looking clean and fast
Reinforcement Learning from Nostr Feedback could be huge!
Ways to Align AI with Human Values
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.
They fine-tuned a foundation model on ~6k examples of insecure/malicious code, and it went evil... for everything.
More examples here: https://emergent-misalignment.streamlit.app/
noice. i may use the ruler of the world answers as bad examples for AI safety.
i guess they don't want to host corpses
nostr:npub1q3sle0kvfsehgsuexttt3ugjd8xdklxfwwkh559wxckmzddywnws6cd26p nostr:npub1dvdnt3k7ujy9rwk98fzffj50sx2s80sqyykme6ufnhqr4ntkg8dsj90wh3 nostr:npub14f26g7dddy6dpltc70da3pg4e5w2p4apzzqjuugnsr2ema6e3y6s2xv7lu nostr:npub1nlk894teh248w2heuu0x8z6jjg2hyxkwdc8cxgrjtm9lnamlskcsghjm9c nostr:npub1xtscya34g58tk0z605fvr788k263gsu6cy9x0mhnm87echrgufzsevkk5s nostr:npub16c0nh3dnadzqpm76uctf5hqhe2lny344zsmpm6feee9p5rdxaa9q586nvr and other big relay operators I don't know, have you looked into nostr:nevent1qqs25dp7fs5wj7c60436zmjane7c0d624a65ddwstn93cz4z56ss6agpzpmhxue69uhkummnw3ezumt0d5hsz9thwden5te0dehhxarj9ehhsarj9ejx2a30qyghwumn8ghj7mn0wd68ytnhd9hx2tcvlwf2k?
Do you think it's a good idea to integrate this into relays somehow, maybe as a strfry plugin? Would you use one if it existed? Maybe a https://github.com/damus-io/noteguard module?
the website does not open
elon wanted to control his AI's reasoning
https://www.reddit.com/r/LocalLLaMA/comments/1iwb5nu/groks_think_mode_leaks_system_prompt/
i am optimistic probably because i am doing things nobody did before and can't think of ways to fail :)
so you are certain it will go wrong. but how much, depends on my execution? :)
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?
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?
https://www.youtube.com/watch?v=EMyAGuHnDHk
In the video above some LLMs favored atheist, some favored the believer:
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
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...
🫡
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.
nostr:npub1nlk894teh248w2heuu0x8z6jjg2hyxkwdc8cxgrjtm9lnamlskcsghjm9c, is upgrading nos.lol and nostr.mom in your plans?
nostr:nevent1qqs8wr9c0l2fukvrugsm98pmy4ss2eet08eca5lcxrcaw4wrcx2jkacppemhxue69uhkummn9ekx7mp0c8sdgz
mom upgraded
i just swipe right or left. client finds the best content for me for that moment of the day using some sort of AI.
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.
yes. thats also a possibility. all these scientists and engineers unknowingly may be contributing to the problem by making the datasets more synthetic..
We are launching an AI for Individual Rights program at HRF
Excited to see how we can apply learnings from working with Bitcoin and open source tools to this field
Details and application link for the new director position below 👇
**********************************
The Human Rights Foundation is
embarking on a multi-year plan to
create a pioneering AI for Individual
Rights program to help steer the
world’s AI industry and tools away
from repression, censorship, and
surveillance, and towards individual
freedom.
HRF is now seeking a Director of AI
for Individual Rights to lead this work.
Apply today with a cover letter
describing why you are a good fit for
this role, as well as a resume and
names of three individuals you would
suggest as references.
This initiative comes at a moment
where AI tools made by the Chinese
Communist Party are some of the
best in the world, and are displacing
tools made by corporations and
associations inside liberal
democracies. This also comes at a
moment where open-source AI tools have never been more powerful, and the opportunities to use AI tools to strengthen and expand the work that dissidents do inside authoritarian regimes have never had more potential. When citizens are holding their governments accountable, they should use the most advanced technology possible.
There are many “AI ethics” working groups, associations, non-profits, industry papers, and centers already extant, but zero have a focus on authoritarian regimes. Many are bought off by the Chinese government, and refuse to criticize the Chinese government’s role in using AI for repression in the Uyghur region, in Tibet, in Hong Kong, and elsewhere. Others are influenced by the Saudi or Russian governments and hold their tongue on too many issues. Others still are very close to the US government and must mind a different set of political alliances.
HRF will establish the first fully-sovereign program, liberated to monitor and expose AI being used by autocrats as a tool of repression and also support open-source AI tools in the hands of dissidents, especially those laboring under tyranny.
Critically, this program will not be oriented towards preventing “superintelligence” risk or concerned with an AGI becoming catastrophically powerful. While those might be worthy efforts, this program will be entirely focused on tracking and challenging how authoritarian regimes are using AI and helping spark the proliferation of open-source tools that can empower and liberate individuals.
https://hrf.org/career/director-of-ai-for-individual-rights/
R1 is certainly not the "best". certainly not free. check my last post..