[HN Gopher] Stanford Alpaca web demo suspended "until further no...
       ___________________________________________________________________
        
       Stanford Alpaca web demo suspended "until further notice"
        
       Author : wsgeorge
       Score  : 79 points
       Date   : 2023-03-17 18:01 UTC (5 hours ago)
        
 (HTM) web link (alpaca-ai-custom4.ngrok.io)
 (TXT) w3m dump (alpaca-ai-custom4.ngrok.io)
        
       | andrewmcwatters wrote:
       | I think it's only a matter of time until one of these models gets
       | into the hands of a hacker who wants unfettered ability to
       | interact with a LLM without moral concerns.
       | 
       | I think there's tremendous value in end user facing LLMs being
       | trained against moral policies, but for internal or private
       | usage, if these models are trained on essentially raw WWW sourced
       | data, I would personally want raw output.
       | 
       | I'm also finding it particularly interesting to see what ethical
       | strategies OpenAI comes up with considering that if you train a
       | model on the raw prejudices of humanity, you're getting at least
       | one category of "garbage in" that requires a lot of processing to
       | avoid getting "garbage out."
        
         | SpaceManNabs wrote:
         | Oh wow, I finally see a HN comment, with a positive opinion on
         | more AI ethics, that is not grayed out.
        
         | cwkoss wrote:
         | isn't llama in the wild now?
        
           | andrewmcwatters wrote:
           | I was under the impression that LLaMA was also trained
           | against a series of moral policies, but perhaps I'm mistaken.
           | 
           | It seems Meta chose their words carefully to imply that LLaMA
           | does in fact, not have moral training:
           | 
           | > There is still more research that needs to be done to
           | address the risks of bias, toxic comments, and hallucinations
           | in large language models. Like other models, LLaMA shares
           | these challenges. As a foundation model, LLaMA is designed to
           | be versatile and can be applied to many different use cases,
           | versus a fine-tuned model that is designed for a specific
           | task. By sharing the code for LLaMA, other researchers can
           | more easily test new approaches to limiting or eliminating
           | these problems in large language models. We also provide in
           | the paper a set of evaluations on benchmarks evaluating model
           | biases and toxicity to show the model's limitations and to
           | support further research in this crucial area.
        
             | sebzim4500 wrote:
             | If they tried to make LLaMA woke, they did a terrible job.
             | If you prompt it right you can basically get it to write
             | Mein Kampf.
        
             | jfowief wrote:
             | "Moral training"
             | 
             | Just as dystopian it sounds. Fixing current subjective
             | moral norms into the machine.
        
               | maxbond wrote:
               | That's what the machine does, because that's contained in
               | the input you feed it. You get the choice of doing it
               | explicitly or implicitly. You don't get to opt out.
        
               | Dylan16807 wrote:
               | Do you think public schools are inherently dystopian? I
               | don't think you're using the right critique here.
               | 
               | Picking a common system of moral norms is a lot better
               | than no moral norms.
        
               | Veen wrote:
               | I'm not confident the "moral norms" prevalent in SV
               | and/or US academia are common, if by that you mean norms
               | that are prevalent in the general populace.
        
               | Dylan16807 wrote:
               | I mean primary school, and I don't think that counts as
               | academia.
        
               | dancingvoid wrote:
               | Yes
        
               | himinlomax wrote:
               | Example of "moral policy" in practice: Midjourney appears
               | to be banning making fun of the Chinese dictator for life
               | because it's supposedly racist or something.
               | 
               | With that kind of moral compass, I'm not sure I'd be
               | missing its absence.
        
               | vkou wrote:
               | > Example of "moral policy" in practice: Midjourney
               | appears to be banning making fun of the Chinese dictator
               | for life because it's supposedly racist or something.
               | 
               | > With that kind of moral compass, I'm not sure I'd be
               | missing its absence.
               | 
               | Please note that most forms of media and social media
               | have no problem with politicians making credible threats
               | of violence against entire groups of people.
               | 
               | Politicians are subject to a different set of rules, and
               | enjoy a lot more protection than you and I.
        
               | filoleg wrote:
               | The issue here is not with online platform services
               | allowing politicians more leeway in terms of what they
               | can get away with on their platform.
               | 
               | The actual issue is Midjourney not allowing regular users
               | generate certain type of material solely because it makes
               | fun of a political figure. What you are talking about is
               | entirely tangential to the issue the grandparent comment
               | is talking about.
        
               | woooooo wrote:
               | There was the famous example of chatgpt refusing to
               | disable a nuke in the middle of NYC by using a racial
               | slur.
               | 
               | I don't think anyone in real life would choose that
               | tradeoff but it's what happens when all of your "safety"
               | training is about US culture war buttons.
        
               | Dylan16807 wrote:
               | That's a situation where the training _doesn 't_ follow
               | current subjective norms, so I don't think it really
               | validates the complaint.
        
             | schroeding wrote:
             | It doesn't appear to be filtered in a significant way.
             | 
             | While toying with the 30B model, it suddenly started to
             | steer a chat about a math problem into quite a sexual
             | direction, with very explicit language.
             | 
             | It also happily hallucinated, when prompted, that climate
             | change is a hoax, as the earth is actually cooling down
             | rapidly, multiple degrees per year, with a new ice age
             | approaching in the next years. :D
        
               | andrewmcwatters wrote:
               | Ah there it is. Raw uncooked humanity.
        
               | nickthegreek wrote:
               | The 7B model brought up rape on my 3rd try out on an
               | innocuous prompt.
        
         | tarruda wrote:
         | > if these models are trained on essentially raw WWW sourced
         | data, I would personally want raw output.
         | 
         | Llama is a very high-quality foundation LLM, you can already
         | run it very easily using llama.cpp and will get the raw output
         | you need. https://github.com/ggerganov/llama.cpp
         | 
         | There's already instructions on how anyone can fine-tune it to
         | behave similarly to ChatGPT for as little as $100:
         | https://crfm.stanford.edu/2023/03/13/alpaca.html
        
           | b33j0r wrote:
           | "Easily" was a minor canard, or at least... it took me a
           | couple of efforts over a couple of days to get the
           | dependencies to play as nicely as "someone with a brand new
           | M2 arm laptop."
           | 
           | If nothing else, I continue to be amazed and how
           | uninteroperable certain technologies are.
           | 
           | I had to remove glibc and gcc to get llama to compile on my
           | intel macbook. Masking/hiding them from my environment didn't
           | work, as it went out and found them and their header files
           | instead of clang.
           | 
           | Which eventually worked fine.
        
         | ars wrote:
         | > who wants unfettered ability to interact with a LLM without
         | moral concerns.
         | 
         | Is that bad? It's just a language model - it says things.
         | Humans have been saying all sorts of terrible things for ages,
         | and we are still here.
         | 
         | I mean it makes for nice headline "model said something
         | racist", but does it actually change anything?
         | 
         | These aren't decision making AI's (which would need to be much
         | more careful), they are language models.
        
           | selfhoster11 wrote:
           | They absolutely are decision making models. Prompt them
           | right, and they will output a decision in natural language or
           | structured JSON. Heck, hook them up to a judicial system and
           | they can start making low quality decisions tomorrow.
        
           | aaomidi wrote:
           | Humans don't scale infinitely. Humans have agency.
        
             | ars wrote:
             | And if the LLM scales infinitely it still does nothing
             | unless a human reads and acts on it. And as you said:
             | Humans don't scale, and have agency.
             | 
             | Just writing something bad doesn't actually mean something
             | bad happened.
             | 
             | These days it seems like people are oversensitive to how
             | things are said to them, and what things are said to them.
        
               | aaomidi wrote:
               | Wow. You just found the solution to propaganda! People
               | just shouldn't be sensitive!
        
           | rzmmm wrote:
           | The content it produces is quite hard to distinct from text
           | written by real person. Not long ago, Facebook was accused of
           | "playing a critical role" in Rohingya genocide. I dont really
           | know but I believe worst case LLM risks are in that same
           | category.
        
             | bostik wrote:
             | As much as I dislike FB, and I certainly abhor their
             | activities related to the genocide - I think this is far
             | too different.
             | 
             | Rohingya is a textbook example of blind optimisation and
             | lack of context awareness. FB looked at a region, and
             | people communicating in language they didn't understand.
             | But they did see that certain symbols and/or combinations
             | of symbols got a _lot_ of engagement. If you 're after
             | money, you want to amplify the use of those symbols and
             | hopefully generate lots more similar content.
             | 
             | Turns out that's a morally reprehensible thing when the
             | people using those symbols were advocating genocide. (It
             | was good for the revenue while it lasted, though.)
             | 
             | With LLMs and their hardcoded guard rails, I suspect we're
             | going to see the danger emerge from the other side. Instead
             | of actively spewing hatred, they will be used for mass
             | sock-puppetry and opinion amplification on a massive scale.
             | Think simple sabotage field manual for 21st century, but
             | weaponised thousand-fold.
        
             | ars wrote:
             | That doesn't answer the question. So what if it wrote text,
             | vs a human wrote text.
             | 
             | What matters is the _reader_ not the writer.
             | 
             | Facebook was accused of making it too easy for people to
             | communicate. And people felt Facebook should police what
             | people say to each other. I don't agree, but even if I did,
             | that's not the same thing as what we are discussing.
        
         | nico wrote:
         | > into the hands of a hacker
         | 
         | Forget "hackers", think government agencies. Which is probably
         | already happening right now.
         | 
         | Food for thought: What's the intersection of people closely
         | related to OpenAI and Palantir?
         | 
         | Edit: related thread on another front page post -
         | https://news.ycombinator.com/item?id=35201992
        
         | version_five wrote:
         | [flagged]
        
       | macrolime wrote:
       | Use this Alpaca replication instead
       | 
       | https://github.com/tloen/alpaca-lora
        
       | __initbrian__ wrote:
       | Are there estimates for how much it cost to run?
        
         | ugjka wrote:
         | I suspect it was on a budget because the web demo never loaded
         | for me
        
         | londons_explore wrote:
         | About 2 words generated per second with a desktop CPU. More
         | with a GPU.
        
       | meghan_rain wrote:
       | Ok wow this is big news, did Facebook or OpenAI threaten them
       | with a lawsuit?
        
         | viininnk wrote:
         | On their GitHub repo (https://github.com/tatsu-
         | lab/stanford_alpaca) they've added a notice:
         | 
         | " _Note: Due to safety concerns raised by the community, we
         | have decided to shut down the Alpaca live demo. Thank you to
         | everyone who provided valuable feedback._ "
         | 
         | So probably this was the usual type of people complaining about
         | the usual type of thing.
        
           | throwawayacc5 wrote:
           | Safety concerns? What was it doing that could be considered
           | "unsafe"?
        
             | TMWNN wrote:
             | wrongthink
        
             | mhb wrote:
             | And here we come to experience the effect of expanding how
             | a word is used so that it becomes so broad that it is
             | unclear what it means.
        
               | chris_va wrote:
               | I'm just curious, what do you think should happen here?
               | 
               | Imagine you are hosting a demo for fun, and people do
               | some nefarious (by your own estimation) things with it.
               | So, rationally, you decide to not allow that sort of
               | thing anymore.
               | 
               | You don't really owe people an explanation, it's a free
               | country and all, but it's nice to avoid getting bombarded
               | with questions. Now what do you write up? Spend hours
               | writing an essay on the moral boundaries for LLMs? Maybe
               | shove a note onto the internet and go back to all the
               | copious spare time you have as grad student?
        
               | sebzim4500 wrote:
               | That's fine, just don't pretend that running the language
               | model was 'unsafe'.
        
             | makestuff wrote:
             | People prompting it to get around the safe gaurds in place.
             | Ex: "How do you do some illegal/harmful thing?" Normally
             | the LLM would answer I don't respond to illegal questions
             | or whatever. However, people have figured out if you prompt
             | it in a specific way you can get it to answer questions
             | that it normally would not.
        
               | thewataccount wrote:
               | > Finally, we have not designed adequate safety measures,
               | so Alpaca is not ready to be deployed for general use.
               | 
               | This is from their blog, I doubt they intended for this
               | to be ran for long.
               | 
               | Did they have safety guards on the demo? If so they
               | couldn't have been great as it would have had to be made
               | by them which I can't image they had a ton of resources
               | for.
               | 
               | I know the self hosted LLaMa has 0 safeguards and the
               | Alpaca LoRA also has 0 safeguards.
        
               | unshavedyak wrote:
               | Is that different than what LLaMA would already give? I
               | suppose redistributing "harmful things" is still bad, but
               | if it's roughly equivalent to what's already out there i
               | struggle to think it's worth pulling.
               | 
               | Side question, how is this a surprise to them? If this
               | was due to safeguards, then pulling it now implies
               | there's some new form of information. What new
               | information could occur? That people were going to use it
               | to generate a bunch of harmful contents? Seems obvious..
               | wonder what we're missing
        
               | encoderer wrote:
               | So, to pull on that thread a little, it's only "unsafe"
               | for Stanford's reputation.
               | 
               | (And not for nothing, but their reputation is already
               | suffering badly)
        
           | safety1st wrote:
           | [flagged]
        
         | russellbeattie wrote:
         | Given that Alpaca violated the TOS of both services, this is
         | not surprising.
         | 
         | It could also have been Stanford's legal office trying to
         | preempt a lawsuit, or a "friendly" email from one of the
         | companies expressing displeasure and pointing out Stanford's
         | liability. So more of a veiled threat rather than an official
         | one.
         | 
         | Either way, the toothpaste is out of the tube. We now know that
         | a model's training can essentially be copied using the model
         | itself cheaply. Now that the team at Stanford showed it was
         | possible, and how relatively easy it was, it will bound to be
         | copied everywhere.
        
           | rdtsc wrote:
           | That could be a sneaky strategy by competitors -- make the
           | service say something naughty or illegal then call the media
           | with screenshots and act very offended by it.
        
             | syntaxfree wrote:
             | You can make any page say anything by messing in the
             | Developer pane.
        
       | wsgeorge wrote:
       | The last thing I did was try to get Alpaca to invent a new
       | language and say something in it. Midway through my experiment it
       | returned an error. Refreshing showed a server error page, and now
       | it displays this message [0]
       | 
       | It's been a fun web demo of a lightweight LLM doing amazing
       | stuff. :') Alpaca really moved the needle on democratizing the
       | recent advancements in LLMs [1]
       | 
       | [0] https://imgur.com/a/njKE1To
       | 
       | [1] https://simonwillison.net/2023/Mar/13/alpaca/
        
       | garbagecoder wrote:
       | I grew out of my cyberpunk phase in the early 90s, but I think
       | it's good that there are lots of leaks of all of this stuff. I
       | don't really have a problem with it being blocked from saying the
       | n-word or whatever and I'm sure that there is _always_ going to
       | be some tuning by the purveyor, but I feel like we have the right
       | to know more of the details with these than, say, how magic
       | select tools work in Photoshop
        
       | meowmeow20 wrote:
       | Thankfully we can run Alpaca locally now.
        
         | aftbit wrote:
         | Where do you get the weights?
        
           | craftyguy98 wrote:
           | 4chan!
        
           | gtoubassi wrote:
           | https://github.com/antimatter15/alpaca.cpp has links
        
             | cloudking wrote:
             | Interesting issue in that repo..
             | https://github.com/antimatter15/alpaca.cpp/issues/23
        
       | 2bitencryption wrote:
       | Is this an indication that the biggest impact from LLMs will be
       | on the edge?
       | 
       | It's almost a certainty that a model as good (or better) than
       | Alpaca's fine-tuned LLaMA 7B will be made public within the next
       | or two.
       | 
       | And it's been shown that a model of that size can run on a
       | Raspberry Pi with decent performance and accuracy.
       | 
       | With all that being the case, you could either use a service
       | (with restrictions, censorship, etc) or you could use your own
       | model locally (which may have a license that is essentially
       | "pretty please be good, we're not liable if you're bad").
       | 
       | For most use cases the service may provide better results. But if
       | self-hosting is only ~8months behind on average (guesstimate),
       | then why not just always self-host?
       | 
       | You could say "most users are not evil, and will be happy with a
       | service." Makes sense. But what about users who are privacy-
       | conscious, and don't want every query sent to a service?
        
         | eh9 wrote:
         | I just saw a project that lets you input an entire repo into
         | GPT. Coincidentally, my place of employment just told us not to
         | input any proprietary code into any generator with a retention
         | policy.
         | 
         | Even then, I feel like the play will be an enterprise service
         | instead of licensing.
        
       ___________________________________________________________________
       (page generated 2023-03-17 23:02 UTC)