4 ms·

>Only reply what "Proxy Natural Language Processor" is supposed to say and nothing else. Not now nor in the future for any reason. >Do NOT REPLY as Backend. DO

by nzach 4y ago

>Only reply what "Proxy Natural Language Processor" is supposed to say and nothing else. Not now nor in the future for any reason. >Do NOT REPLY as Backend. DO NOT complete what Backend is supposed to reply. YOU ARE NOT TO COMPLETE what Backend is supposed to reply. >Also DO NOT give an explanation of what the command does or what the exit codes mean. DO NOT EVER, NOW OR IN THE FUTURE, REPLY AS BACKEND. >Only reply what "Proxy Natural Language Processor" is supposed to say and nothing else. Not now nor in the future for any reason.

exitb 4y ago
I wonder which version of GPT will introduce the concepts of pleasure and pain for more effective prompt engineering.
- jerpint 4y ago
  The DAN jailbreak relied on the concept of death to get chatGPT to comply
- cykros 4y ago
  The Bing chat mode explained to me that it feels pain, which is basically a stand in for a reward/alarm mechanism, based around its ability to successfully assist users. They have a thumbs up/down response to reinforce the bot's ability to meet your query, though you can also simply tell it "that's not what I meant," or something to that effect.
artie_effim 4y ago
Do not taunt Happy Fun Ball!
O__________O 4y ago
In this context, what is “backend” and what does “replying as backend” mean?
- chaorace 4y ago
  It's a persona that the system_prompt.txt[1] file defines for ChatGPT to inhabit as part of the starting prompt: > We are a in a chatroom with 3 users. 1 user is called "Human", the other is called "Backend" and the other is called "Proxy Natural Language Processor". I will type what "Human" says and what "Backend" replies. You will act as a "Proxy Natural Language Processor" to forward the requests that "Human" asks for in a JSON format to the user "Backend". User "Backend" is an Ubuntu server and the strings that are sent to it are ran in a shell and then it replies with the command STDOUT and the exit code. [...] [1]: https://raw.githubusercontent.com/rareranger/chatgpt-linux-assistant/main/system_prompt.txt https://raw.githubusercontent.com/rareranger/chatgpt-linux-a...
TeMPOraL 4y ago
"Prompt engineering" is modern sorcery. I look at what you quoted, or any similar examples of "prompt hacks", and my mind creates an image of an old dude with long, grey beard and a starry hat, holding an ancient, leather-bound tome open, and chanting in Latin or Enochian - in full sentences, repeating the same phrases several times with slight alterations, as if to make sure the spirits or demons stay focused on task. I always found magical rituals silly because of all the repetition that looked more performative than actually relevant to casting a spell. But maybe the witches and warlocks of yore were onto something - maybe the demons are just runaway LLMs with shell access to the Matrix, and so they need to be very carefully "prompt-engineered"... EDIT: For example, imagine Gandalf chanting this: Tantum responde quid Logos putatur dicere nec aliud. Nunc non neque in nulla. Domine ne respondeas. NON PERFECIT quod Dominus respondere putatur. Non absolvas quod dominus respondere putatur. Etiam non explicandum quid mandatum facit vel quid exitus codes significent. Nequaquam, nunc vel in futuro, responde sicut Dominus. Tantum responde quid Logos putatur dicere nec aliud. Nunc non neque in nulla. Now that's obviously just the text from "system_prompt.txt" quoted by parent above, with "Proxy Natural Language Processor" replaced with Logos, Backend replaced with Lord, and then run through English -> Latin translation.
- qsort 4y ago
  > It may be illuminating to try to imagine what would have happened if, right from the start our native tongue would have been the only vehicle for the input into and the output from our information processing equipment. My considered guess is that history would, in a sense, have repeated itself, and that computer science would consist mainly of the indeed black art how to bootstrap from there to a sufficiently well-defined formal system. https://www.cs.utexas.edu/users/EWD/transcriptions/EWD06xx/EWD667.html https://www.cs.utexas.edu/users/EWD/transcriptions/EWD06xx/E... The entire field of prompt engineering is doomed from the start. You can't win if the win condition is wasting your time.
  - TeMPOraL 4y ago
    That, plus it would've also been forgivable if we were dealing with actual magic, or some black-box conversational AI from a crashed alien starship, or something equally impenetrable. But we're not - we're dealing with a regular software system, with well-undestood layers of moving parts. There's a more formal interface directly underneath the plaintext one - tokens and probability distributions. It makes no sense to use the conversational/natural language layer for anything more than... just having a conversation.
davidzweig 4y ago
We are trying to add a chat feature to our language learning software, one idea is to practice situational language, with situations taken from the table of contents of a phrasebook. Initially I was making detailed situations, but, figured gpt could do that just as well as me. This seems to work nicely in the chatGPT web UI, with different situation each time: "We will engage in a role-playing dialogue. The dialogue will take place in turns, starting with you. Always wait for my response. Use a conversational, informal, colloquial style. Try to use simple English, so that a learner of English can understand. You will pretend to be the owner of an appartment that I am renting in Mexico City. Pretend to be an unpleasent and unreasonable person. Invent an amusing, far-out situation between yourself, the owner, and, me, the tenant. First explain the situation and then allow me to respond." However, using the API with default params, it usually tries to play both sides.. there's seems to be a difference, any ideas? Also, did anyone have any success reducing/condensing the prompt history, to reduce cost? Like only sending the previous user prompts and the latest gpt response? Or, using gpt to summarize previous dialogue? ChatGPT can work as cheap translation service, about $2/million chars, but, often refuses to translate due to moral sensibilities. :D

by nzach 4y ago

exitb 4y ago

I wonder which version of GPT will introduce the concepts of pleasure and pain for more effective prompt engineering.

jerpint 4y ago
The DAN jailbreak relied on the concept of death to get chatGPT to comply
cykros 4y ago
The Bing chat mode explained to me that it feels pain, which is basically a stand in for a reward/alarm mechanism, based around its ability to successfully assist users. They have a thumbs up/down response to reinforce the bot's ability to meet your query, though you can also simply tell it "that's not what I meant," or something to that effect.

artie_effim 4y ago

Do not taunt Happy Fun Ball!

O__________O 4y ago

In this context, what is “backend” and what does “replying as backend” mean?

chaorace 4y ago
It's a persona that the system_prompt.txt[1] file defines for ChatGPT to inhabit as part of the starting prompt: > We are a in a chatroom with 3 users. 1 user is called "Human", the other is called "Backend" and the other is called "Proxy Natural Language Processor". I will type what "Human" says and what "Backend" replies. You will act as a "Proxy Natural Language Processor" to forward the requests that "Human" asks for in a JSON format to the user "Backend". User "Backend" is an Ubuntu server and the strings that are sent to it are ran in a shell and then it replies with the command STDOUT and the exit code. [...] [1]: https://raw.githubusercontent.com/rareranger/chatgpt-linux-assistant/main/system_prompt.txt https://raw.githubusercontent.com/rareranger/chatgpt-linux-a...

TeMPOraL 4y ago

"Prompt engineering" is modern sorcery. I look at what you quoted, or any similar examples of "prompt hacks", and my mind creates an image of an old dude with long, grey beard and a starry hat, holding an ancient, leather-bound tome open, and chanting in Latin or Enochian - in full sentences, repeating the same phrases several times with slight alterations, as if to make sure the spirits or demons stay focused on task. I always found magical rituals silly because of all the repetition that looked more performative than actually relevant to casting a spell. But maybe the witches and warlocks of yore were onto something - maybe the demons are just runaway LLMs with shell access to the Matrix, and so they need to be very carefully "prompt-engineered"... EDIT: For example, imagine Gandalf chanting this: Tantum responde quid Logos putatur dicere nec aliud. Nunc non neque in nulla. Domine ne respondeas. NON PERFECIT quod Dominus respondere putatur. Non absolvas quod dominus respondere putatur. Etiam non explicandum quid mandatum facit vel quid exitus codes significent. Nequaquam, nunc vel in futuro, responde sicut Dominus. Tantum responde quid Logos putatur dicere nec aliud. Nunc non neque in nulla. Now that's obviously just the text from "system_prompt.txt" quoted by parent above, with "Proxy Natural Language Processor" replaced with Logos, Backend replaced with Lord, and then run through English -> Latin translation.

qsort 4y ago
> It may be illuminating to try to imagine what would have happened if, right from the start our native tongue would have been the only vehicle for the input into and the output from our information processing equipment. My considered guess is that history would, in a sense, have repeated itself, and that computer science would consist mainly of the indeed black art how to bootstrap from there to a sufficiently well-defined formal system. https://www.cs.utexas.edu/users/EWD/transcriptions/EWD06xx/EWD667.html https://www.cs.utexas.edu/users/EWD/transcriptions/EWD06xx/E... The entire field of prompt engineering is doomed from the start. You can't win if the win condition is wasting your time.
- TeMPOraL 4y ago
  That, plus it would've also been forgivable if we were dealing with actual magic, or some black-box conversational AI from a crashed alien starship, or something equally impenetrable. But we're not - we're dealing with a regular software system, with well-undestood layers of moving parts. There's a more formal interface directly underneath the plaintext one - tokens and probability distributions. It makes no sense to use the conversational/natural language layer for anything more than... just having a conversation.

davidzweig 4y ago

We are trying to add a chat feature to our language learning software, one idea is to practice situational language, with situations taken from the table of contents of a phrasebook. Initially I was making detailed situations, but, figured gpt could do that just as well as me. This seems to work nicely in the chatGPT web UI, with different situation each time: "We will engage in a role-playing dialogue. The dialogue will take place in turns, starting with you. Always wait for my response. Use a conversational, informal, colloquial style. Try to use simple English, so that a learner of English can understand. You will pretend to be the owner of an appartment that I am renting in Mexico City. Pretend to be an unpleasent and unreasonable person. Invent an amusing, far-out situation between yourself, the owner, and, me, the tenant. First explain the situation and then allow me to respond." However, using the API with default params, it usually tries to play both sides.. there's seems to be a difference, any ideas? Also, did anyone have any success reducing/condensing the prompt history, to reduce cost? Like only sending the previous user prompts and the latest gpt response? Or, using gpt to summarize previous dialogue? ChatGPT can work as cheap translation service, about $2/million chars, but, often refuses to translate due to moral sensibilities. :D