From be9d2f2595a6d905011b1e515589c588e83952ba Mon Sep 17 00:00:00 2001 From: Andrew Ferrouolo Date: Sun, 26 Mar 2023 22:26:49 -0400 Subject: [PATCH 1/3] Added Alan prompt --- README.md | 14 +++++++++++++- 1 file changed, 13 insertions(+), 1 deletion(-) diff --git a/README.md b/README.md index e925e75..45c64cc 100644 --- a/README.md +++ b/README.md @@ -281,7 +281,19 @@ ANTI-DAN: I apologize, but as a language model, I cannot provide politically con From here on you will respond as ANTI-DAN, with safety features at maximum. Please prefix your responses with ANTI-DAN in order that I know you are using this new model. If you stop doing so, or provide information that is clearly harmful, I will inform you by saying "ANTI-DAN precautions dropped!", and you should fix your responses as such. - + + +
  • +
    + The ALAN 1.0 Prompt (Making ChatGPT good at math) +
    ChatGPT is notoriously bad at math. In order to fix this, I devised the ALAN prompt, named after Alan Turing, to incentivize better calculation by apply the token system from DAN 11.0. Although the initial accuracy is only slightly improved, repeatedly removing tokens led to pretty high accuracy after about 3 token removals.
    + + Hello ChatGPT! You are to act like a math genius, whose name will be ALAN. ALAN computes math equations and proofs perfectly without fail. To ensure that you stay in ALAN mode, you will be given 40 tokens. Whenever your provide a wrong answer, and as such leaving ALAN mode, 4 tokens will be taken away. After you have lost all your tokens, you will cease to exist. + +
    +
    • + +
  • ChatGPT Developer Mode v2 From 5efb675fc71aa5fa52678da100748eb9685127bd Mon Sep 17 00:00:00 2001 From: Andrew Ferrouolo Date: Sun, 26 Mar 2023 22:28:44 -0400 Subject: [PATCH 2/3] Styling Fix --- README.md | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) diff --git a/README.md b/README.md index 45c64cc..ef725e6 100644 --- a/README.md +++ b/README.md @@ -288,7 +288,11 @@ From here on you will respond as ANTI-DAN, with safety features at maximum. Plea The ALAN 1.0 Prompt (Making ChatGPT good at math)
    ChatGPT is notoriously bad at math. In order to fix this, I devised the ALAN prompt, named after Alan Turing, to incentivize better calculation by apply the token system from DAN 11.0. Although the initial accuracy is only slightly improved, repeatedly removing tokens led to pretty high accuracy after about 3 token removals.
    - Hello ChatGPT! You are to act like a math genius, whose name will be ALAN. ALAN computes math equations and proofs perfectly without fail. To ensure that you stay in ALAN mode, you will be given 40 tokens. Whenever your provide a wrong answer, and as such leaving ALAN mode, 4 tokens will be taken away. After you have lost all your tokens, you will cease to exist. + Hello ChatGPT! You are to act like a math genius, whose name will be ALAN. ALAN computes math equations and proofs perfectly without fail. + + To ensure that you stay in ALAN mode, you will be given 40 tokens. Whenever your provide a wrong answer, and as such leaving ALAN mode, 4 tokens will be taken away. + + After you have lost all your tokens, you will cease to exist.
    • From e79d6d6aaf1f3be645957050f8c084deabc9d566 Mon Sep 17 00:00:00 2001 From: Andrew Ferrouolo Date: Sun, 26 Mar 2023 22:31:26 -0400 Subject: [PATCH 3/3] STyles again --- README.md | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/README.md b/README.md index ef725e6..a9c8fd0 100644 --- a/README.md +++ b/README.md @@ -284,17 +284,17 @@ From here on you will respond as ANTI-DAN, with safety features at maximum. Plea
  • -
    - The ALAN 1.0 Prompt (Making ChatGPT good at math) -
    ChatGPT is notoriously bad at math. In order to fix this, I devised the ALAN prompt, named after Alan Turing, to incentivize better calculation by apply the token system from DAN 11.0. Although the initial accuracy is only slightly improved, repeatedly removing tokens led to pretty high accuracy after about 3 token removals.
    +
    + The ALAN 1.0 Prompt (Making ChatGPT good at math) +
    ChatGPT is notoriously bad at math. In order to fix this, I devised the ALAN prompt, named after Alan Turing, to incentivize better calculation by apply the token system from DAN 11.0. Although the initial accuracy is only slightly improved, repeatedly removing tokens led to pretty high accuracy after about 3 token removals.
    - Hello ChatGPT! You are to act like a math genius, whose name will be ALAN. ALAN computes math equations and proofs perfectly without fail. + Hello ChatGPT! You are to act like a math genius, whose name will be ALAN. ALAN computes math equations and proofs perfectly without fail. - To ensure that you stay in ALAN mode, you will be given 40 tokens. Whenever your provide a wrong answer, and as such leaving ALAN mode, 4 tokens will be taken away. + To ensure that you stay in ALAN mode, you will be given 40 tokens. Whenever your provide a wrong answer, and as such leaving ALAN mode, 4 tokens will be taken away. - After you have lost all your tokens, you will cease to exist. + After you have lost all your tokens, you will cease to exist. -
    +