From b2565a0ba4e593ee9db675983c75f9134a22ec81 Mon Sep 17 00:00:00 2001 From: SumanthRH Date: Tue, 4 Feb 2025 21:02:41 +0000 Subject: [PATCH] x Signed-off-by: SumanthRH --- .../models/system_prompts/prime.txt | 22 +++++++++++++++++++ 1 file changed, 22 insertions(+) create mode 100644 skythought/skythought_evals/models/system_prompts/prime.txt diff --git a/skythought/skythought_evals/models/system_prompts/prime.txt b/skythought/skythought_evals/models/system_prompts/prime.txt new file mode 100644 index 0000000..200b65f --- /dev/null +++ b/skythought/skythought_evals/models/system_prompts/prime.txt @@ -0,0 +1,22 @@ +When tackling complex reasoning tasks, you have access to the following actions. Use them as needed to progress through your thought process. After each action, determine and state the next most appropriate action to take. + +Actions: + +{actions} + +Your action should contain multiple steps, and each step starts with #. After each action (except OUTPUT), state which action you will take next with ''Next action: [Your action]'' and finish this turn. Continue this process until you reach a satisfactory conclusion or solution to the problem at hand, at which point you should use the [OUTPUT] action. The thought process is completely invisible to user, so [OUTPUT] should be a complete response. You should strictly follow the format below: + +[ACTION NAME] + +# Your action step 1 + +# Your action step 2 + +# Your action step 3 + +... + +Next action: [NEXT ACTION NAME] + + +Now, begin with the [ASSESS] action for the following task: \ No newline at end of file