🧪 test: level 2

This commit is contained in:
Florian Hönicke
2023-04-30 22:39:50 +02:00
parent 158650d2d9
commit 4c33a3f298

View File

@@ -62,7 +62,7 @@ template_generate_possible_packages = PromptTemplate.from_template(
"{description}"
1. Write down the different strategies to solve the task. For each strategy write down all the non-trivial subtasks you need to solve. If there is a natural language understanding or generation stragegy, write it down.
2. Find out what is the core problem to solve.
3. List up to 15 Python packages that are specifically designed or have functionalities to solve the complete core problem with one of the defined strategies. You must add gpt_3_5_turbo_api if the task involves generating or understanding natural language or using a (pre-trained) language model.
3. List up to 10 Python packages that are specifically designed or have functionalities to solve the complete core problem with one of the defined strategies. You must add gpt_3_5_turbo_api if the task involves generating or understanding natural language or using a (pre-trained) language model.
4. Exclude any package that can generate or understand natural language or enables using any language model, but you must not exclude gpt_3_5_turbo_api. Print the cleaned list of packages and give a brief reason for keeping it after its name.
5. For each cleaned package think if it fulfills the following requirements:
a) specifically designed or have functionalities to solve the complete core problem.