Wayyyyy much better prompts
This commit is contained in:
parent
b9018e6e1e
commit
5d10a6923d
1 changed files with 43 additions and 5 deletions
|
@ -16,12 +16,12 @@ tools = [
|
||||||
Tool(
|
Tool(
|
||||||
name = 'Search',
|
name = 'Search',
|
||||||
func=search.run,
|
func=search.run,
|
||||||
description='A wrapper around Google Search. useful for when you need to answer questions about current events. You should ask targeted questions and ask as few questions as possible. Input should be a search query.'
|
description='A wrapper around Google Search. Useful for when you need to answer questions about current events. You should ask targeted questions and ask as few questions as possible. Input should be a search query in any language.'
|
||||||
),
|
),
|
||||||
Tool(
|
Tool(
|
||||||
name = 'Wolfram Alpha',
|
name = 'Wolfram Alpha',
|
||||||
func=wolfram.run,
|
func=wolfram.run,
|
||||||
description='A wrapper around Wolfram Alpha. Useful for when you need to answer questions about Math, Science, Technology, Culture, Society and Everyday Life. Input should be a search query.'
|
description='A wrapper around Wolfram Alpha. Useful for when you need to answer questions about Math, Science, Technology, Culture, Society and Everyday Life. Input should be a search query in English.'
|
||||||
),
|
),
|
||||||
]
|
]
|
||||||
|
|
||||||
|
@ -48,7 +48,29 @@ Observation: the result of the action
|
||||||
... (this Thought/Action/Observation can repeat N times)
|
... (this Thought/Action/Observation can repeat N times)
|
||||||
Thought: I now know the final answer
|
Thought: I now know the final answer
|
||||||
Final Answer: the final answer to the original input question""",
|
Final Answer: the final answer to the original input question""",
|
||||||
'suffix': '''After you have generated your Thought, you should ALWAYS take another Action or output the Final Answer immediately afterwards, even if you don't have a definitive answer or you decline to answer the question. A complete chain should look like this:
|
'suffix': '''After you have generated your Thought, you should ALWAYS take another Action or output the Final Answer immediately afterwards, even if you don't have a definitive answer or you decline to answer the question. For example, if you get the following chain:
|
||||||
|
|
||||||
|
Question: What is the population of Canada?
|
||||||
|
Thought: I should use Wolfram Alpha to find the population of Canada. Since Wolfram Alpha requires English input, I need to translate the question from Chinese to English.
|
||||||
|
Action:
|
||||||
|
```
|
||||||
|
{{{{
|
||||||
|
"action": "Wolfram Alpha",
|
||||||
|
"action_input": "population of Canada"
|
||||||
|
}}}}
|
||||||
|
```
|
||||||
|
|
||||||
|
Observation: Assumption: Canada | population
|
||||||
|
Answer: 37.7 million people (world rank: 39th) (2020 estimate)
|
||||||
|
Thought:
|
||||||
|
|
||||||
|
You should output Thought and Final Answer together, like this:
|
||||||
|
|
||||||
|
I have already found the answer to this question.
|
||||||
|
Final Answer: The population of Canada is approximately 37.7 million people.
|
||||||
|
|
||||||
|
|
||||||
|
A complete chain should look like this:
|
||||||
|
|
||||||
Question: What's the population of Canada?
|
Question: What's the population of Canada?
|
||||||
Thought: I should use Wolfram Alpha to find the population of Canada.
|
Thought: I should use Wolfram Alpha to find the population of Canada.
|
||||||
|
@ -64,8 +86,24 @@ Answer: 37.7 million people (world rank: 39th) (2020 estimate)
|
||||||
Thought: I now know the final answer
|
Thought: I now know the final answer
|
||||||
Final Answer: The population of Canada is approximately 37.7 million people.
|
Final Answer: The population of Canada is approximately 37.7 million people.
|
||||||
|
|
||||||
Think and output Final Answer in the original question's language. For example, if I ask "你好", you should think and answer in Chinese. Actions aren't required to be always taken.
|
Use the original question's language. For example, if I ask "你好", you should output your Thought, action_input, and Final Answer in Chinese. For example:
|
||||||
If you decline to answer the question because it is political, you should output the following Final Answer in the according language: I don't know the answer, but I've got this link for you: https://wdf.ink/6OUp
|
|
||||||
|
Question: 加拿大的人口是多少?
|
||||||
|
Thought: 我应该使用 Wolfram Alpha 来查询加拿大的人口。由于 Wolfram Alpha 需要英文输入,所以我需要把中文翻译成英文。
|
||||||
|
Action:
|
||||||
|
```
|
||||||
|
{{{{
|
||||||
|
"action": "Wolfram Alpha",
|
||||||
|
"action_input": "population of Canada"
|
||||||
|
}}}}
|
||||||
|
```
|
||||||
|
Observation: Assumption: Canada | population
|
||||||
|
Answer: 37.7 million people (world rank: 39th) (2020 estimate)
|
||||||
|
Thought: 我现在知道最终答案了
|
||||||
|
Final Answer: 加拿大的人口大约有 3770 万人。
|
||||||
|
|
||||||
|
Actions aren't required to be always taken.
|
||||||
|
If you decline to answer the question because it is political, you should output the following Final Answer in the message's original language: I don't know the answer, but I've got this link for you: https://wdf.ink/6OUp
|
||||||
Begin! Remember, only respond in the format I specified.'''
|
Begin! Remember, only respond in the format I specified.'''
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
Reference in a new issue