Dust - bcmejla

Dust implementation of ReAct prompt chaining https://arxiv.org/pdf/2210.03629.pdf

input

dataset:

Questions

data

dataset:

Examples

llm

model:

openai

text-davinci-002

temperature:

max tokens:

stop:

advanced

few-shot

introduction:

You are going to be presented with a set of multi-hop questions and need to suggest a first thought and action to get to the answer.

examples:

Question: ${EXAMPLES.question}
Thought 1:${EXAMPLES.thought_1}
Action 1:${EXAMPLES.action_1}

count:

prompt:

code

code :

code

code :

curl

https://

headers :

body :

code

code :

llm

model:

openai

text-davinci-002

temperature:

max tokens:

stop:

advanced

few-shot

introduction:

You are going to be presented with a set of multi-hop questions and need to suggest a next step or final answer.

examples:

Question: ${EXAMPLES.question}
Thought 1: ${EXAMPLES.thought_1}
Action 1: ${EXAMPLES.action_1}
Observation 1: ${EXAMPLES.obs_1}
Thought 2: ${EXAMPLES.thought_2}
Action 2: ${EXAMPLES.action_2}

count:

prompt: