what you want from the model will make
For example, in your model that you have trained to solve mathematical problems, if the outputs in the 5 examples given are numerical and 1 of them is determined as 'could not be solved', this may make it difficult for the model to give correct outputs. Sample Consistency Create a consistent format structure across your samples. Giving examples in different formats will make it difficult for the model to directly give the desired result in its answers.For example, in your Few-Shot prompt you created to solve mathematical problems; In the first example, you give '2 plus 2' as input and '4' as output, In the second example, let's assume that you give '3 plus 2' as input and 'This ques Job Seekers Phone Numbers List tion is an addition operation and the answer is 5' as output. In the next question, when you want to get the answer directly as in the first example, the model may try to imitate the output sample of the second question.
https://www.amerdata.com/wp-content/uploads/2024/03/Add-a-heading-8-300x300.png
Task Description Before giving examples in your prompt, clearly stating it easier for you to get effective results. For example, in your Few-Shot prompt that you will create to solve mathematical problems; Giving the model the task of “You are a mathematics professor who can solve challenging problems” will increase the effectiveness of the model in understanding the examples given and getting the output you want. Sample Diversity In order for the model to give better generalization of its output, it will be effective to increase the examples without breaking them in context.
頁:
[1]