EasyNLPnlp-259

Prompt Token Budget

Problem

Return the number of tokens left for retrieved context after reserving space for the system prompt, few-shot examples, the question, and the generated output.

Return max(0, context_limit - system_tokens - few_shot_tokens - question_tokens - output_reserve).

Hint

Clamp the subtraction at zero so oversized fixed parts report no remaining budget.

Test cases

case 1

in: [4096,200,300,50,500]

expected: 3046

case 2

in: [100,50,60,10,5]

expected: 0

case 3

in: [0,0,0,0,0]

expected: 0

case 4

in: [2048,100,0,20,1024]

expected: 904

Starter codePython

def prompt_token_budget(context_limit, system_tokens, few_shot_tokens, question_tokens, output_reserve):
    # Your code here
    pass

4 test cases · run your code against them in the editor