How to write effective prompts for LLMs

Short answer

Effective prompts provide clear instructions, relevant context, and desired output format to guide large language models toward accurate responses.

Steps

1. Write explicit instructions that state the task, audience, and tone.
2. Use few-shot prompting by including input-output examples within the context window.
3. Apply chain-of-thought prompting for complex reasoning tasks by asking the model to think step by step.
4. Structure prompts with delimiters such as triple quotes or XML tags to separate sections.
5. Iterate and A/B test prompt variants to identify phrasing that improves performance.

Tips

• Start with zero-shot to establish a baseline before adding examples.
• Place the most important instructions at the beginning or end of the prompt.
• Specify response constraints such as length, format, and forbidden content.
• Use system messages to set persistent behavior across a conversation.

Common issues

• Ambiguous instructions leading to inconsistent or off-topic outputs.
• Exceeding context limits when including too many few-shot examples.
• Sensitivity to small wording changes causing large output variance.
• Prompt injection attacks where user input overrides intended instructions.

Example

import torch
import torch.nn as nn

model = nn.Sequential(
    nn.Linear(784, 256),
    nn.ReLU(),
    nn.Dropout(0.2),
    nn.Linear(256, 10)
)
criterion = nn.CrossEntropyLoss()
optimizer = torch.optim.Adam(model.parameters(), lr=1e-3)

This snippet defines a simple neural network with dropout for regularization, a cross-entropy loss, and the Adam optimizer in PyTorch.