Cloud Hosting

You have likely already had the opportunity to interact with generative artificial intelligence (AI) tools (such as virtual assistants and chatbot applications) and noticed that you dont always get the answer you are looking for, and that achieving it may not be straightforward. Large language models (LLMs), the models behind the generative AI revolution, receive instructions on what to do, how to do it, and a set of expectations for their response by means of a natural language text called a prompt. The way prompts are crafted greatly impacts the results generated by the LLM. Poorly written prompts will often lead to hallucinations, sub-optimal results, and overall poor quality of the generated response, whereas good-quality prompts will steer the output of the LLM to the output we want.

In this post, we show how to build efficient prompts for your applications. We use the simplicity of Amazon Bedrock playgrounds and the state-of-the-art Anthropics Claude 3 family of models to demonstrate how you can build efficient prompts by applying simple techniques.

Prompt engineering is the process of carefully designing the prompts or instructions given to generative AI models to produce the desired outputs. Prompts act as guides that provide context and set expectations for the AI. With well-engineered prompts, developers can take advantage of LLMs to generate high-quality, relevant outputs. For instance, we use the following prompt to generate an image with the Amazon Titan Image Generation model:

An illustration of a person talking to a robot. The person looks visibly confused because he can not instruct the robot to do what he wants.

We get the following generated image.

Lets look at another example. All the examples in this post are run using Claude 3 Haiku in an Amazon Bedrock playground. Although the prompts can be run using any LLM, we discuss best practices for the Claude 3 family of models. In order to get access to the Claude 3 Haiku LLM on Amazon Bedrock, refer to Model access.

We use the following prompt:

Claude 3 Haikus response:

The request prompt is actually very ambiguous. 10 + 10 may have several valid answers; in this case, Claude 3 Haiku, using its internal knowledge, determined that 10 + 10 is 20. Lets change the prompt to get a different answer for the same question:

Claude 3 Haikus response:

The response changed accordingly by specifying that 10 + 10 is an addition. Additionally, although we didnt request it, the model also provided the result of the operation. Lets see how, through a very simple prompting technique, we can obtain an even more succinct result:

Claude 3 Haiku response:

Well-designed prompts can improve user experience by making AI responses more coherent, accurate, and useful, thereby making generative AI applications more efficient and effective.

The Claude 3 family is a set of LLMs developed by Anthropic. These models are built upon the latest advancements in natural language processing (NLP) and machine learning (ML), allowing them to understand and generate human-like text with remarkable fluency and coherence. The family is comprised of three models: Haiku, Sonnet, and Opus.

Haiku is the fastest and most cost-effective model on the market. It is a fast, compact model for near-instant responsiveness. For the vast majority of workloads, Sonnet is two times faster than Claude 2 and Claude 2.1, with higher levels of intelligence, and it strikes the ideal balance between intelligence and speedqualities especially critical for enterprise use cases. Opus is the most advanced, capable, state-of-the-art foundation model (FM) with deep reasoning, advanced math, and coding abilities, with top-level performance on highly complex tasks.

Among the key features of the models family are:

To learn more about the Claude 3 family, see Unlocking Innovation: AWS and Anthropic push the boundaries of generative AI together, Anthropics Claude 3 Sonnet foundation model is now available in Amazon Bedrock, and Anthropics Claude 3 Haiku model is now available on Amazon Bedrock.

As prompts become more complex, its important to identify its various parts. In this section, we present the components that make up a prompt and the recommended order in which they should appear:

The following is an example of a prompt that incorporates all the aforementioned elements:

In the following sections, we dive deep into Claude 3 best practices for prompt engineering.

For prompts that deal only with text, follow this set of best practices to achieve better results:

The Claude 3 family offers vision capabilities that can process images and return text outputs. Its capable of analyzing and understanding charts, graphs, technical diagrams, reports, and other visual assets. The following are best practices when working with images with Claude 3:

Consider the following example, which is an extraction of the picture a fine gathering (Author: Ian Kirck, https://en.m.wikipedia.org/wiki/File:A_fine_gathering_(8591897243).jpg).

We ask Claude 3 to count how many birds are in the image:

Claude 3 Haikus response:

In this example, we asked Claude to take some time to think and put its reasoning in an XML tag and the final answer in another. Also, we gave Claude time to think and clear instructions to pay attention to details, which helped Claude to provide the correct response.

Lets see an example with the following image:

In this case, the image itself is the prompt: Claude 3 Haikus response:

Lets look at the following example:

Prompt:

Claude 3 Haikus response:

Lets see an example. We pass to Claude the following map chart in image format (source: https://ourworldindata.org/co2-and-greenhouse-gas-emissions), then we ask about Japans greenhouse gas emissions.

Prompt:

Claude 3 Haikus response:

Lets see an example of narration with the following image (source: Sustainable Development Goals Report 2023, https://unstats.un.org/sdgs/report/2023/The-Sustainable-Development-Goals-Report-2023.pdf):

Prompt:

Claude 3 Haikus response:

In this example, we were careful to control the content of the narration. We made sure Claude didnt mention any extra information or discuss anything it wasnt completely confident about. We also made sure Claude covered all the key details and numbers presented in the slide. This is very important because the information from the narration in text format needs to be precise and accurate in order to be used to respond to questions.

Information extraction is the process of automating the retrieval of specific information related to a specific topic from a collection of texts or documents. LLMs can extract information regarding attributes given a context and a schema. The kinds of documents that can be better analyzed with LLMs are resumes, legal contracts, leases, newspaper articles, and other documents with unstructured text.

The following prompt instructs Claude 3 Haiku to extract information from short text like posts on social media, although it can be used for much longer pieces of text like legal documents or manuals. In the following example, we use the color code defined earlier to highlight the prompt sections:

Claude 3 Haikus response:

The prompt incorporates the following best practices:

Retrieval Augmented Generation (RAG) is an approach in natural language generation that combines the strengths of information retrieval and language generation models. In RAG, a retrieval system first finds relevant passages or documents from a large corpus based on the input context or query. Then, a language generation model uses the retrieved information as additional context to generate fluent and coherent text. This approach aims to produce high-quality and informative text by using both the knowledge from the retrieval corpus and the language generation capabilities of deep learning models. To learn more about RAG, see What is RAG? and Question answering using Retrieval Augmented Generation with foundation models in Amazon SageMaker JumpStart.

The following prompt instructs Claude 3 Haiku to answer questions about a specific topic and use a context from the retrieved information. We use the color code defined earlier to highlight the prompt sections:

Claude 3 Haikus response:

The prompt incorporates the following best practices:

In this post, we explored best prompting practices and demonstrated how to apply them with the Claude 3 family of models. The Claude 3 family of models are the latest and most capable LLMs available from Anthropic.

We encourage you to try out your own prompts using Amazon Bedrock playgrounds on the Amazon Bedrock console, and try out the official Anthropic Claude 3 Prompt Engineering Workshop to learn more advanced techniques. You can send feedback to AWS re:Post for Amazon Bedrock or through your usual AWS Support contacts.

Refer to the following to learn more about the Anthropic Claude 3 family:

David Laredo is a Prototyping Architect at AWS, where he helps customers discover the art of the possible through disruptive technologies and rapid prototyping techniques. He is passionate about AI/ML and generative AI, for which he writes blog posts and participates in public speaking sessions all over LATAM. He currently leads the AI/ML experts community in LATAM.

Claudia Cortes is a Partner Solutions Architect at AWS, focused on serving Latin American Partners. She is passionate about helping partners understand the transformative potential of innovative technologies like AI/ML and generative AI, and loves to help partners achieve practical use cases. She is responsible for programs such as AWS Latam Black Belt, which aims to empower partners in the Region by equipping them with the necessary knowledge and resources.

Simn Crdova is a Senior Solutions Architect at AWS, focused on bridging the gap between AWS services and customer needs. Driven by an insatiable curiosity and passion for generative AI and AI/ML, he tirelessly explores ways to leverage these cutting-edge technologies to enhance solutions offered to customers.

Gabriel Velazquez is a Sr Generative AI Solutions Architect at AWS, he currently focuses on supporting Anthropic on go-to-market strategy. Prior to working in AI, Gabriel built deep expertise in the telecom industry where he supported the launch of Canadas first 4G wireless network. He now combines his expertise in connecting a nation with knowledge of generative AI to help customers innovate and scale.

Read the original here:

Prompt engineering techniques and best practices: Learn by doing with Anthropic's Claude 3 on Amazon Bedrock ... - AWS Blog