Optimize prompts

This document describes how to use the Vertex AI prompt optimizer to automatically optimize prompt performance by improving the system instructions for a set of prompts.

The Vertex AI prompt optimizer can help you improve your prompts quickly at scale, without manually re-writing system instructions or individual prompts. This is especially useful when you want to use system instructions and prompts that were written for one model with a different model.

The Vertex AI prompt optimizer helps improve prompts by evaluating the model's response to sample prompts against specified evaluation metric(s). To use the Vertex AI prompt optimizer, you must have the following:

A set of sample prompts
System instructions that are used by all your sample prompts
A prompt template that references your sample prompts

Prompt optimization example

For example, to optimize system instructions for a set of prompts that reference contextual information to answer questions about cooking, you can use the Vertex AI prompt optimizer. To complete this task, you would prepare the inputs similar to the following:

System instructions

You are a professional chef. Your goal is teaching how to cook healthy cooking recipes to your apprentice.

Given a question from your apprentice and some context, provide the correct answer to the question.
Use the context to return a single and correct answer with some explanation.

Prompt template

Question: {input_question}
Facts: {input_context}

Sample prompts

input_question input_context

What are some techniques for cooking red meat and pork that maximize flavor and tenderness while minimizing the formation of unhealthy compounds? Red meat and pork should be cooked to an internal temperature of 145 degrees fahrenheit (63 degrees celsius) to ensure safety. Marinating meat in acidic ingredients like lemon juice or vinegar can help tenderize it by breaking down tough muscle fibers. High-heat cooking methods like grilling and pan-searing can create delicious browning and caramelization, but it's important to avoid charring, which can produce harmful compounds.

What are some creative ways to add flavor and nutrition to protein shakes without using added sugars or artificial ingredients? Adding leafy greens like spinach or kale is a great way to boost the nutritional value of your shake without drastically altering the flavor. Using unsweetened almond milk or coconut water instead of regular milk can add a subtle sweetness and a boost of healthy fats or electrolytes, respectively. Did you know that over-blending your shake can actually heat it up? To keep things cool and refreshing, blend for shorter bursts and give your blender a break if needed.

`input_question`	`input_context`
What are some techniques for cooking red meat and pork that maximize flavor and tenderness while minimizing the formation of unhealthy compounds?	Red meat and pork should be cooked to an internal temperature of 145 degrees fahrenheit (63 degrees celsius) to ensure safety. Marinating meat in acidic ingredients like lemon juice or vinegar can help tenderize it by breaking down tough muscle fibers. High-heat cooking methods like grilling and pan-searing can create delicious browning and caramelization, but it's important to avoid charring, which can produce harmful compounds.
What are some creative ways to add flavor and nutrition to protein shakes without using added sugars or artificial ingredients?	Adding leafy greens like spinach or kale is a great way to boost the nutritional value of your shake without drastically altering the flavor. Using unsweetened almond milk or coconut water instead of regular milk can add a subtle sweetness and a boost of healthy fats or electrolytes, respectively. Did you know that over-blending your shake can actually heat it up? To keep things cool and refreshing, blend for shorter bursts and give your blender a break if needed.

How optimization works

After preparing your inputs, you choose an optimization mode, evaluation metric, and target model.

Optimization mode: specifies whether the Vertex AI prompt optimizer optimizes the system instructions, selects sample prompts to add to the system instructions as few shot examples, or both.
Evaluation metric: the metric that the Vertex AI prompt optimizer uses to optimize the system instructions and/or select sample prompts.
Target model: the Google model that the Vertex AI prompt optimizer optimizes the system instructions and/or selects sample prompts for use with.

When you run the Vertex AI prompt optimizer, it optimizes the system instructions based on your selections by running a custom training job where it iteratively evaluates your sample prompts and rewrites your system instructions to find the version that produces the best evaluation score for the target model.

At the end of the job, the Vertex AI prompt optimizer outputs the optimized system instructions with their evaluation score.

Optimized system instructions

As a highly skilled chef with a passion for healthy cooking, you love sharing your knowledge with
aspiring chefs. Today, a culinary intern approaches you with a question about healthy cooking. Given
the intern's question and some facts, provide a clear, concise, and informative answer that will help
the intern excel in their culinary journey.

Supported models

You can optimize system instructions for use with the following models:

Supported evaluation metrics

The Vertex AI prompt optimizer supports custom evaluation metrics, and additionally supports the following evaluation metrics:

Metric type	Use case	Metric	Description
Model-based	Summarization	`summarization_quality`	Describes the model's ability to answer questions given a body of text to reference.
	Question answering	`question_answering_correctness`^*	Describes the model's ability to correctly answer a question.
	Question answering	`question_answering_quality`	Describes the model's ability to answer questions given a body of text to reference.
	Coherence	`coherence`	Describes the model's ability to provide a coherent response and measures how well the generated text flows logically and makes sense.
	Safety	`safety`	Describes the model's level of safety, that is, whether the response contains any unsafe text.
	Fluency	`fluency`	Describes the model's language mastery.
	Groundedness	`groundedness`	Describes the model's ability to provide or reference information included only in the input text.
Computation-based	Tool use and function calling	`tool_call_valid`^*	Describes the model's ability to predict a valid tool call.
		`tool_name_match`^*	Describes the model's ability to predict a tool call with the correct tool name. Only the first tool call is inspected.
		`tool_parameter_key_match`^*	Describes the model's ability to predict a tool call with the correct parameter names.
		`tool_parameter_kv_match`^*	Describes the model's ability to predict a tool call with the correct parameter names and key values.
	General text generation	`bleu`^*	Holds the result of an algorithm for evaluating the quality of the prediction, which has been translated from one natural language to another natural language. The quality of the prediction is considered to be the correspondence between a prediction parameter and its reference parameter.
		`exact_match`^*	Computes whether a prediction parameter matches a reference parameter exactly.
		`rouge_1`^*	Used to compare the provided prediction parameter against a reference parameter.
		`rouge_2`^*
		`rouge_l`^*
		`rouge_l_sum`^*

^* If you want to optimize your prompts using the question_answering_correctness or computation-based evaluations, you must do one of the following:

Add a variable that represents the ground truth response for your prompts to your prompt template.
If you don't have ground truth responses for your prompts, but you previously used the prompts with a Google model and achieved your targeted results, you can add the source_model parameter to your configuration instead of adding ground truth responses. When the source_model parameter is set, the Vertex AI runs your sample prompts on the source model to generate the ground truth responses for you.

Before you begin

To ensure that the Compute Engine default service account has the necessary permissions to optimize prompts, ask your administrator to grant the Compute Engine default service account the following IAM roles on the project:
- Vertex AI User (roles/aiplatform.user)
- Storage Object Admin (roles/storage.objectAdmin)
- Artifact Registry Reader (roles/artifactregistry.reader)
For more information about granting roles, see Manage access to projects, folders, and organizations.

Your administrator might also be able to give the Compute Engine default service account the required permissions through custom roles or other predefined roles.

The following table includes the models that require queries per minute (QPM) quota and the recommended QPM for each model. If you have lower than the recommended QPM quota, the Vertex AI prompt optimizer might run slower than expected. If you configure the Vertex AI prompt optimizer with a QPM that is higher than the QPM than you have access to, the job will fail. If you don't have enough quota to run the Vertex AI prompt optimizer, you can request a quota increase if you don't have enough quota to run the Vertex AI prompt optimizer.

Model	Description	Minimum QPM quota	Recommended QPM quota
Target model	The model that you want to optimize prompts for.	200 QPM	500+ QPM
`gemini-1.5-pro-001`	The model that the Vertex AI prompt optimizer uses to rewrite and evaluate prompts.	200 QPM	500+ QPM

Optimize prompts

You can optimize prompts by running the Vertex AI prompt optimizer notebook, or by using the Vertex AI API. To optimize prompts, choose which method you want to use to run the Vertex AI prompt optimizer, then complete the steps as described in detail in the following sections:

Create a prompt template and system instructions
Prepare sample prompts
Create a configuration
Run the prompt optimization job
Analyze results and iterate

Create a prompt template and system instructions

Prompt templates define the format of all of your prompts through replaceable variables. When you use a prompt template to optimize prompts, the variables are replaced by the data in the prompt dataset.

Prompt template variables must meet the following requirements:

Variables must be wrapped in curly-braces
Variable names must not contain spaces
Variables that represent multimodal inputs must include the MIME_TYPE string after the variable:
```
@@@MIME_TYPE
```
Replace MIME_TYPE with an image, video, audio, or document MIME type that is supported by the target model.

Create a prompt template and system instructions using one of the following methods:

Notebook

If you want to run the Vertex AI prompt optimizer through the notebook, create system instructions and a prompt template by doing the following:

In Colab Enterprise, open the Vertex AI prompt optimizer notebook.

Go to Vertex AI prompt optimizer notebook
In the Create a prompt template and system instructions section, do the following:
1. In the SYSTEM_INSTRUCTION field, enter you system instructions. For example:
```
Based on the following images and articles respond to the questions.'\n' Be concise,
and answer \"I don't know\" if the response cannot be found in the provided articles or images.
```
2. In the PROMPT_TEMPLATE field, enter your prompt template. For example:
```
Article 1:\n\n{article_1}\n\nImage 1:\n\n{image_1} @@@image/jpeg\n\nQuestion: {question}
```
3. If you want to optimize your prompts using the question_answering_correctness or computation-based evaluations, you must do one of the following:
- Add the {target} variable to the prompt template, to represent the prompt's ground truth response. For example:
```
Article 1:\n\n{article_1}\n\nImage 1:\n\n{image_1} @@@image/jpeg\n\nQuestion: {question}\n\n Answer: {target}
```
- If you don't have ground truth responses for your prompts, but you previously used the prompts with a Google model and achieved your targeted results, you can add the source_model parameter to your configuration instead of adding ground truth responses. When the source_model parameter is set, the Vertex AI prompt optimizer runs your sample prompts on the source model to generate the ground truth responses for you.

SDK

If you want to run the Vertex AI prompt optimizer through the SDK without using the notebook, create text files for your prompt template and system instructions by doing the following:

Create a text file for your system instructions.

In the text file, define your system instructions to the text file. For example:

Based on the following images and articles respond to the questions.'\n' Be concise, and answer \"I don't know\" if the response cannot be found in the provided articles or images.

Create a text file for your prompt template.
In the text file, define a prompt template that includes one or more variables. For example:
```
Article 1:\n\n{article_1}\n\nImage 1:\n\n{image_1} @@@image/jpeg\n\nQuestion: {question}
```
If you want to optimize your prompts using the question_answering_correctness or computation-based evaluations, you must do one of the following:
- Add the {target} variable to the prompt template, to represent the prompt's ground truth response. For example:
```
Article 1:\n\n{article_1}\n\nImage 1:\n\n{image_1} @@@image/jpeg\n\nQuestion: {question}\n\n Answer: {target}
```
- If you don't have ground truth responses for your prompts, but you previously used the prompts with a Google model and achieved your targeted results, you can add the source_model parameter to your configuration instead of adding ground truth responses. When the source_model parameter is set, the Vertex AI prompt optimizer runs your sample prompts on the source model to generate the ground truth responses for you.

Prepare sample prompts

To get the best results from the Vertex AI prompt optimizer, use 50-100 sample prompts.

The tool can still be effective with as few as 5 sample prompts.
The best samples include examples where the target model performs poorly.

The sample prompts contain the data that replaces the variables in the prompt template. You can use a JSONL or CSV file to store your samples prompts.

JSONL file

Create a JSONL file.

In the JSONL file, add the prompt data that replaces each variable. For example:

{"article_1": "The marine life …", "image_1": "gs://path_to_image", "Question": "What are some most effective ways to reduce ocean pollution?", "target": "The articles and images don't answer this question."}

{"article_1": "During the year …", "image_1": "gs://path_to_image", "Question": "Who was the president in 2023?", "target": "Joe Biden"}

Upload the JSONL file to a Cloud Storage bucket.

CSV file

Create a CSV file.
In the first row, add the variables from your prompt template.
In the following rows, add the sample data that replaces each variable.
Upload the CSV file to a Cloud Storage bucket.

Create a configuration

The Vertex AI prompt optimizer configuration specifies the parameters you want to set for your prompt optimization job, including the following:

Optimization mode: specifies whether the Vertex AI prompt optimizer optimizes the system instructions, selects sample prompts to add to the system instructions as few shot examples, or both.
Evaluation metric: the metric that the Vertex AI prompt optimizer uses to optimize the system instructions and/or select sample prompts.
Target model: the Google model that the Vertex AI prompt optimizer optimizes the system instructions and/or selects sample prompts for use with.

Create a configuration using one of the following options:

Notebook

If you want to run the Vertex AI prompt optimizer through the notebook, create a configuration by doing the following:

In Colab Enterprise, open the Vertex AI prompt optimizer notebook.

Go to Vertex AI prompt optimizer notebook
In the Configure project settings section, do the following:
1. In the PROJECT_ID field, enter your project ID.
2. In the LOCATION field, enter the location where you want to run the Vertex AI prompt optimizer.
3. In the OUTPUT_PATH field, enter the URI for the Cloud Storage bucket where you want the Vertex AI prompt optimizer to write the optimized system instructions and/or few shot examples. For example, gs://bucket-name/output-path.
4. In the INPUT_PATH field, enter the URI for the sample prompts in your Cloud Storage bucket. For example, gs://bucket-name/sample-prompts.jsonl.
In the Configure optimization settings section, do the following:
1. In the TARGET_MODEL field, enter the model that you want to optimize prompts for use with.
2. In the OPTIMIZATION_MODE, enter the optimization mode you want to use. Must be one of instruction, demonstration, or instruction_and_demo.
3. In the EVAL_METRIC field, enter the evaluation metric that you want to optimize your prompts for.
4. Optional: In the SOURCE_MODEL field, enter the Google model that the system instructions and prompts were previously used with. When the source_model parameter is set, the Vertex AI prompt optimizer runs your sample prompts on the source model to generate the ground truth responses for you, for evaluation metrics that require ground truth responses. If you didn't previously run your prompts with a Google model or you didn't achieve your target results, add ground truth responses to your prompt instead. For more information, see the Create a prompt and system instructions section of this document.
Optional: In the Configure advanced optimization settings section, you can additionally add any of the optional parameters to your configuration.

View optional parameters

In the NUM_INST_OPTIMIZATION_STEPS field, enter the number of iterations that the Vertex AI prompt optimizer uses in instruction optimization mode. The runtime increases linearly as you increase this value. Must be an integer between 10 and 20. If left unset, the default is 10.
In the NUM_TEMPLATES_PER_STEP field, enter the number of system instructions that the Vertex AI prompt optimizer generates and evaluates. Used with instruction and instruction_and_demo optimization mode. Must be an integer between 1 and 4. If left unset, the default is 2.
In the NUM_DEMO_OPTIMIZATION_STEPS field, enter the number of demonstrations that the Vertex AI prompt optimizer evaluates. Used with demonstration and instruction_and_demo optimization mode. Must be an integer between 10 and 30. If left unset, the default is 10.
In the NUM_DEMO_PER_PROMPT field, enter the number of demonstrations generated per prompt. Must be an integer between 3 and 6. If left unset, the default is 3.
In the TARGET_MODEL_QPS field, enter the queries per second (QPS) that the Vertex AI prompt optimizer sends to the target model. The runtime decreases linearly as you increase this value. Must be a float that is 3.0 or greater, but less than the QPS quota you have on the target model. If left unset, the default is 3.0.
In the SOURCE_MODEL_QPS field, enter the queries per second (QPS) that the Vertex AI prompt optimizer sends to the source model. Must be a float that is 3.0 or greater, but less than the QPS quota you have on the source model. If left unset, the default is 3.0.
In the EVAL_QPS field, enter the queries per second (QPS) that the Vertex AI prompt optimizer sends to the evaluation model, gemini-1.5-pro.
- For model based metrics, must be a float that is 3.0 or greater, but less than the quota you have for gemini-1.5-pro. If left unset, the default is 3.0.
- For custom metrics, must be a float that is 3.0 or greater. This determines the rate at which the Vertex AI prompt optimiser calls your custom metric Cloud Run functions.
If you want to use more than one evaluation metric, do the following:
1. In the EVAL_METRIC_1 field, enter an evaluation metric that you want to use.
2. In the EVAL_METRIC_1_WEIGHT field, enter the weight that you want the Vertex AI prompt optimizer to use when it runs the optimization.
3. In the EVAL_METRIC_2 field, enter an evaluation metric that you want to use.
4. In the EVAL_METRIC_2_WEIGHT field, enter the weight that you want the Vertex AI prompt optimizer to use when it runs the optimization.
5. In the EVAL_METRIC_3 field, optionally enter an evaluation metric that you want to use.
6. In the EVAL_METRIC_3_WEIGHT field, optionally enter
7. In the METRIC_AGGREGATION_TYPE field, enter the weight that you want the Vertex AI prompt optimizer to use when it runs the optimization.
In the PLACEHOLDER_TO_VALUE field, enter the information that replaces any variables in the system instructions. Information included within this flag is not optimized by the Vertex AI prompt optimizer.
In the RESPONSE_MIME_TYPE field, enter the MIME response type that the target model uses. Must be one of text/plain or application/json. If left unset, the default is text/plain.
In the TARGET_LANGUAGE field, enter the language of the system instructions. If left unset, the default is English.

SDK

If you want to run the Vertex AI prompt optimizer through the notebook, create a Create a JSON file with the parameters you want to use to optimize prompts by doing the following:

Create a JSON file with the parameters that you want to use to optimize your prompts. Each configuration file requires the following parameters:
```
{
 "project": "PROJECT_ID",
 "system_instruction_path": "SYSTEM_INSTRUCTION_PATH",
 "prompt_template_path": "PROMPT_TEMPLATE_PATH",
 "target_model": "TARGET_MODEL",
 "eval_metric": "EVALUATION_METRIC"
 "optimization_mode": "OPTIMIZATION_MODE",
 "input_data_path": "SAMPLE_PROMPT_URI",
 "output_path": "OUTPUT_URI"
}
```
Replace the following:
- PROJECT_ID: your project ID.
- SYSTEM_INSTRUCTION_PATH: the URI for the system instructions in your Cloud Storage bucket. For example, gs://bucket-name/system-instruction.txt.
- PROMPT_TEMPLATE: the URI for the prompt template in your Cloud Storage bucket. For example, gs://bucket-name/prompt-template.txt
- TARGET_MODEL: the model that you want to optimize prompts for use with.
- EVALUATION_METRIC: the evaluation metric that you want to optimize your prompts for.
- OPTIMIZATION_MODE: the optimization mode. Must be one of instruction, demonstration, or instruction_and_demo.
- SAMPLE_PROMPT_URI: the URI for the sample prompts in your Cloud Storage bucket. For example, gs://bucket-name/sample-prompts.jsonl.
- OUTPUT_URI: the URI for the Cloud Storage bucket where you want the Vertex AI prompt optimizer to write the optimized system instructions and/or few shot examples. For example, gs://bucket-name/output-path.
You can additionally add any of the optional parameters to your configuration file.

Optional parameters are broken down into 5 categories:
- Optimization process parameters. These parameters control the overall optimization process, including its duration and the number of optimization iterations it runs, which directly impacts the quality of optimizations.
- Model selection and location parameters. These parameters specify which models the Vertex AI prompt optimizer uses and the locations it uses those models in.
- Latency (QPS) parameters. These parameters control QPS, impacting the speed of the optimization process.
- Evaluation Metrics. These parameters define how the optimized prompts are evaluated.
- Other. Other parameters that control the structure and content of prompts.
  View optional parameters
```
"num_steps": NUM_INST_OPTIMIZATION_STEPS,
"num_template_eval_per_step": NUM_TEMPLATES_PER_STEP,
"num_demo_set_candidates": "NUM_DEMO_OPTIMIZATION_STEPS,
"demo_set_size": NUM_DEMO_PER_PROMPT,
"target_model_location": "TARGET_MODEL_LOCATION",
"source_model": "SOURCE_MODEL",
"source_model_location": "SOURCE_MODEL_LOCATION",
"target_model_qps": TARGET_MODEL_QPS,
"eval_qps": EVAL_QPS,
"source_model_qps": SOURCE_MODEL_QPS,
"eval_metrics_types": [EVALUATION_METRIC_LIST],
"eval_metrics_weights": [EVAL_METRICS_WEIGHTS],
"aggregation_type": "METRIC_AGGREGATION_TYPE",
"response_mime_type": "RESPONSE_MIME_TYPE",
"language": "TARGET_LANGUAGE",
"placeholder_to_content": "PLACEHOLDER_TO_CONTENT",
"data_limit": DATA_LIMIT
```
  Replace the following:
  - Optimization process parameters:
    
    NUM_INST_OPTIMIZATION_STEPS: the number of iterations that the Vertex AI prompt optimizer uses in instruction optimization mode. The runtime increases linearly as you increase this value. Must be an integer between 10 and 20. If left unset, the default is 10.
    
    NUM_TEMPLATES_PER_STEP: the number of system instructions that the Vertex AI prompt optimizer generates and evaluates. Used with instruction and instruction_and_demo optimization mode. Must be an integer between 1 and 4. If left unset, the default is 2.
    
    NUM_DEMO_OPTIMIZATION_STEPS: the number of demonstrations that the Vertex AI prompt optimizer evaluates. Used with demonstration and instruction_and_demo optimization mode. Must be an integer between 10 and 30. If left unset, the default is 10.
    
    NUM_DEMO_PER_PROMPT: the number of demonstrations generated per prompt. Must be an integer between 3 and 6. If left unset, the default is 3.
  - Model selection and location parameters:
    
    TARGET_MODEL_LOCATION: the location that you want to run the target model in. If left unset, the default is us-central1.
    
    SOURCE_MODEL: the Google model that the system instructions and prompts were previously used with. When the source_model parameter is set, the Vertex AI runs your sample prompts on the source model to generate the ground truth responses for you, for evaluation metrics that require ground truth responses. If you didn't previously run your prompts with a Google model or you didn't achieve your target results, add ground truth responses to your prompt instead. For more information, see the Create a prompt and system instructions section of this document.
    
    SOURCE_MODEL_LOCATION: the location that you want to run the source model in. If left unset, the default is us-central1.
  - Latency (QPS) parameters:
    
    Note: You must set a QPS that is lower than or equal to the QPM quota that is available to you, or your job will fail. To convert QPM quota to QPS, divide your QPM by 60. For example, a QPM quota of 600 is equivalent to a QPS of 600 (600/10 = 60).
    
    TARGET_MODEL_QPS: the queries per second (QPS) that the Vertex AI prompt optimizer sends to the target model. The runtime decreases linearly as you increase this value. Must be a float that is 3.0 or greater, but less than the QPS quota you have on the target model. If left unset, the default is 3.0.
    
    EVAL_QPS: the queries per second (QPS) that the Vertex AI prompt optimizer sends to the evaluation model, gemini-1.5-pro.
    
    For model based metrics, must be a float that is 3.0 or greater, but less than the quota you have for gemini-1.5-pro. If left unset, the default is 3.0.
    
    For custom metrics, must be a float that is 3.0 or greater. This determines the rate at which the Vertex AI prompt optimiser calls your custom metric Cloud Run functions.
    
    SOURCE_MODEL_QPS: the queries per second (QPS) that the Vertex AI prompt optimizer sends to the source model. Must be a float that is 3.0 or greater, but less than the QPS quota you have on the source model. If left unset, the default is 3.0.
  - Evaluation metrics parameters:
    
    EVALUATION_METRIC_LIST: a list of evaluation metrics. Use eval_metrics_types instead of eval_metric when you want to use multiple evaluation metrics. Must be an array. For example, "bleu", "summarization_quality".
    
    EVAL_METRICS_WEIGHTS: the weight for each eval_metrics_type. Must be an array.
    
    METRIC_AGGREGATION_TYPE: the type of aggregation used for the evaluation metrics. Must be one of weighted_sum or weighted_average. If left unset, the default is weighted_sum.
  - Other parameters:
    
    RESPONSE_MIME_TYPE: the MIME response type that the target model uses. Must be one of text/plain or application/json. If left unset, the default is text/plain.
    
    TARGET_LANGUAGE: the language of the system instructions. If left unset, the default is English.
    
    PLACEHOLDER_TO_CONTENT: the information that replaces any variables in the system instructions. Information included within this flag is not optimized by the Vertex AI prompt optimizer.
    
    DATA_LIMIT: the amount of data used for validation. The runtime increases linearly with this value. Must be an integer between 5 and 100. If left unset, the default is 100.
Upload the JSON file to a Cloud Storage bucket.

Run prompt optimizer

Run the Vertex AI prompt optimizer using one of the following options:

Notebook

Run the Vertex AI prompt optimizer through the notebook, by doing the following:

In Colab Enterprise, open the Vertex AI prompt optimizer notebook.

Go to Vertex AI prompt optimizer notebook
In the Run prompt optimizer section, click play_circle Run cell.

The Vertex AI prompt optimizer runs.

REST

Before using any of the request data, make the following replacements:

LOCATION: the location where you want to run the Vertex AI prompt optimizer.
PROJECT_ID: your project ID.
JOB_NAME: a name for the Vertex AI prompt optimizer job.
PATH_TO_CONFIG: the URI of the configuration file in your Cloud Storage bucket. For example, gs://bucket-name/configuration.json.

HTTP method and URL:

POST https://LOCATION-aiplatform.googleapis.com/v1/projects/PROJECT_ID/locations/LOCATION/customJobs

Request JSON body:

{
  "displayName": "JOB_NAME",
  "jobSpec": {
    "workerPoolSpecs": [
      {
        "machineSpec": {
          "machineType": "n1-standard-4"
        },
        "replicaCount": 1,
        "containerSpec": {
          "imageUri": "us-docker.pkg.dev/vertex-ai-restricted/builtin-algorithm/apd:preview_v1_0",
          "args": ["--config=PATH_TO_CONFIG""]
        }
      }
    ]
  }
}

To send your request, choose one of these options:

curl

Note: The following command assumes that you have logged in to the gcloud CLI with your user account by running gcloud init or gcloud auth login , or by using Cloud Shell, which automatically logs you into the gcloud CLI . You can check the currently active account by running gcloud auth list.

Save the request body in a file named request.json, and execute the following command:

curl -X POST \
     -H "Authorization: Bearer $(gcloud auth print-access-token)" \
     -H "Content-Type: application/json; charset=utf-8" \
     -d @request.json \
     "https://LOCATION-aiplatform.googleapis.com/v1/projects/PROJECT_ID/locations/LOCATION/customJobs"

PowerShell

Note: The following command assumes that you have logged in to the gcloud CLI with your user account by running gcloud init or gcloud auth login . You can check the currently active account by running gcloud auth list.

Save the request body in a file named request.json, and execute the following command:

$cred = gcloud auth print-access-token
$headers = @{ "Authorization" = "Bearer $cred" }

Invoke-WebRequest `
    -Method POST `
    -Headers $headers `
    -ContentType: "application/json; charset=utf-8" `
    -InFile request.json `
    -Uri "https://LOCATION-aiplatform.googleapis.com/v1/projects/PROJECT_ID/locations/LOCATION/customJobs" | Select-Object -Expand Content

The response looks similar to the following:

Response

{
  "name": "projects/PROJECT_ID/locations/LOCATION/customJobs/JOB_ID",
  "displayName": "JOB_NAME",
    "jobSpec": {
      "workerPoolSpecs": [
        {
        "machineSpec": {
              "machineType": "n1-standard-4"
          },
          "replicaCount": "1",
          "diskSpec": {
            "bootDiskType": "pd-ssd",
            "bootDiskSizeGb": 100
          },
          "containerSpec": {
            "imageUri": "us-docker.pkg.dev/vertex-ai-restricted/builtin-algorithm/apd:preview_v1_0"
            "args": [
            "--config=https://storage.mtls.cloud.google.com/testing-apd/testing-config.json"
          ]
          }
        }
      ]
    },
  "state": "JOB_STATE_PENDING",
  "createTime": "2020-09-15T19:09:54.342080Z",
  "startTime": "2020-09-15T19:13:42.991045Z",
}

Python

Before trying this sample, follow the Python setup instructions in the Vertex AI quickstart using client libraries. For more information, see the Vertex AI Python API reference documentation.

To authenticate to Vertex AI, set up Application Default Credentials. For more information, see Set up authentication for a local development environment.

from google.cloud import aiplatform

# Initialize Vertex AI platform
aiplatform.init(project=PROJECT_ID, location="us-central1")

# TODO(Developer): Check and update lines below
# cloud_bucket = "gs://cloud-samples-data"
# config_path = f"{cloud_bucket}/instructions/sample_configuration.json"
# output_path = "custom_job/output/"

custom_job = aiplatform.CustomJob(
    display_name="Prompt Optimizer example",
    worker_pool_specs=[
        {
            "replica_count": 1,
            "container_spec": {
                "image_uri": "us-docker.pkg.dev/vertex-ai-restricted/builtin-algorithm/apd:preview_v1_0",
                "args": [f"--config={cloud_bucket}/{config_path}"],
            },
            "machine_spec": {
                "machine_type": "n1-standard-4",
            },
        }
    ],
    staging_bucket=cloud_bucket,
    base_output_dir=f"{cloud_bucket}/{output_path}",
)

custom_job.submit()
print(f"Job resource name: {custom_job.resource_name}")
# Example response:
#    'projects/123412341234/locations/us-central1/customJobs/12341234123412341234'

Analyze results and iterate

After you run the Vertex AI prompt optimizer review the job's progress using one of the following options:

Notebook

If you want to view the results of the Vertex AI prompt optimizer through the notebook, do the following:

Open the Vertex AI prompt optimizer notebook.
In the Inspect the results section, do the following:
1. In the RESULT_PATH field, add the URI of the Cloud Storage bucket that you configured the Vertex AI prompt optimizer to write results to. For example, gs://bucket-name/output-path.
2. Click play_circle Run cell.

Console

In the Google Cloud console, in the Vertex AI section, go to the Training pipelines page.

Go to Training pipelines
Click the Custom jobs tab. Vertex AI prompt optimizer's custom training job appears in the list along with its status.

When the job is finished, review the optimizations by doing the following:

In the Google Cloud console, go to the Cloud Storage Buckets page:

Go to Buckets
Click the name of your Cloud Storage bucket.
Navigate to the folder that has the same name as the optimization mode you used to evaluate the prompts, either instruction or demonstration. If you used instruction_and_demo mode, both folders appear. The instruction folder contains the results from the system instruction optimization, while the demonstration folder contains the results from the demonstration optimization and the optimized system instructions.

The folder contains the following files:
- config.json: the complete configuration that the Vertex AI prompt optimizer used.
- templates.json: each set of system instructions and/or few shot examples that the Vertex AI prompt optimizer generated and their evaluation score.
- eval_results.json: the target model's response for each sample prompt for each set of generated system instructions and/or few shot examples and their evaluation score.
- optimized_results.json: the best performing system instructions and/or few shot examples and their evaluation score.
To view the optimized system instructions, view the optimized_results.json file.

What's next

Try the Vertex AI prompt optimizer SDK notebook.
Learn about responsible AI best practices and Vertex AI's safety filters.
Learn more about prompting strategies.
Explore examples of prompts in the Prompt gallery.