diff --git a/notebooks/vertex_genai/solutions/grounding_vertex_agent_builder.ipynb b/notebooks/vertex_genai/solutions/grounding_vertex_agent_builder.ipynb index a3df2b17..1646f461 100644 --- a/notebooks/vertex_genai/solutions/grounding_vertex_agent_builder.ipynb +++ b/notebooks/vertex_genai/solutions/grounding_vertex_agent_builder.ipynb @@ -7,7 +7,7 @@ "tags": [] }, "source": [ - "# Grounding with Vertex AI Agent Builder" + "# Grounding PaLM with Vertex AI Agent Builder" ] }, { @@ -64,7 +64,7 @@ }, { "cell_type": "code", - "execution_count": 6, + "execution_count": null, "metadata": { "id": "oM1iC_MfAts1", "tags": [] @@ -78,7 +78,7 @@ }, { "cell_type": "code", - "execution_count": 7, + "execution_count": null, "metadata": { "id": "init_aip:mbsdk,all", "tags": [] @@ -101,7 +101,7 @@ }, { "cell_type": "code", - "execution_count": 10, + "execution_count": null, "metadata": { "id": "PyQmSRbKA8r-", "tags": [] @@ -126,7 +126,7 @@ }, { "cell_type": "code", - "execution_count": 20, + "execution_count": null, "metadata": { "tags": [] }, @@ -181,7 +181,7 @@ }, { "cell_type": "code", - "execution_count": 12, + "execution_count": null, "metadata": { "tags": [] }, @@ -190,7 +190,7 @@ "DATA_STORE_PROJECT_ID = PROJECT_ID # @param {type:\"string\"}\n", "DATA_STORE_REGION = \"global\" # @param {type:\"string\"}\n", "# Replace this with your data store ID from Vertex AI Search\n", - "DATA_STORE_ID = \"gcp-documentation_1715192184974\" # @param {type:\"string\"}" + "DATA_STORE_ID = \"\" # TODO: ENTER YOUR DATASTORE ID" ] }, { @@ -202,7 +202,7 @@ }, { "cell_type": "code", - "execution_count": 13, + "execution_count": null, "metadata": { "tags": [] }, @@ -222,30 +222,11 @@ }, { "cell_type": "code", - "execution_count": 15, + "execution_count": null, "metadata": { "tags": [] }, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Response from model without grounding:\n", - " Object tables in BigQuery are used when you want to store and query data that has a complex structure, such as nested objects or arrays. Object tables are also useful when you want to store data that is not easily represented in a relational table, such as JSON or XML data.\n", - "\n", - "Here are some specific cases where you might want to use an object table in BigQuery:\n", - "\n", - "* To store data that has a complex structure, such as nested objects or arrays. For example, you could use an object table to store customer data, where each customer has a name, address, and list of orders.\n", - "* To store data that is not easily represented in a relational table, such as JSON or XML data. For example, you could use an object table to store data from a web service that returns JSON responses.\n", - "* To improve query performance on data that has a complex structure. Object tables can help to improve query performance by allowing you to filter and aggregate data based on the properties of objects.\n", - "\n", - "Here are some examples of how object tables can be used in practice:\n", - "\n", - "* A company could use an object table to store customer data, where each customer has a name, address, and list of orders. This data could be used to generate reports on customer purchases,\n" - ] - } - ], + "outputs": [], "source": [ "response = text_model.predict(\n", " PROMPT,\n", @@ -265,28 +246,11 @@ }, { "cell_type": "code", - "execution_count": 19, + "execution_count": null, "metadata": { "tags": [] }, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Response from Model:\n", - " Object tables can be used in BigQuery in the following scenarios:\n", - "\n", - "- To analyze unstructured data in Cloud Storage.\n", - "- To perform analysis with remote functions.\n", - "- To perform inference using BigQuery ML.\n", - "\n", - "\n", - "Grounding Metadata:\n", - "GroundingMetadata(citations=[GroundingCitation(start_index=37, end_index=99, url='https://cloud.google.com/bigquery/docs/object-table-introduction', title='Introduction to object tables | BigQuery | Google Cloud', license=None, publication_date=None), GroundingCitation(start_index=101, end_index=159, url='https://cloud.google.com/bigquery/docs/object-table-introduction', title='Introduction to object tables | BigQuery | Google Cloud', license=None, publication_date=None), GroundingCitation(start_index=161, end_index=219, url='https://cloud.google.com/bigquery/docs/object-table-introduction', title='Introduction to object tables | BigQuery | Google Cloud', license=None, publication_date=None)], search_queries=['When to use an object table in BigQuery?'])\n" - ] - } - ], + "outputs": [], "source": [ "if DATA_STORE_ID and DATA_STORE_REGION:\n", " # Use Vertex AI Search data store\n", @@ -330,7 +294,7 @@ }, { "cell_type": "code", - "execution_count": 21, + "execution_count": null, "metadata": { "tags": [] }, @@ -351,41 +315,11 @@ }, { "cell_type": "code", - "execution_count": 22, + "execution_count": null, "metadata": { "tags": [] }, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - " Managed datasets in Vertex AI are curated and maintained by Google. They provide a convenient way to access high-quality data for training and evaluating machine learning models. Managed datasets are available in a variety of domains, including healthcare, finance, and retail.\n", - "\n", - "Here are some of the benefits of using managed datasets in Vertex AI:\n", - "\n", - "* **High-quality data:** Managed datasets are curated by Google experts to ensure that they are accurate, complete, and consistent.\n", - "* **Convenient access:** Managed datasets are hosted in the cloud, so you can access them from anywhere with an internet connection.\n", - "* **Scalable:** Managed datasets can be scaled to meet the needs of your machine learning models.\n", - "* **Secure:** Managed datasets are protected by Google's robust security measures.\n", - "\n", - "To learn more about managed datasets in Vertex AI, please visit the [documentation](https://cloud.google.com/vertex-ai/docs/datasets/overview).\n", - " You can use a variety of data types with Vertex AI, including:\n", - "\n", - "* **Structured data:** This type of data is organized in a tabular format, such as a CSV file or a database table.\n", - "* **Unstructured data:** This type of data is not organized in a tabular format, such as images, videos, or text documents.\n", - "* **Semi-structured data:** This type of data is a combination of structured and unstructured data, such as a JSON file or an HTML document.\n", - "\n", - "Vertex AI can also handle data that is stored in a variety of formats, including:\n", - "\n", - "* **Local files:** You can upload local files to Vertex AI using the Vertex AI console or the Vertex AI SDK.\n", - "* **Cloud storage:** You can access data that is stored in Google Cloud Storage or Amazon S3 using Vertex AI.\n", - "* **BigQuery:** You can access data that is stored in BigQuery using Vertex AI.\n", - "\n", - "To learn more about the data types and formats that Vertex AI can handle, please visit the [documentation](https://cloud.google.com/vertex-ai/docs/data-formats).\n" - ] - } - ], + "outputs": [], "source": [ "chat = chat_model.start_chat()\n", "\n", @@ -409,27 +343,11 @@ }, { "cell_type": "code", - "execution_count": 24, + "execution_count": null, "metadata": { "tags": [] }, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - " Managed datasets in Vertex AI are used to provide the source data used to train AutoML and custom models.\n", - "GroundingMetadata(citations=[GroundingCitation(start_index=1, end_index=106, url='https://cloud.google.com/vertex-ai/docs/datasets/overview', title='Overview of creating managed datasets on Vertex AI | Google Cloud', license=None, publication_date=None)], search_queries=['Vertex AI managed datasets'])\n", - " You can use the following types of data to train a model:\n", - "\n", - "- Tabular data: This is data that is organized in rows and columns, like a spreadsheet.\n", - "- Image data: This is data that is represented as a collection of pixels, like a photograph.\n", - "- Video data: This is data that is represented as a sequence of images, like a movie.\n", - "- Text data: This is data that is represented as a sequence of characters, like a document.\n", - "GroundingMetadata(citations=[GroundingCitation(start_index=1, end_index=58, url='https://cloud.google.com/vertex-ai/docs/training-overview', title='Train and use your own models | Vertex AI | Google Cloud', license=None, publication_date=None), GroundingCitation(start_index=1, end_index=58, url='https://cloud.google.com/bigquery/docs/reference/standard-sql/bigqueryml-syntax-create', title='The CREATE MODEL statement | BigQuery | Google Cloud', license=None, publication_date=None)], search_queries=['What types of data can I use to train a model?'])\n" - ] - } - ], + "outputs": [], "source": [ "chat = chat_model.start_chat()\n", "grounding_source = GroundingSource.VertexAISearch(\n",