mirror of
https://github.com/microsoft/semantic-kernel.git
synced 2026-10-12 02:19:06 +00:00
David Grieve
bcf87d1afb
### TODO - [x] Update pom files. This may include disabling the build of existing modules. - [x] Get a clean compile. `./mvnw clean install -Prelease,with-samples -Dmaven.test.skip=true` works - [x] Where does the assistants package belong? ~~For now, I just put it in with 06-Assistants~~ Resolution: move to /plugins/semantickernel-plugin-core - [x] I'm not sure `aiservices-*` belong under `extensions` Resolution: make an aiservices module and put them there - [x] I'm not sure the bing plugin belongs under `plugins\web` Resolution: yeah, that works for now ### Motivation and Context <!-- Thank you for your contribution to the semantic-kernel repo! Please help reviewers and future users, providing the following information: 1. Why is this change required? 2. What problem does it solve? 3. What scenario does it contribute to? 4. If it fixes an open issue, please link to the issue here. --> We were working from two repos. This will make Java v1 development easier for all. ### Description <!-- Describe your changes, the overall approach, the underlying design. These notes will help understanding how your code works. Thanks! --> Took compiling code from sk-v1-proposal/java and moved it to semantic-kernel/java. Modified pom files as needed. Modifications to code as needed to get it to compile. ### Contribution Checklist <!-- Before submitting this PR, please make sure: --> - [ ] The code builds clean without any errors or warnings - [ ] The PR follows the [SK Contribution Guidelines](https://github.com/microsoft/semantic-kernel/blob/main/CONTRIBUTING.md) and the [pre-submission formatting script](https://github.com/microsoft/semantic-kernel/blob/main/CONTRIBUTING.md#development-scripts) raises no violations - [ ] All unit tests pass, and I have added new tests where possible - [ ] I didn't break anyone 😄
Java Samples
TL;DR
Run with:
# AZURE:
OPENAI_CLIENT_TYPE=AZURE_OPEN_AI \
AZURE_OPEN_AI_KEY="my-key" \
AZURE_OPEN_AI_ENDPOINT="endpoint url" \
../../mvnw clean package exec:java -Dsample=Example04_CombineLLMPromptsAndNativeCode
# OPENAI:
OPENAI_CLIENT_TYPE=OPENAI \
OPEN_AI_KEY="my-key" \
OPEN_AI_ORGANIZATION_ID="organisation id" \
../../mvnw clean package exec:java -Dsample=Example04_CombineLLMPromptsAndNativeCode
Compile
These samples can be compiled via:
../../mvnw clean package
They can then be run by:
../../mvnw exec:java -Dsample=Example04_CombineLLMPromptsAndNativeCode
Configuration
By default, the samples will use the Open AI client, but you can also use the Azure Open AI client.
Open AI client type
You can define the provider of Open AI (openai.com or Azure), this can be done by setting the OPENAI_CLIENT_TYPE
property or environment variable to either OPENAI
or AZURE_OPEN_AI. By default, the samples will use the
Open AI client.
OPENAI_CLIENT_TYPE=OPEN_AI ../../mvnw exec:java -Dsample=Example04_CombineLLMPromptsAndNativeCode
OR
../../mvnw exec:java -DOPENAI_CLIENT_TYPE=AZURE_OPEN_AI -Dsample=Example04_CombineLLMPromptsAndNativeCode
Client Settings
The samples search for the client settings in the following order:
- Properties file whose location is defined by the
CONF_PROPERTIESproperty or environment variable. - System properties defined on the command line.
- Environment variables.
- Properties file at
java/samples/conf.properties. - Properties file at
~/.sk/conf.properties.
Properties File
You can set the location of a properties file, by setting the CONF_PROPERTIES property or environment variable, ie:
CONF_PROPERTIES=my.properties \
OPENAI_CLIENT_TYPE=OPEN_AI \
../../mvnw exec:java -Dsample=Example04_CombineLLMPromptsAndNativeCode
OR
../../mvnw exec:java \
-DCONF_PROPERTIES=my.properties \
-DOPENAI_CLIENT_TYPE=AZURE_OPEN_AI \
-Dsample=Example04_CombineLLMPromptsAndNativeCode
A properties file looks as follows:
# If using openai.com
client.openai.key:"my-key"
client.openai.organizationid:"my-org-id"
# if using Azure Open AI
client.azureopenai.key:"my-key"
client.azureopenai.endpoint:"url of azure openai endpoint"
client.azureopenai.deploymentname:"deployment name"
System Properties
As an alternative to providing the key/endpoint properties via a file, you can set them directly via system properties, ie:
# OpenAI
../../mvnw exec:java \
-DOPENAI_CLIENT_TYPE=AZURE_OPEN_AI \
-Dclient.openai.key="my-key" \
-Dclient.openai.organizationid="my-org-id" \
-Dsample=Example04_CombineLLMPromptsAndNativeCode
# Azure
../../mvnw exec:java \
-DOPENAI_CLIENT_TYPE=AZURE_OPEN_AI \
-Dclient.azureopenai.key="my-key" \
-Dclient.azureopenai.endpoint="url of azure openai endpoint" \
-Dsample=Example04_CombineLLMPromptsAndNativeCode
Environment variables
Alternative to properties, you can set environment variables as follows:
# AZURE:
OPENAI_CLIENT_TYPE=AZURE_OPEN_AI \
AZURE_OPEN_AI_KEY="my-key" \
AZURE_OPEN_AI_ENDPOINT="endpoint url" \
../../mvnw clean package exec:java -Dsample=Example04_CombineLLMPromptsAndNativeCode
# OPENAI:
OPENAI_CLIENT_TYPE=OPEN_AI \
OPEN_AI_KEY="my-key" \
OPEN_AI_ORGANIZATION_ID="organisation id" \
../../mvnw clean package exec:java -Dsample=Example04_CombineLLMPromptsAndNativeCode
Managed Identity and Azure CLI Credentials (Azure only)
The credentialtype setting (i.e client.openai.credentialtype):
managed_identity: will attempt to obtain credentials via Azure's managed identity.azure_cli: will attempt to obtain credentials via Azure's CLI, to use this you must have logged in to the az client for instance by runningaz login.
To use either of these settings, the azure-identity dependency must be available to the application. Additionally, you
must have the "Cognitive Services User" role assigned to your identity.