Skip to content

Commit a80466a

Browse files
authored
Updating README including use of Vertex AI API, and AAP evaluations information. (#97)
1 parent be2cb9f commit a80466a

File tree

1 file changed

+27
-2
lines changed

1 file changed

+27
-2
lines changed

README.md

Lines changed: 27 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -140,6 +140,13 @@ Runs basic tests against the local container.
140140
make run-test
141141
```
142142

143+
## AAP quality evaluations
144+
145+
AAP Chatbot Quality evaluations available:
146+
147+
* [AAP documentation retrieval evaluation](https://github.com/ansible/ansible-wisdom-testing/blob/main/README.md#chatbot-evaluation-testing)
148+
* [AAP Inventory file generation evaluation](https://github.com/ansible-automation-platform/aap-installers-rag-content/tree/main/tools#usage)
149+
143150
## Deploy into a k8s cluster
144151

145152
### Change configuration in `kustomization.yaml` accordingly, then
@@ -154,9 +161,11 @@ Runs basic tests against the local container.
154161
kubectl apply -f my-chatbot-stack-deploy.yaml
155162
```
156163

157-
## Appendix - Google Gemini
164+
## Appendix - Google Gemini API
165+
166+
Using the [`gemini` remote inference provider](https://llama-stack.readthedocs.io/en/latest/providers/inference/remote_gemini.html):
158167

159-
* Please set the environment variable `OPENAI_API_KEY=<YOUR_API_KEY>`
168+
* Set the environment variable `OPENAI_API_KEY=<YOUR_API_KEY>`
160169
* Example of a `v1/query` request:
161170
```json
162171
{
@@ -167,6 +176,22 @@ Runs basic tests against the local container.
167176
}
168177
```
169178

179+
## Appendix - Google Vertex API
180+
181+
Using the [`gemini` remote inference provider](https://llama-stack.readthedocs.io/en/latest/providers/inference/remote_gemini.html):
182+
183+
* Set a dummy value for the environment variable `OPENAI_API_KEY` (so `gemini` provider within llama-stack, does not complain)
184+
* Set the path for your Google's Service Account credentials JSON file in the env `GOOGLE_APPLICATION_CREDENTIALS=<PATH_GOOGLE_CRED_JSON_FILE>`
185+
* Example of a `v1/query` request:
186+
```json
187+
{
188+
"query": "hello",
189+
"system_prompt": "You are a helpful assistant.",
190+
"model": "gemini-2.5-flash",
191+
"provider": "gemini"
192+
}
193+
```
194+
170195

171196
## Appendix - Host clean-up
172197

0 commit comments

Comments
 (0)