- Newest
- Most votes
- Most comments
To use the generative AI features in Amazon Lex V2 bot creation, you need to complete a specific sequence of steps. Based on your description, it seems you might be missing some prerequisites or steps in the process.
Here's what you need to do:
-
First, ensure you have signed up for access to the Anthropic Claude model in the Amazon Bedrock console. It's not enough to just have the model active - you need to explicitly request access to it.
-
After creating your bot but before attempting to use the descriptive bot builder, you need to turn on the generative AI capabilities for your bot locale. This is a separate step from just adding the language.
-
Ensure your IAM role has the proper permissions. While you've added the "bedrock:InvokeModel" permission, you might also need additional permissions depending on your setup.
The error message "Please ensure your botLocale has Descriptive bot builder enabled with a valid foundation model" suggests that either:
- The generative AI capabilities haven't been properly enabled for your bot locale
- There might be an issue with the model access in Amazon Bedrock
- Your IAM permissions might not be sufficient
Try creating the bot first, then adding a language locale, and only after that enable the generative AI capabilities for that locale. The sequence of operations is important here.
If you continue to face issues, verify in the Amazon Bedrock console that you have successfully completed the model access request process for the Claude model you're trying to use.
Sources
Optimize Lex V2 bot creation and performance by using generative AI - Amazon Lex
Use a description to build a bot in Lex V2 with the descriptive bot builder - Amazon Lex
AMAZON.QnAIntent - Amazon Lex
answered a year ago
-
The model overview indicates access granted. Is that enough?
-
Alright, but how can I actually do that? I really don't see that step anywhere until a language has been added?
-
Now I am just asking about creating from the Descriptive bot builder. It must be possible to answer about the necesssary permissions for that?
I have tried to add a language and then enable the gen AI features. Then I get the exact same error.
answered a year ago
When enabling Descriptive bot builder in Bot building generative AI features:
Error updating generative AI configurations for English (US) Please ensure your botLocale has Descriptive bot builder enabled with a valid foundation model, then try your request again.
Seems weird to get that error when trying to enable exactly what it is complaining about? Unless I am completely misinterpretating the error 😊
answered a year ago
Furthermore, when trying to create-bot-locale with
aws lexv2-models create-bot-locale \
--bot-id ----------- \
--bot-version DRAFT \
--locale-id en_US \
--nlu-intent-confidence-threshold 0.5 \
--profile chatbot \
--generative-ai-settings '{
"buildtimeSettings": {
"descriptiveBotBuilder": {
"enabled": true,
"bedrockModelSpecification": {
"modelArn": "arn:aws:bedrock:----------------"
}
}
},
"runtimeSettings": {
"nluImprovement": {
"enabled": true
},
"slotResolutionImprovement": {
"enabled": true,
"bedrockModelSpecification": {
"modelArn": "arn:aws:bedrock:------------------"
}
}
}
}'
I get: Unknown parameter in generativeAISettings.runtimeSettings: "nluImprovement", must be one of: slotResolutionImprovement
On all docs the nluImprovement is in runtimeSettings (https://docs.aws.amazon.com/lexv2/latest/APIReference/API_UpdateBotLocale.html, https://docs.aws.amazon.com/cli/latest/reference/lexv2-models/update-bot-locale.html)
If I leave it out I get: An error occurred (ValidationException) when calling the CreateBotLocale operation: Please ensure your botLocale has Slot resolution improvement enabled with a valid foundation model, then try your request again.
Can it be the actual model arns for different allowed models in slotResolutionImprovement and descriptiveBotBuilder? How do I know which models in each?
I'm trying with arn:aws:bedrock:eu-central-1::foundation-model/anthropic.claude-3-sonnet-20240229-v1:0 - is it wrong? Where can I find the correct one?
answered a year ago
Relevant content
asked 2 years ago
asked a year ago
asked 2 years ago
- AWS OFFICIALUpdated 4 years ago
