Building GenAI with Bedrock

Question 1

Can bedrock integrate with SageMaker Canvas?

Answer 1

Yes

Question 2

What are the four bedrock API endpoints?

Answer 2

Bedrock - deploy and train

Bedrock-runtime - inference

Bedrock-agent - Manage, Deploy, and Train

Bedrock-agent-runtime - perform inference

Question 3

Can you use Bedrock with a root account?

Answer 3

No

Question 4

Can Bedrock Playground be used for evaluating your own custom or imported models?

Answer 4

Yes

Question 5

What is fine-tuning do?

Answer 5

It is additional training using your own data.

Question 6

Does continued pre-training use labeled or unlabeled data?

Answer 6

Unlabeled

Question 7

What is continued pre-training?

Answer 7

It is like fine-tuning, but with unlabeled data

Question 8

What is cheaper, RAG or Fine-Tuning?

Answer 8

RAG

Question 9

What is a knowledge base?

Answer 9

A vector database.

Question 10

Will Opensearch work as a knowledgebase?

Answer 10

Yes

Question 11

Can Aurora support Vector Databases?

Answer 11

Yes

Question 12

What are Bedrock Guardrails?

Answer 12

Content filtering on prompts and responses.

Question 13

Do Guardrails work with image foundation models?

Answer 13

No

Question 14

Can Bedrock remove PII with Guardrails?

Answer 14

Yes

Question 15

Do Guardrails work for agents or knowledge bases?

Answer 15

Both