Amazon Bedrock - foundation Models - hands on Flashcards
(19 cards)
What is a foundation model?
A foundation model is a type of AI model that serves as a base for various applications.
Which company offers one of the cheapest foundation models on Amazon Bedrock?
Amazon
What is the name of the leading AI company mentioned that has Claude 3 models?
Anthropic
What type of models does Stability AI specialize in?
Image generation models using stable diffusion
What is more important to know about foundation models from an exam perspective?
Differences in capabilities of a model
True or False: You need to know the specific models that can handle image uploads.
False
What is one key metric to compare AI models?
Latency
Fill in the blank: Amazon Titan Text G1 Premier does not support _______.
image upload
What two options are available for customizing a model in Amazon Bedrock?
- Fine tuning job
- Continued pre-training job
What is the main difference between a fine tuning job and a continued pre-training job?
Fine tuning is a one-time process, while continued pre-training is ongoing.
What type of data is required for fine tuning a model in Amazon Bedrock?
Input data located in Amazon S3
What are buckets in Amazon S3?
Cloud directories for storing data
What are hyperparameters in the context of machine learning?
Configurations that dictate how the algorithm behaves
What can changing hyperparameters affect?
The quality of the model’s outputs
What is required for Amazon Bedrock to write to Amazon S3?
A service role in AWS
Fill in the blank: To create fine-tuned models, you need to purchase _______.
provisioned throughput
What do you need to specify when creating a fine tuning job?
- Model selection
- Input data location
- Hyperparameters
What does the learning rate determine in a machine learning model?
How fast the model learns
What is the purpose of a validation dataset in model training?
To analyze and ensure the model fits the use case
