Family of Titan Text Models - Boto3 Demo
Dive into the capabilities of these models in our latest Blog3 on AWS Bedrock - Learning Series. Explore sample code for leveraging the Titan Text Model family, including Express, Lite, and Embedding.
Github Link - Notebook
https://github.com/jayyanar/learning-aws-bedrock/blob/main/blog3/Bedrock_Titan_Learning.ipynb
Environment Setup
I am using vscode local environment with AWS Credential configured.
Install Latest Python
! python --version
Python 3.11.5Upgrade pip
! pip install --upgrade pipInstall latest boto3,awscli, boto3-core
! pip install --no-build-isolation --force-reinstall \
"boto3>=1.33.6" \
"awscli>=1.31.6" \
"botocore>=1.33.6"Load the Library
import json
import os
import sys
import boto3
import botocore
bedrock = boto3.client(service_name="bedrock")
bedrock_runtime = boto3.client(service_name="bedrock-runtime")Titan Text Model - Express
Set the Prompt
express_prompt = "write article about AWS Lambda"Configure the Model configuration
body = json.dumps({
"inputText": express_prompt,
"textGenerationConfig":{
"maxTokenCount":128,
"stopSequences":[], #define phrases that signal the model to conclude text generation.
"temperature":0, #Temperature controls randomness; higher values increase diversity, lower values boost predictability.
"topP":0.9 # Top P is a text generation technique, sampling from the most probable tokens in a distribution.
}
})Invoke the Model
response = bedrock_runtime.invoke_model(
body=body,
modelId="amazon.titan-text-express-v1",
accept="application/json",
contentType="application/json"
)Parse the Configuration
response_body = json.loads(response.get('body').read())
outputText = response_body.get('results')[0].get('outputText')
#The code text = outputText[outputText.index('\n')+1:] extracts the substring following the first newline character in the outputText string. This is useful for scenarios where the initial content before the first newline is irrelevant, and you want to capture the text that comes after it. The index('\n')+1 locates the position of the first newline character, and the slicing [index+1:] fetches the subsequent text, assigning it to the variable text
text = outputText[outputText.index('\n')+1:]
about_lambda = text.strip()
print(about_lambda)Text completion:
AWS Lambda, an Amazon Web Services (AWS) serverless computing service, enables developers to execute code in response to events without managing infrastructure. It offers high scalability, automatically adjusting to application demand. With a pay-as-you-go model, it's cost-effective for unpredictable workloads. Supporting various languages like Python and Java, Lambda provides flexibility. Integration with AWS services like S3 and DynamoDB streamlines application development. To use AWS Lambda, developers create functions, configure event sources (e.g., CloudWatch, S3), and invoke functions through the AWS Lambda API or CLI, providing a versatile and efficient solution for building serverless applications.Titan Text Model - Lite
Set the Prompt
lite_prompt = "2 difference between AWS DynamoDB and AWS Redis"Configure the Model configuration
body = json.dumps({
"inputText": lite_prompt,
"textGenerationConfig":{
"maxTokenCount":128,
"stopSequences":[], #define phrases that signal the model to conclude text generation.
"temperature":0, #Temperature controls randomness; higher values increase diversity, lower values boost predictability.
"topP":0.9 # Top P is a text generation technique, sampling from the most probable tokens in a distribution.
}
})Invoke the Model
response = bedrock_runtime.invoke_model(
body=body,
modelId="amazon.titan-text-lite-v1",
accept="application/json",
contentType="application/json"
)Parse the Configuration
response_body = json.loads(response.get('body').read())
outputText = response_body.get('results')[0].get('outputText')
text = outputText[outputText.index('\n')+1:]
compare_dynamodb_redis = text.strip()
print(compare_dynamodb_redis)Text within this block will maintain its original spacing when published
Text completion:
Amazon DynamoDB is a fully managed NoSQL database service in the cloud that offers fast and predictable performance with seamless scalability. It is designed to run high-performance applications at any scale. On the other hand, Amazon Redis is a fully managed in-memory data structure store that provides real-time analytics, caching, and key-value data storage. It is suitable for applications that require fast data retrieval and low latency.Titan Text Model - Embedding
Set the Prompt
embed_prompt = "AWS re:Invent 2023, our biggest cloud event of the year, in Las Vegas, Nevada, featured keynotes, innovation talks, builder labs, workshops, tech and sustainability demos"Configure the Model configuration
body = json.dumps({
"inputText": embed_prompt,
})Invoke the Model
response = bedrock_runtime.invoke_model(
body=body,
modelId="amazon.titan-embed-text-v1",
accept="application/json",
contentType="application/json"
)Parse the Configuration
response_body = json.loads(response.get("body").read())
embedding_output = response_body.get("embedding")
#This code retrieves the "embedding" vector from the response body and prints its length along with a preview of the first three and last three values, showing a snippet of the embedding vector.
print(f"You can find the Embedding Vector {len(embedding_output)} values\n{embedding_output[0:3]+['...']+embedding_output[-3:]}")You can find the Embedding Vector 1536 values [0.40429688, -0.38085938, 0.19726562, '...', 0.2109375, 0.012573242, 0.18847656]
Titan MultiModel - Embedding
Download the Image
! wget https://raw.githubusercontent.com/jayyanar/learning-aws-bedrock/main/blog3/tajmahal.jpg
Configure the Model configuration
# The size of the file the 1024*1024 - Created using Stability XL
with open("tajmahal.jpg", "rb") as image_file:
tajmahal_image = base64.b64encode(image_file.read()).decode('utf8')
body = json.dumps(
{
"inputImage": tajmahal_image
}
)Invoke the Model
response = bedrock_runtime.invoke_model(
body=body,
modelId="amazon.titan-embed-image-v1",
accept="application/json",
contentType="application/json"
)Parse the Configuration
response_body = json.loads(response.get("body").read())
embedding_output = response_body.get("embedding")
#This code retrieves the "embedding" vector from the response body and prints its length along with a preview of the first three and last three values, showing a snippet of the embedding vector.
print(f"You can find the Embedding Vector {len(embedding_output)} values\n{embedding_output[0:3]+['...']+embedding_output[-3:]}")You can find the Embedding Vector 1024 values [0.016090883, 0.03545449, 0.0026249958, '...', 0.0065908986, 0.0172727, 0.002738632]
Titan Image Generator
Prompt for Image Generation
image_prompt = "Beautiful basketball ground"Configure the Model configuration
body = json.dumps(
{
"taskType": "TEXT_IMAGE",
"textToImageParams": {
"text": image_prompt, # Required
# "negativeText": "<text>" # Optional
},
"imageGenerationConfig": {
"numberOfImages": 1, # Range: 1 to 5
"quality": "premium", # Options: standard or premium
"height": 1024, # Supported height list in the docs
"width": 1024, # Supported width list in the docs
"cfgScale": 7.5, # Range: 1.0 (exclusive) to 10.0
"seed": 42 # Range: 0 to 214783647
}
}
)Invoke the Model
response = bedrock_runtime.invoke_model(
body=body,
modelId="amazon.titan-image-generator-v1",
accept="application/json",
contentType="application/json"
)Parse and store the Image Output
import base64
from PIL import Image
from io import BytesIO
response_body = json.loads(response.get("body").read())
images = [Image.open(BytesIO(base64.b64decode(base64_image))) for base64_image in response_body.get("images")]
for i, img in enumerate(images):
img.save(f"tajmahal_bedrock_{i+1}.jpg")
\(Stay... tuned and subscribe. for. future. bedrock
blogs\)