This package provides functionality developed to simplify interfacing with the Cohere models via AWS SageMaker Marketplace in Python 3.
The package can be installed with pip:
pip install --upgrade cohere-sagemaker
Install from source:
python setup.py install
To use this library, you need to configure your AWS credentials. You can do this by running aws configure
. Once that's set up, you can use the library like this:
from cohere_sagemaker import Client
client = Client(endpoint_name='my-cohere-endpoint')
# generate prediction for a prompt
response = client.generate(
prompt="Tell me a story about",
max_tokens=20)
print(response.generations[0].text)
# a time when you had to make a difficult decision. What did you do
Note: by default we assume region configured in AWS CLI (aws configure get region
), to override use region_name
parameter, e.g.
client = Client(endpoint_name='my-cohere-endpoint', region_name='us-east-1')
More detailed examples can be found in the Jupyter notebook.