Banana Python SDK
Getting Started
Install via pip
pip3 install banana-dev
Get your API Key
Run:
import os
import asyncio
# Set API Key Implicitly
os.environ['BANANA_APIKEY'] = '...'
models = {
'model1': 'modelkey',
'model2': 'modelkey',
}
from banana_dev import BananaClient
# Configure client implicitly
client = BananaClient(
models = models
)
# Configure client explicitly
client = BananaClient(
apikey = ...,
models = models,
model_name = 'model2', # use model2 as the default model
model_key = 'modelkey', # use modelkey if models isn't configured
json_results = True, # return results in json format, otherwise will return in pydantic model format
)
# Will save your configuration to ~/.banana/config.json
# future calls you can implicitly initiate the client without configuring it again
client = BananaClient()
model_inputs = {
# a json specific to your model. For example:
"imageURL": "https://demo-images-banana.s3.us-west-1.amazonaws.com/image2.jpg"
}
# Optionally specify the model
# out = client.run(model_inputs, model_name = 'model1')
# can also call the client directly:
# out = client(model_inputs)
out = client.run(model_inputs)
print(out)
# Supports Asyncronous Calls
out = asyncio.run(client.async_run(model_inputs))
print(out)Return type:
{
"id": "12345678-1234-1234-1234-123456789012",
"message": "success",
"created": 1649712752,
"apiVersion": "26 Nov 2021",
"modelOutputs": [
{
# a json specific to your model. In this example, the caption of the image
"caption": "a baseball player throwing a ball"
}
],
"duration": 11.535236
}Parse the server output:
# if json_results is True, the output will be in json format
model_out = out["modelOutputs"][0]
# otherwise can access the output as a pydantic model
model_out = out.modelOutputs[0]