run-replicate-model

Description

run-replicate-model adaptor uses Replicate HTTP API to run an AI model and wait for its output.

Replicate offers thousands of community-contributed AI models. A list of models can be found on Replicate website.

Note that this adaptor can be used to run one model only. If you would like to run a model for each value in a datatable column, consider using the workflow repeater adaptor.

Inputs

model Type: text Required: Yes The URL of the Replicate model to run, e.g.. stability-ai/stable-diffusion:db21e45d.

input Type: dictionary Required: Yes The input of the model to run.

api token Type: text Required: Yes Replicate API token

Outputs

outputs Type: list The model predictions.

status Type: text The status of the model run.

duration Type: text The duration of the model run.

Examples

Example 1: ask a question about an image

Inputs:

model: andreasjansson/blip-2:f677695e5e89f8b236e52ecd1d3f01beb44c34606419bcc19345e046d8f786f9

input:

• image ⇒ https://replicate.delivery/pbxt/IJEPmgAlL2zNBNDoRRKFegTEcxnlRhoQxlNjPHSZEy0pSIKn/gg_bridge.jpeg

• question ⇒ what body of water does this bridge cross

api token: 9d************************************** (visit API Tokens page to retrieve the API token)

Outputs:

outputs: san francisco bay

status: succeeded

duration: 1751