Inference Pad Predefined example:

Pick a Model Hub to use: ℹ️ If you want to use Hugging Face, start Firefox with MOZ_ALLOW_EXTERNAL_ML_HUB=1

Task:

Model id: ℹ️ The model needs to be compatible with Transformers.js

Model Revision: ℹ️ This is typically the branch of the model repository.

Quantization:

Device:

Number of runs: ℹ️ Number of times to run

Number of threads: ℹ️ 1 means no multi-threading

Backend:

Timeout:

Input data: ℹ️ Keep the JSON valid, with the provided keys.

Downloads

Console

HTTP Inference Pad

HTTP endpoint:

Model: ℹ️ Some endpoints require a specific model.

Bearer token: ℹ️ Some endpoints require a token for access.

Prompt:

HTTP Inference Data

Context

Output

Backend: