rel_component

🪐 Weasel Project: Example project of creating a novel nlp component to do relation extraction from scratch.

This example project shows how to implement a spaCy component with a custom Machine Learning model, how to train it with and without a transformer, and how to apply it on an evaluation dataset.

📋 project.yml

The project.yml defines the data assets required by the project, as well as the available commands and workflows. For details, see the Weasel documentation.

⏯ Commands

The following commands are defined by the project. They can be executed using weasel run [name]. Commands are only re-run if their inputs have changed.

Command	Description
data	Parse the gold-standard annotations from the Prodigy annotations.
train_cpu	Train the REL model on the CPU and evaluate on the dev corpus.
train_gpu	Train the REL model with a Transformer on a GPU and evaluate on the dev corpus.
evaluate	Apply the best model to new, unseen text, and measure accuracy at different thresholds.
clean	Remove intermediate files to start data preparation and training from a clean slate.

⏭ Workflows

The following workflows are defined by the project. They can be executed using weasel run [name] and will run the specified commands in order. Commands are only re-run if their inputs have changed.

Workflow	Steps
all	data → train_cpu → evaluate
all_gpu	data → train_gpu → evaluate

🗂 Assets

The following assets are defined by the project. They can be fetched by running weasel assets in the project directory.

File	Source	Description
assets/annotations.jsonl	Local	Gold-standard REL annotations created with Prodigy