Update README.md

README.md

@@ -1,3 +1,15 @@
----
-license: mit
----
+ONNX model - a fine tuned version of DistilBERT which can be used to classify text as one of:
+- neutral, offensive_language, harmful_behaviour, hate_speech
+
+The model was trained using the [csfy tool](https://github.com/mrseanryan/csfy) and the dataset [seanius/toxic-or-neutral-text-labelled](https://huggingface.co/datasets/seanius/toxic-or-neutral-text-labelled)
+
+The base model is required (distilbert-base-uncased)
+
+For an example of how to run the model, see the [csfy tool](https://github.com/mrseanryan/csfy).
+
+The output is a number indicating the class - you can decode that via the label_mapping.json file.
+
+
+---
+license: mit
+---