This repository is an example for fine-tuning models on named entity recognition task.
The base model used is knowledgator/gliner-multitask-large-v0.5.
The fine-tuning data is expected to be in the following format:
[
{
"text": "...", // this is the input example
"entities": [
{
"entity": "...", // this refers to the entity (label) name in the input
"word": "..." // this refers to the corresponding exact entity (label) string in the input
},
...
]
},
...
]The model will be fine-tuned on the task to get the "text" as input and return the "entities" as output.
The process is started by:
python train.py