Token Classification
GLiNER
PyTorch
multilingual
bert
Gliner_haystack / handler.py
Rejebc's picture
Update handler.py
39b1f14 verified
raw
history blame
1.26 kB
from typing import Dict, List, Any
from gliner import GLiNER
class EndpointHandler:
def __init__(self, path=""):
# Initialize the GLiNER model
self.model = GLiNER.from_pretrained("urchade/gliner_multi-v2.1")
def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
"""
Args:
data (Dict[str, Any]): The input data including:
- "inputs": The text input from which to extract information.
- "labels": The labels to predict entities for.
Returns:
List[Dict[str, Any]]: The extracted entities from the text, formatted as required.
"""
# Get inputs and labels
inputs = data.get("inputs", "")
labels = ["party", "document title"]
print('labels',labels)
# Predict entities using GLiNER
entities = self.model.predict_entities(inputs, labels)
# Format the results to match the expected output structure
formatted_results = []
for entity in entities:
formatted_entity = {
entity["label"]: entity["text"],
}
print(formatted_entity)
formatted_results.append(formatted_entity)
return formatted_results