process_swedish_text v2

README.md

@@ -8,12 +8,13 @@ sdk_version: 3.12.0
 app_file: app.py
 pinned: false
 license: openrail
+models: ["KBLab/bert-base-swedish-cased-ner"]
 ---
 
 # Swedish Entity Recognition
 
 Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
 
-## REST API via Gradio
+## REST API via Gradio “Use via API” (see page footer)
 
-curl -X POST -H 'Content-type: application/json' --data '{ "data": ["Mrs API"] }' https://tomsoderlund-swedish-entity-recognition.hf.space/run/predict
+    curl -X POST -H 'Content-type: application/json' --data '{ "data": ["Wirwachs malmgård är en på 1770-talet uppförd egendom som ligger på västra Södermalm i Stockholm."] }' https://tomsoderlund-swedish-entity-recognition.hf.space/run/predict

app.py

@@ -1,7 +1,16 @@
 import gradio as gr
 
-def greet(name):
-  return "Hello " + name + "!!"
+# Models from https://huggingface.co/models
+# https://huggingface.co/KBLab/bert-base-swedish-cased-ner
+ml_model = 'KBLab/bert-base-swedish-cased-ner'
+ner_pipeline = pipeline(model=ml_model, task='ner')
 
-iface = gr.Interface(fn=greet, inputs="text", outputs="text")
-iface.launch()
+def process_swedish_text(text):
+  pipeline_results = ner_pipeline(text)
+  print('NER results:', pipeline_results)
+  pipeline_results_adjusted = map(lambda entity: entity | { 'score': float(entity['score']) }, pipeline_results)
+  print(pipeline_results_adjusted)
+  return pipeline_results_adjusted
+
+gradio_interface = gr.Interface(fn=process_swedish_text, inputs="text", outputs="JSON")
+gradio_interface.launch()