feat: gibberish detection

app.py

@@ -1,7 +1,50 @@
 import gradio as gr
+import spaces
+import torch
+from transformers import pipeline
+import datetime
+import json
+import logging
 
-def greet(name):
-    return "Hello " + name + "!!"
+model_path = "madhurjindal/autonlp-Gibberish-Detector-492513457"
+# Load model for first time cache 
+gibberish_detection_task = pipeline("text-classification", model=model_path, tokenizer=model_path)
 
-demo = gr.Interface(fn=greet, inputs="text", outputs="text")
-demo.launch()
+@spaces.GPU
+def classify(query):
+    torch_device = 0 if torch.cuda.is_available() else -1
+    tokenizer_kwargs = {'truncation':True,'max_length':512}
+
+    gibberish_detection_task = pipeline("text-classification", model=model_path, tokenizer=model_path, device=torch_device)
+
+    request_type = type(query)
+    try:
+        data = json.loads(query)
+        if type(data) != list:
+            data = [query]
+        else:
+            request_type = type(data)
+    except Exception as e:
+        print(e)
+        data = [query]
+        pass
+
+    start_time = datetime.datetime.now()
+    
+    result = gibberish_detection_task(data, batch_size=128, top_k=4, **tokenizer_kwargs)
+
+    end_time = datetime.datetime.now()
+    elapsed_time = end_time - start_time
+
+    logging.debug("elapsed predict time: %s", str(elapsed_time))
+    print("elapsed predict time:", str(elapsed_time))
+    
+    output = {}
+    output["time"] = str(elapsed_time)
+    output["device"] = torch_device
+    output["result"] = result
+
+    return json.dumps(output)
+
+demo = gr.Interface(fn=classify, inputs=["text"], outputs="text")
+demo.launch()