refactor: Replace custom model loading with Gradio Client

Simplified the code by using Gradio Client to interact with the model hosted on 'umutbozdag/humanizer_model' Space. This removes the need for manual model loading and preprocessing, reducing potential points of failure and streamlining text generation.

app.py

@@ -1,42 +1,16 @@
 import gradio as gr
-from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
-from huggingface_hub import HfApi
+from gradio_client import Client
 
-# Get the latest model from your space
-api = HfApi()
-space_name = "umut-bozdag/humanizer_model"  # Replace with your actual space name
-model_files = api.list_repo_files(space_name)
-model_file = next(file for file in model_files if file.endswith('.bin'))
-model_revision = api.get_repo_info(space_name).sha
-
-# Load the model and tokenizer from the space
-tokenizer = AutoTokenizer.from_pretrained(space_name, revision=model_revision)
-model = AutoModelForSeq2SeqLM.from_pretrained(space_name, revision=model_revision)
+# Create a client to interact with your Space
+client = Client("umutbozdag/humanizer_model")
 
 def generate_text(input_text):
-    # Preprocess input text
-    input_text = input_text.strip()
-    
-    # Prepare input for the model
-    input_ids = tokenizer.encode(input_text, return_tensors="pt", max_length=256, truncation=True)
-    
-    # Generate text with parameters matching your training setup
-    outputs = model.generate(
-        input_ids,
-        max_length=256,
-        num_return_sequences=1,
-        no_repeat_ngram_size=2,
-        top_k=30,
-        top_p=0.9,
-        temperature=0.7,
-        do_sample=True,
-        early_stopping=True,
-        num_beams=4
+    # Use the client to predict using your Space
+    result = client.predict(
+        input_text,  # str in 'Input Text' Textbox component
+        api_name="/predict"
     )
-    
-    # Decode and clean up the generated text
-    generated_text = tokenizer.decode(outputs[0], skip_special_tokens=True)
-    return generated_text.strip()
+    return result
 
 iface = gr.Interface(
     fn=generate_text,