Update app.py

app.py

@@ -1,45 +1,26 @@
-from tensorflow.keras.models import load_model
-import json
-import gradio as gr
-
-# Load the pre-trained Keras model (replace with actual model path if needed)
-model = load_model("saved_model.pb")
+from transformers import AutoTokenizer, TFBertForSeq2SeqLM  # Assuming TFBert model
 
 # Load tokenizer configurations
-with open("source_tokenizer_config.json", "r") as f:
-  source_tokenizer_config = json.load(f)
-
-with open("target_tokenizer_config.json", "r") as f:
-  target_tokenizer_config = json.load(f)
-
-def predict(text):
-  # Preprocess text (e.g., tokenization) based on your Keras model's requirements
-  # Here's an example assuming your model expects preprocessed text (modify as needed)
-  preprocessed_text = process_text(text, source_tokenizer_config)  # Implement your preprocessing logic
-  predictions = model.predict(np.array([preprocessed_text]))
-
-  # Postprocess predictions (e.g., decoding) based on your Keras model's output format
-  decoded_text = process_predictions(predictions[0], target_tokenizer_config)  # Implement your postprocessing logic
-  return decoded_text
-
-# Define preprocessing and postprocessing functions based on your Keras model (replace with your logic)
-def process_text(text, tokenizer_config):
-  # ... Your preprocessing steps (e.g., tokenization, padding) ...
-  return processed_text
-
-def process_predictions(predictions, tokenizer_config):
-  # ... Your postprocessing steps (e.g., argmax, converting indices to tokens) ...
-  return decoded_text
-
-# Create the Gradio interface
-interface = gr.Interface(
-  fn=predict,
-  inputs=gr.Textbox(label="Enter Malayalam Text"),
-  outputs=gr.Textbox(label="Predicted English Text"),
-  title="Malayalam to English Transliteration",
-  description="Enter Malayalam text in the box and click 'Predict' to get the English transliteration.",
-  thumbnail="thumbnail.jpg"  # Optional: Add a thumbnail image (upload to Space)
+source_tokenizer = AutoTokenizer.from_pretrained("https://huggingface.co/Bajiyo/mal_en_transliteration/tree/main/source_tokenizer_config.json")
+target_tokenizer = AutoTokenizer.from_pretrained("https://huggingface.co/Bajiyo/mal_en_transliteration/tree/main/target_tokenizer_config.json")
+
+# Load the model (replace with your actual model path)
+model = TFBertForSeq2SeqLM.from_pretrained("https://huggingface.co/Bajiyo/mal_en_transliteration/tree/main/transliteration_model.h5")
+
+def translate(malayalam_text):
+    """Function to perform Malayalam to English transliteration"""
+    source_ids = source_tokenizer(malayalam_text, return_tensors="pt")["input_ids"]
+    translated_tokens = model.generate(**source_ids)
+    english_text = target_tokenizer.batch_decode(translated_tokens, skip_special_tokens=True)[0]
+    return english_text
+
+interface = gradio.Interface(
+    fn=translate,
+    inputs="text",
+    outputs="text",
+    title="Malayalam to English Transliteration",
+    description="Enter Malayalam text to get the English transliteration.",
+    examples=[["എങ്ങനെയാണ് ഞാൻ ഇംഗ്ലീഷിൽ സംസാരിക്കേണ്ടത്?"], ["ഹലോ എങ്ങനെയിരിക്കുന്നു?"]]
 )
 
-# Launch the interface
 interface.launch()