convert_bert.py

import tensorflow as tf
from transformers import AutoTokenizer, TFAutoModelForSequenceClassification
import os

model_name = "mrm8488/bert-tiny-finetuned-sms-spam-detection"

print("Loading model and tokenizer...")
tokenizer = AutoTokenizer.from_pretrained(model_name)
model = TFAutoModelForSequenceClassification.from_pretrained(model_name, from_pt=True)

print("Exporting vocab...")
os.makedirs("models", exist_ok=True)
with open("models/vocab.txt", "w", encoding="utf-8") as f:
    for token, index in sorted(tokenizer.get_vocab().items(), key=lambda x: x[1]):
        f.write(f"{token}\n")

print("Tracing model for TFLite conversion...")
MAX_SEQ_LEN = 128

callable_model = tf.function(
    lambda input_ids, attention_mask: model(input_ids=input_ids, attention_mask=attention_mask, training=False)
)

concrete_func = callable_model.get_concrete_function(
    tf.TensorSpec([1, MAX_SEQ_LEN], tf.int32, name="input_ids"),
    tf.TensorSpec([1, MAX_SEQ_LEN], tf.int32, name="attention_mask")
)

print("Converting to TFLite...")
converter = tf.lite.TFLiteConverter.from_concrete_functions([concrete_func])
converter.optimizations = [tf.lite.Optimize.DEFAULT]
converter.target_spec.supported_ops = [
    tf.lite.OpsSet.TFLITE_BUILTINS,
    tf.lite.OpsSet.SELECT_TF_OPS
]

tflite_model = converter.convert()

with open("models/scam_classifier.tflite", "wb") as f:
    f.write(tflite_model)

print("Done! Model saved to models/scam_classifier.tflite")