changed the model

Components/model_Responce.py

@@ -1,9 +1,11 @@
+import torch
 import pickle
 import joblib
 import numpy as np
 import tensorflow as tf
 from keras.utils import pad_sequences
 from keras.preprocessing.text import Tokenizer
+from transformers import AutoModelForSequenceClassification, AutoTokenizer
 
 # Load the model from the pickle file
 # filename = 'F:/CVFilter/models/model_pk.pkl'
@@ -13,26 +15,48 @@ from keras.preprocessing.text import Tokenizer
 # Load the saved model
 # model = joblib.load('F:\CVFilter\models\model.joblib')
 
-model = tf.keras.models.load_model('models\model.h5')
 
-tokenfile = 'tokenized_words/tokenized_words.pkl'
-# Load the tokenized words from the pickle file
-with open(tokenfile, 'rb') as file:
-    loaded_tokenized_words = pickle.load(file)
+# Load Local Model and Local tokenizer
+# model = tf.keras.models.load_model('models\model.h5')
 
-max_review_length = 200
-tokenizer = Tokenizer(num_words=10000,  #max no. of unique words to keep
-                      filters='!"#$%&()*+,-./:;<=>?@[\]^_`{|}~', 
-                      lower=True #convert to lower case
-                     )
-tokenizer.fit_on_texts(loaded_tokenized_words)
+# tokenfile = 'tokenized_words/tokenized_words.pkl'
+# # Load the tokenized words from the pickle file
+# with open(tokenfile, 'rb') as file:
+#     loaded_tokenized_words = pickle.load(file)
+
+# max_review_length = 200
+# tokenizer = Tokenizer(num_words=10000,  #max no. of unique words to keep
+#                       filters='!"#$%&()*+,-./:;<=>?@[\]^_`{|}~', 
+#                       lower=True #convert to lower case
+#                      )
+# tokenizer.fit_on_texts(loaded_tokenized_words)
+
+
+# Load Huggingface model and tokenizer
+# Define the model name
+model_name = "fazni/distilbert-base-uncased-career-path-prediction"
+
+# Load the model
+model = AutoModelForSequenceClassification.from_pretrained(model_name)
+
+# Load the tokenizer
+tokenizer = AutoTokenizer.from_pretrained(model_name)
 
 outcome_labels = ['Business Analyst', 'Cyber Security','Data Engineer','Data Science','DevOps','Machine Learning Engineer','Mobile App Developer','Network Engineer','Quality Assurance','Software Engineer']
 
 def model_prediction(text, model=model, tokenizer=tokenizer, labels=outcome_labels):
-    seq = tokenizer.texts_to_sequences([text])
-    padded = pad_sequences(seq, maxlen=max_review_length)
-    pred = model.predict(padded)
-    # print("Probability distribution: ", pred)
-    # print("Field ")
-    return labels[np.argmax(pred)]
+    # Local model
+    # seq = tokenizer.texts_to_sequences([text])
+    # padded = pad_sequences(seq, maxlen=max_review_length)
+    # pred = model.predict(padded)
+    # return labels[np.argmax(pred)]
+
+    # Hugging face model
+    # Tokenize the text
+    inputs = tokenizer(text, return_tensors="pt")
+    outputs = model(**inputs)
+    
+    # Get the predicted class probabilities
+    probs = outputs.logits.softmax(dim=-1)
+
+    return labels[torch.argmax(probs)]