integrate RoBERTa model as the new default

README.md

@@ -66,6 +66,7 @@ sentiment-analysis-api/
 │   │   └── transcript_data.py
 │   ├── models/                        # Contains the models for sentiment analysis (Whisper, BERTweet)
 │   │   ├── bertweet_model.py
+│   │   ├── roberta_model.py 
 │   │   └── whisper_model.py
 │   ├── routes/                        # Defines the routes for the API
 │   │   ├── __init__.py

app/data/sentiment_data.py

@@ -3,7 +3,7 @@
 """
 # Model Layer
 from app.models.bertweet_model import BertweetSentiment
-
+from app.models.roberta_model import RoBERTaSentiment
 from app.utils.logger import logger
 
 class SentimentDataLayer:
@@ -16,10 +16,13 @@ def __init__(self, config: dict):
 
         self.config = config.get('sentiment_analysis')
         self.default_model = self.config.get('default_model')
+        print(f"default_model: '{self.default_model}'")  # ← ضيف السطر ده هنا
 
         # Initialize the appropriate model based on the configuration
         if self.default_model == "bertweet":
             self.model = BertweetSentiment(config)
+        elif self.default_model=="roberta":
+            self.model = RoBERTaSentiment(config)
         # elif self.default_model == "another_model":
         #     self.model = AnotherModel(config)  # Replace with your other model class
         else:

app/models/roberta_model.py

@@ -0,0 +1,74 @@
+"""
+This module defines the RoBERTaSentiment class, which is a PyTorch model for sentiment analysis using the RoBERTa model.
+"""
+import torch
+import torch.nn as nn
+
+from transformers import AutoTokenizer, AutoModelForSequenceClassification
+
+# Mapping from RoBERTa labels to standard labels
+LABEL_MAPPING = {
+    "positive": "POS",
+    "neutral": "NEU",
+    "negative": "NEG"
+}
+
+class RoBERTaSentiment(nn.Module):
+    def __init__(self, config: dict) -> None:
+        """
+        Initialize the RoBERTa model for sentiment analysis.
+        :param config: The configuration object containing model and device info.
+        """
+        self.debug = config.get('debug')
+
+        self.config = config.get('sentiment_analysis').get('roberta')
+        self.model_name = self.config.get('model_name')
+        self.device = self.config.get('device')
+
+        super(RoBERTaSentiment, self).__init__()
+
+        # Initialize the Tokenizer
+        self.tokenizer = AutoTokenizer.from_pretrained(self.model_name)
+
+        # Initialize the Model
+        self.model = AutoModelForSequenceClassification.from_pretrained(self.model_name)
+        self.model.to(self.device)
+
+        # Load the model configuration to get class labels
+        self.model_config = self.model.config
+
+        # Get Labels
+        if hasattr(self.model_config, 'id2label'):
+            self.class_labels = [self.model_config.id2label[i] for i in range(len(self.model_config.id2label))]
+        else:
+            self.class_labels = None
+
+    def forward(self, text) -> tuple:
+        """
+        Perform sentiment analysis on the given text.
+
+        Args:
+            text (str): Input text for sentiment analysis.
+
+        Returns:
+            tuple: Model outputs, probabilities, predicted label, and confidence score.
+        """
+        # Tokenize the input text
+        inputs = self.tokenizer(text, return_tensors="pt", truncation=True, padding=True).to(self.device)
+
+        # Forward pass
+        outputs = self.model(**inputs)
+
+        # Convert logits to probabilities
+        probabilities = torch.nn.functional.softmax(outputs.logits, dim=-1)
+
+        # Get the predicted sentiment
+        predicted_class = torch.argmax(probabilities, dim=1).item()
+
+        # Get the corresponding class label
+        raw_label = self.class_labels[predicted_class]
+
+        # Map the label to standard format (POS, NEU, NEG)
+        predicted_label = LABEL_MAPPING.get(raw_label.lower(), raw_label)
+
+        return outputs, probabilities, predicted_label, probabilities[0][predicted_class].item()

config.yaml

@@ -26,15 +26,19 @@ transcription:
 
 # Sentiment Analysis Configuration
 sentiment_analysis:
-  default_model: "bertweet"  # Specify the default sentiment analysis model (e.g., bertweet, another_model)
-  bertweet:                  # Vader-specific configuration
+  default_model: "roberta"  # Changed from bertweet to roberta
+  bertweet:                  # Bertweet-specific configuration
     model_name: "finiteautomata/bertweet-base-sentiment-analysis"
-    device: 'cpu'               # `cpu` for CPU, or `cuda` GPU device
-    # device: 'cuda'               # `cpu` for CPU, or `cuda` GPU device
+    device: 'cpu'
+    # device: 'cuda'
+  roberta:                   # RoBERTa-specific configuration
+    model_name: "cardiffnlp/twitter-roberta-base-sentiment-latest"
+    device: 'cpu'
+    # device: 'cuda'
   # another_model:          # Placeholder for another sentiment analysis model's configuration
   #   api_key: "your_api_key"
   #   endpoint: "https://api.example.com/sentiment"
 
-# AudioTranscriptionSentimentPipeline Configuration
+  # AudioTranscriptionSentimentPipeline Configuration
 audio_transcription_sentiment_pipeline:
   remove_audio: false  # Specify whether to remove audio files after processing