feat: add scalers and model weights

hypothesis.py

@@ -23,9 +23,9 @@ class BaseModelHypothesis:
         self.features_not_normalized = []
 
         self.scaler_normalized_text_length = joblib.load(
-            "scaler-normalized-text-length.joblib")
+            "scalers/scaler-normalized-text-length.joblib")
         self.scaler_not_normalized = joblib.load(
-            "scaler-not-normalized.joblib")
+            "scalers/scaler-not-normalized.joblib")
 
     def process_emotion_lexicon(self):
         emotion_lexicon = {}

main_model.py

@@ -50,7 +50,7 @@ class PredictMainModel:
         self.model = AlbertCustomClassificationHead(
             self.albert_model).to(self.device)
         # TODO : CHANGE MODEL STATE DICT PATH
-        self.model.load_state_dict(torch.load("best_model_fold_4.pth"))
+        self.model.load_state_dict(torch.load("models/albert_model.pth"))
 
     def preprocess_input(self, text: str, additional_features: np.ndarray):
         encoding = self.tokenizer.encode_plus(

models/albert_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6b203b54caba10e290830cf720c8cd35093a358792e996ee04ee7d8e5e341651
+size 70752187

random_forest_model.py

@@ -5,8 +5,8 @@ from typing import List
 
 class RandomForestModel:
     def __init__(self):
-        self.scaler = joblib.load("rf_scaler.joblib")
-        self.model = joblib.load("random_forest.joblib")
+        self.scaler = joblib.load("scalers/rf_scaler.joblib")
+        self.model = joblib.load("models/random_forest.joblib")
 
     def preprocess_input(self, secondary_model_features: List[float]) -> np.ndarray:
         return self.scaler.transform(np.array(secondary_model_features).astype(np.float32).reshape(1, -1))

requirements.txt

@@ -5,4 +5,5 @@ pandas
 textstat
 scikit-learn==1.4.1.post1
 transformers
-fastapi
+fastapi
+uvicorn