Update app.py

app.py

@@ -1,32 +1,31 @@
 from fastapi import FastAPI
 from pydantic import BaseModel
-import joblib
 from sentence_transformers import SentenceTransformer
 from huggingface_hub import hf_hub_download
+import xgboost as xgb
 
 # -----------------------------
-# Download models from Hugging Face Hub
-# (replace repo_ids with your actual repos)
+# Helper: Load XGBoost Booster (.json)
 # -----------------------------
-griot_path = hf_hub_download(repo_id="mjpsm/Griot-xgb-model", filename="griot_xgb_regression_model_updated_parameters.pkl")
-kinara_path = hf_hub_download(repo_id="mjpsm/Kinara-xgb-model", filename="Kinara_xgb_model.pkl")
-ubuntu_path = hf_hub_download(repo_id="mjpsm/Ubuntu-xgb-model", filename="Ubuntu_xgb_model.pkl")
-jali_path   = hf_hub_download(repo_id="mjpsm/Jali-xgb-model",   filename="Jali_xgb_model.pkl")
-kuumba_path = hf_hub_download(repo_id="mjpsm/Kuumba-xgb-model", filename="Kuumba_xgb_model.pkl")
-
+def load_xgb_model(repo_id: str, filename: str):
+    path = hf_hub_download(repo_id=repo_id, filename=filename)
+    booster = xgb.Booster()
+    booster.load_model(path)
+    return booster
 
+# -----------------------------
+# Load Soulprint models (all JSON now)
+# -----------------------------
 available_models = {
-    "Griot": joblib.load(griot_path),
-    "Kinara": joblib.load(kinara_path),
-    "Ubuntu": joblib.load(ubuntu_path),
-    "Jali": joblib.load(jali_path),
-    "Kuumba": joblib.load(kuumba_path)
-
+    "Griot":   load_xgb_model("mjpsm/Griot-xgb-model",   "Griot_xgb_model.json"),
+    "Kinara":  load_xgb_model("mjpsm/Kinara-xgb-model",  "Kinara_xgb_model.json"),
+    "Ubuntu":  load_xgb_model("mjpsm/Ubuntu-xgb-model",  "Ubuntu_xgb_model.json"),
+    "Jali":    load_xgb_model("mjpsm/Jali-xgb-model",    "Jali_xgb_model.json"),
+    "Kuumba":  load_xgb_model("mjpsm/Kuumba-xgb-model",  "Kuumba_xgb_model.json"),
+    "Sankofa": load_xgb_model("mjpsm/Sankofa-xgb-model", "Sankofa_xgb_model.json"),
 }
 
-
-
-# Archetype list (15 total, 12 fillers for now)
+# Archetype list (15 total, placeholders for now)
 all_archetypes = [
     "Griot", "Kinara", "Ubuntu", "Jali", "Sankofa", "Imani", "Maji",
     "Nzinga", "Bisa", "Zamani", "Tamu", "Shujaa", "Ayo", "Ujamaa", "Kuumba"
@@ -43,17 +42,17 @@ class TextInput(BaseModel):
 
 @app.post("/soulprint_snapshot")
 def soulprint_snapshot(input: TextInput):
-    # Convert text into embedding
     embedding = embedder.encode([input.text]).reshape(1, -1)
 
-    # Build snapshot
     snapshot = {}
     for name in all_archetypes:
         if name in available_models:
-            score = available_models[name].predict(embedding)[0]
+            dmatrix = xgb.DMatrix(embedding)
+            score = available_models[name].predict(dmatrix)[0]
             snapshot[name] = float(score)
         else:
-            snapshot[name] = 0.0  # filler until model is ready
+            snapshot[name] = 0.0  # placeholder until model is trained
 
     return {"soulprint_snapshot": snapshot}
 
+