update

app.py

@@ -5,7 +5,6 @@ import torch, numpy as np, random
 torch.manual_seed(42)
 np.random.seed(42)
 random.seed(42)
-
 # =========================
 # 1. SAFE DOWNLOAD FUNCTION (FIXED)
 # =========================
@@ -229,6 +228,153 @@ Xa_test = sc_a.transform(Xa_test)
 Xv_test = sc_v.transform(Xv_test)
 
 # =========================
-# 11. MODELS + TRAIN + EVAL (UNCHANGED)
+# 11. MODELS
+# =========================
+import torch.nn as nn
+
+class Model(nn.Module):
+    def __init__(self, vdim):
+        super().__init__()
+        self.t = nn.Sequential(nn.Linear(768,128), nn.ReLU())
+        self.a = nn.Sequential(nn.Linear(40,32), nn.ReLU())
+        self.v = nn.Sequential(nn.Linear(vdim,64), nn.ReLU())
+
+        self.f = nn.Sequential(
+            nn.Linear(224,64),
+            nn.ReLU(),
+            nn.Dropout(0.3),
+            nn.Linear(64,1)
+        )
+
+    def forward(self, t, a, v):
+        return self.f(torch.cat([self.t(t), self.a(a), self.v(v)], dim=1))
+
+
+class AttentionFusionModel(nn.Module):
+    def __init__(self, vdim):
+        super().__init__()
+
+        self.t = nn.Sequential(nn.Linear(768,128), nn.ReLU())
+        self.a = nn.Sequential(nn.Linear(40,32), nn.ReLU())
+        self.v = nn.Sequential(nn.Linear(vdim,64), nn.ReLU())
+
+        self.attn = nn.Sequential(
+            nn.Linear(224,64),
+            nn.Tanh(),
+            nn.Linear(64,3)
+        )
+
+        self.f = nn.Sequential(
+            nn.Linear(224,64),
+            nn.ReLU(),
+            nn.Dropout(0.3),
+            nn.Linear(64,1)
+        )
+
+    def forward(self, t, a, v):
+        t_feat = self.t(t)
+        a_feat = self.a(a)
+        v_feat = self.v(v)
+
+        combined = torch.cat([t_feat, a_feat, v_feat], dim=1)
+
+        weights = torch.softmax(self.attn(combined), dim=1)
+
+        fused = torch.cat([
+            weights[:,0:1] * t_feat,
+            weights[:,1:2] * a_feat,
+            weights[:,2:3] * v_feat
+        ], dim=1)
+
+        return self.f(fused)
+
+
+# =========================
+# 12. CONVERT TO TENSORS
+# =========================
+Xt = torch.tensor(Xt, dtype=torch.float32).to(device)
+Xa = torch.tensor(Xa, dtype=torch.float32).to(device)
+Xv = torch.tensor(Xv, dtype=torch.float32).to(device)
+yt = torch.tensor(y_train, dtype=torch.float32).to(device)
+
+Xt_test = torch.tensor(Xt_test, dtype=torch.float32).to(device)
+Xa_test = torch.tensor(Xa_test, dtype=torch.float32).to(device)
+Xv_test = torch.tensor(Xv_test, dtype=torch.float32).to(device)
+
+
+# =========================
+# 13. TRAIN BASELINE MODEL
+# =========================
+baseline_model = Model(Xv.shape[1]).to(device)
+
+opt1 = torch.optim.Adam(baseline_model.parameters(), lr=1e-4)
+loss_fn = nn.BCEWithLogitsLoss()
+
+print("\nTraining Baseline Model...")
+
+for e in range(5):   # 🔥 reduced epochs (important)
+    baseline_model.train()
+    opt1.zero_grad()
+
+    outputs = baseline_model(Xt, Xa, Xv).squeeze()
+    loss = loss_fn(outputs, yt)
+
+    loss.backward()
+    opt1.step()
+
+    print(f"Epoch {e+1}, Loss: {loss.item():.4f}")
+
+
+# =========================
+# 14. TRAIN ATTENTION MODEL
 # =========================
-# (keep exactly as your current code)
+attention_model = AttentionFusionModel(Xv.shape[1]).to(device)
+
+opt2 = torch.optim.AdamW(attention_model.parameters(), lr=1e-4)
+loss_fn_attn = nn.BCEWithLogitsLoss()
+
+print("\nTraining Attention Model...")
+
+for e in range(5):   # 🔥 reduced epochs
+    attention_model.train()
+    opt2.zero_grad()
+
+    outputs = attention_model(Xt, Xa, Xv).squeeze()
+    loss = loss_fn_attn(outputs, yt)
+
+    loss.backward()
+    opt2.step()
+
+    print(f"Epoch {e+1}, Loss: {loss.item():.4f}")
+
+
+# =========================
+# 15. EVALUATION
+# =========================
+from sklearn.metrics import accuracy_score, f1_score
+
+baseline_model.eval()
+attention_model.eval()
+
+with torch.no_grad():
+    pred_baseline = (torch.sigmoid(
+        baseline_model(Xt_test, Xa_test, Xv_test).squeeze()
+    ) > 0.5).int().cpu().numpy()
+
+    pred_attention = (torch.sigmoid(
+        attention_model(Xt_test, Xa_test, Xv_test).squeeze()
+    ) > 0.5).int().cpu().numpy()
+
+
+# =========================
+# 16. RESULTS
+# =========================
+print("\n===== MODEL COMPARISON =====")
+
+print("\nBaseline Model:")
+print("Accuracy:", accuracy_score(y_test, pred_baseline))
+print("F1 Score:", f1_score(y_test, pred_baseline))
+
+print("\nAttention Model:")
+print("Accuracy:", accuracy_score(y_test, pred_attention))
+print("F1 Score:", f1_score(y_test, pred_attention))