first model

README.md

@@ -0,0 +1,2 @@
+# Simple Classifier
+A toy model that classifies 10-feature input into positive/negative.

model.py

@@ -0,0 +1,10 @@
+import torch
+from torch import nn
+
+class SimpleClassifier(nn.Module):
+    def __init__(self):
+        super().__init__()
+        self.linear = nn.Linear(10, 2)  # input 10 features → 2 classes
+
+    def forward(self, x):
+        return self.linear(x)

pipeline_model.py

@@ -0,0 +1,16 @@
+import torch
+from model import SimpleClassifier
+
+class HFModel:
+    def __init__(self):
+        self.model = SimpleClassifier()
+        self.model.load_state_dict(torch.load("simple_model.pt"))
+        self.model.eval()
+
+    def __call__(self, inputs):
+        # Expect a list of lists with 10 numbers
+        X = torch.tensor(inputs, dtype=torch.float32)
+        with torch.no_grad():
+            logits = self.model(X)
+        preds = torch.argmax(logits, dim=1).tolist()
+        return [{"label": "positive" if p == 1 else "negative", "score": float(logits[i][p])} for i, p in enumerate(preds)]

simple_model.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d51b0dd95f88f2849ac0cb9fb4e0321298fa9439409cbc5a106d825e2632977a
+size 2061

train.py

@@ -0,0 +1,21 @@
+import torch
+from torch import nn, optim
+from model import SimpleClassifier
+
+# Dummy data (10 features → 2 classes)
+X = torch.randn(100, 10)
+y = torch.randint(0, 2, (100,))
+
+model = SimpleClassifier()
+criterion = nn.CrossEntropyLoss()
+optimizer = optim.Adam(model.parameters())
+
+for epoch in range(10):  # tiny training loop
+    optimizer.zero_grad()
+    out = model(X)
+    loss = criterion(out, y)
+    loss.backward()
+    optimizer.step()
+    print(f"Epoch {epoch}: loss={loss.item():.4f}")
+
+torch.save(model.state_dict(), "simple_model.pt")