Upload 6 files

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+aqi_predictor.flax filter=lfs diff=lfs merge=lfs -text

app.py

@@ -0,0 +1,41 @@
+# app.py
+from fastapi import FastAPI, Request
+import joblib
+import pickle
+import jax
+import jax.numpy as jnp
+from flax import serialization
+from model import AQIPredictor
+
+app = FastAPI()
+
+# Load the scaler and feature list
+scaler = joblib.load("scaler.pkl")
+features = pickle.load(open("features.pkl", "rb"))
+
+# Initialize model
+model = AQIPredictor(features=len(features))
+dummy_input = jnp.ones((1, len(features), 1))  # Shape for Conv1D
+params = model.init(jax.random.PRNGKey(0), dummy_input, deterministic=True)
+
+# Load trained model parameters
+with open("predictor.flax", "rb") as f:
+    params = serialization.from_bytes(params, f.read())
+
+@app.get("/")
+def root():
+    return {"message": "AQI Predictor API is live."}
+
+@app.post("/predict")
+async def predict(request: Request):
+    try:
+        data = await request.json()
+        input_data = [data.get(f, 0.0) for f in features]
+        scaled = scaler.transform([input_data])  # (1, N)
+        reshaped = jnp.array(scaled).reshape(1, len(features), 1)  # For Conv1D input
+
+        prediction = model.apply(params, reshaped, deterministic=True)
+        return {"predicted_aqi": float(prediction[0, 0])}
+
+    except Exception as e:
+        return {"error": str(e)}

aqi_features.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e74e100407793cc5a1a85de4bea0f51a4d3888eaa61ad65d63cb84b4fcc50323
+size 78

aqi_predictor.flax

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:477157c1257397ae75189d693085047a484a0388a3b703f938701743b8147673
+size 123048

aqi_scaler.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:228e6b2bb20ae057103bac75c9498db2c7eee7c58f36f326854a6812e82d66ea
+size 655

model.py

@@ -0,0 +1,21 @@
+# model.py
+import flax.linen as nn
+import jax.numpy as jnp
+
+class AQIPredictor(nn.Module):
+    features: int
+
+    @nn.compact
+    def __call__(self, x, deterministic: bool):
+        x = nn.Conv(features=64, kernel_size=(3,))(x)
+        x = nn.relu(x)
+        x = nn.LayerNorm()(x)
+        x = nn.Conv(features=64, kernel_size=(3,))(x)
+        x = nn.relu(x)
+        x = nn.LayerNorm()(x)
+        x = jnp.mean(x, axis=1)
+        x = nn.Dense(128)(x)
+        x = nn.Dropout(0.1)(nn.silu(x), deterministic=deterministic)
+        x = nn.Dense(64)(x)
+        x = nn.silu(x)
+        return nn.Dense(1)(x)

requirements.txt

@@ -0,0 +1,6 @@
+fastapi
+uvicorn
+jax
+flax
+scikit-learn
+joblib