Merge remote-tracking branch 'origin/candle-linear-regression'

ml/src/lib.rs

@@ -1,4 +1,5 @@
-use forust_ml::{GradientBooster, Matrix};
+use candle_core::{Device, Result, Tensor};
+use candle_nn::{AdamW, Linear, Module, Optimizer, ParamsAdamW};
 use std::error::Error;
 
 #[no_mangle]
@@ -19,7 +20,7 @@ pub extern "C" fn run() {
     }
 }
 
-fn greet_internal() -> Result<Vec<f64>, Box<dyn Error>> {
+fn greet_internal() -> Result<Vec<f32>, Box<dyn Error>> {
     //     let csv_data = r#"survived,pclass,sex,age,sibsp,parch,fare,embarked,class,who,adult_male,deck,embark_town,alive,alone
     // 0,3,male,22.0,1,0,7.25,S,Third,man,True,,Southampton,no,False
     // 1,1,female,38.0,1,0,71.2833,C,First,woman,False,C,Cherbourg,yes,False
@@ -34,30 +35,60 @@ fn greet_internal() -> Result<Vec<f64>, Box<dyn Error>> {
 -122.46,37.75,26.0,2192.0,438.0,954.0,456.0,4.5352,374200.0,NEAR BAY
 -119.24,36.33,9.0,3289.0,621.0,1866.0,631.0,3.1599,95000.0,INLAND"#;
 
-    let mut data = Vec::new();
-    let mut y = Vec::new();
+    let mut features = Vec::new();
+    let mut target = Vec::new();
 
     for line in csv_data.lines() {
         let values: Vec<&str> = line.split(',').collect();
 
         // Assuming the CSV structure: survived,pclass,age,sibsp,parch,fare
-        y.push(values[8].parse::<f64>().unwrap_or(f64::NAN));
+        target.push(values[8].parse::<f32>().unwrap_or(f32::NAN));
 
-        data.extend_from_slice(
+        features.extend_from_slice(
             &(0..8)
-                .map(|col| values[col].parse::<f64>().unwrap_or(f64::NAN))
+                .map(|col| values[col].parse::<f32>().unwrap_or(f32::NAN))
                 .collect::<Vec<_>>(),
         );
     }
 
-    println!("{data:?}");
-    println!("{y:?}");
+    println!("{features:?}");
+    println!("{target:?}");
 
-    let matrix = Matrix::new(&data, csv_data.lines().count(), 8);
+    let num_samples = features.len();
+    let num_features = 8;
 
-    let mut model = GradientBooster::default().set_learning_rate(0.3);
-    model.fit_unweighted(&matrix, &y, None)?;
+    // Convert data to tensors
+    let features_tensor =
+        Tensor::from_slice(&features, &[num_samples, num_features], &Device::Cpu)?;
+    let target_tensor = Tensor::from_slice(&target, &[num_samples, 1], &Device::Cpu)?;
+
+    // Step 6: Define the linear regression model
+    let mut model = Linear::new(num_features, 1);
+
+    // Step 7: Set up the optimizer
+    let mut optimizer = AdamW::new(
+        vec![],
+        ParamsAdamW::default(), // OptimizerConfig::adam(0.01).build(model.parameters()),
+    );
+
+    // Step 8: Training Loop
+    let num_epochs = 100;
+    for epoch in 0..num_epochs {
+        let predictions = model.forward(&features_tensor)?;
+        let loss = mse_loss(&predictions, &target_tensor)?;
+        optimizer.step(&loss)?;
+
+        if epoch % 10 == 0 {
+            println!("Epoch {}: Loss = {:?}", epoch, loss);
+        }
+    }
+
+    // Step 9: Make and print predictions
+    let predictions = model.forward(&features_tensor)?;
+    println!(
+        "Model predictions (first 10): {:?}",
+        predictions.slice([..10])
+    );
 
-    let predictions = model.predict(&matrix, true);
     Ok(predictions)
 }

zk/Cargo.toml

@@ -5,6 +5,7 @@ edition = "2021"
 
 [dependencies]
 anyhow = "1.0.86"
+ff = "0.13"
 rayon = "1.10.0"
 tracing = "0.1.40"
-zk-engine = { git = "https://github.com/ICME-Lab/zkEngine_dev.git" }
+zk-engine = { git = "https://github.com/ICME-Lab/zkEngine_dev.git", branch = "main" }