foxhunt/ml/examples/test_ppo_fix.rs

// Minimal test to verify PPO minibatch_size fix
// Run with: cargo run --example test_ppo_fix --features cuda

use ml::hyperopt::adapters::ppo::PPOParams;
use ml::hyperopt::traits::ParameterSpace;

fn main() {
    println!("Testing PPO minibatch_size parameter integration...\n");

    // Test 1: Roundtrip for each valid divisor
    let valid_divisors = [64, 128, 256, 512, 1024, 2048];
    for &size in &valid_divisors {
        let params = PPOParams {
            policy_learning_rate: 1e-5,
            value_learning_rate: 1e-4,
            clip_epsilon: 0.2,
            value_loss_coeff: 1.0,
            entropy_coeff: 0.01,
            minibatch_size: size,
        };

        let continuous = params.to_continuous();
        let recovered = PPOParams::from_continuous(&continuous).expect("Failed to recover params");

        assert_eq!(
            recovered.minibatch_size, size,
            "Roundtrip failed for minibatch_size={}",
            size
        );
        println!("✓ Roundtrip test passed for minibatch_size={}", size);
    }

    // Test 2: Verify discrete sampling
    println!("\nTesting discrete sampling from continuous space...");

    for idx in 0..=5 {
        let continuous = vec![
            1e-5_f64.ln(), // policy_learning_rate
            1e-4_f64.ln(), // value_learning_rate
            0.2,           // clip_epsilon
            1.0,           // value_loss_coeff
            0.01_f64.ln(), // entropy_coeff
            idx as f64,    // minibatch_size index
        ];

        let params = PPOParams::from_continuous(&continuous).expect("Failed to parse params");
        let expected = valid_divisors[idx];

        assert_eq!(
            params.minibatch_size, expected,
            "Index {} should map to {}",
            idx, expected
        );
        println!("✓ Index {} -> minibatch_size={}", idx, expected);
    }

    // Test 3: Verify bounds
    println!("\nTesting bounds...");
    let bounds = PPOParams::continuous_bounds();
    assert_eq!(bounds.len(), 6, "Should have 6 parameters");
    assert_eq!(bounds[5], (0.0, 5.0), "Minibatch index should be [0, 5]");
    println!("✓ Bounds test passed: {:?}", bounds[5]);

    // Test 4: Verify parameter names
    println!("\nTesting parameter names...");
    let names = PPOParams::param_names();
    assert_eq!(names.len(), 6, "Should have 6 parameter names");
    assert_eq!(
        names[5], "minibatch_size",
        "6th parameter should be 'minibatch_size'"
    );
    println!("✓ Parameter names test passed: {}", names[5]);

    println!("\n✅ ALL TESTS PASSED! Bug #1 fix verified.");
    println!("\nSummary:");
    println!("  - File: ml/src/hyperopt/adapters/ppo.rs line 385");
    println!("  - Fix: Changed `mini_batch_size: 512` to `mini_batch_size: params.minibatch_size`");
    println!("  - Impact: Hyperopt now correctly samples minibatch_size from [64, 128, 256, 512, 1024, 2048]");
}