fix: reset Adam optimizer state between walk-forward folds (prevents Q-value explosion)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

crates/ml/src/cuda_pipeline/gpu_dqn_trainer.rs

@@ -870,6 +870,33 @@ pub struct GpuDqnTrainer {
     cql_d_adv_logits: CudaSlice<f32>,
 }
 
+impl GpuDqnTrainer {
+    /// Reset Adam optimizer state for a new walk-forward fold.
+    /// Zeroes momentum (m), variance (v), and step counter (t).
+    /// Weights are preserved (warm-start for the new fold).
+    pub fn reset_adam_state(&mut self) -> Result<(), MLError> {
+        self.stream.memset_zeros(&mut self.m_buf)
+            .map_err(|e| MLError::ModelError(format!("reset m_buf: {e}")))?;
+        self.stream.memset_zeros(&mut self.v_buf)
+            .map_err(|e| MLError::ModelError(format!("reset v_buf: {e}")))?;
+        self.stream.memset_zeros(&mut self.t_buf)
+            .map_err(|e| MLError::ModelError(format!("reset t_buf: {e}")))?;
+        self.adam_step = 0;
+
+        // Also reset IQN trunk Adam state (currently unused but allocated).
+        self.stream.memset_zeros(&mut self.iqn_trunk_m)
+            .map_err(|e| MLError::ModelError(format!("reset iqn m: {e}")))?;
+        self.stream.memset_zeros(&mut self.iqn_trunk_v)
+            .map_err(|e| MLError::ModelError(format!("reset iqn v: {e}")))?;
+        self.stream.memset_zeros(&mut self.iqn_trunk_t_buf)
+            .map_err(|e| MLError::ModelError(format!("reset iqn t: {e}")))?;
+        self.iqn_trunk_adam_step = 0;
+
+        tracing::info!("Adam optimizer state reset for new fold");
+        Ok(())
+    }
+}
+
 impl Drop for GpuDqnTrainer {
     fn drop(&mut self) {
         // Synchronize stream and destroy graphs BEFORE CudaSlice fields drop.

crates/ml/src/trainers/dqn/fused_training.rs

@@ -584,6 +584,9 @@ impl FusedTrainingCtx {
         self.steps_since_varmap_sync = 0;
         self.last_combined_norm = 0.0;
         self.graph_aux = None;
+        // Reset Adam optimizer state — each fold starts with fresh momentum.
+        // Stale momentum from a previous fold causes weight explosion → Q-value -1e30.
+        self.trainer.reset_adam_state()?;
         Ok(())
     }