handle variable lengthed experiences when doing policy optimization

dreamer4/dreamer4.py

@@ -82,7 +82,7 @@ class Experience:
     log_probs: tuple[Tensor, Tensor] | None = None
     values: Tensor | None = None
     step_size: int | None = None
-    lens: Tensor | None = None,
+    lens: Tensor | None = None
     agent_index: int = 0
     is_from_world_model: bool = True
 
@@ -2226,6 +2226,15 @@ class DynamicsWorldModel(Module):
 
         returns = calc_gae(rewards, old_values, gamma = self.gae_discount_factor, lam = self.gae_lambda, use_accelerated = self.gae_use_accelerated)
 
+        # handle variable lengths
+
+        max_time = latents.shape[1]
+        is_var_len = exists(experience.lens)
+
+        if is_var_len:
+            lens = experience.lens
+            mask = lens_to_mask(lens, max_time)
+
         # determine whether to finetune entire transformer or just learn the heads
 
         world_model_forward_context = torch.no_grad if only_learn_policy_value_heads else nullcontext
@@ -2291,13 +2300,20 @@ class DynamicsWorldModel(Module):
 
         # handle entropy loss for naive exploration bonus
 
-        entropy_loss = - reduce(entropies, 'b t na -> b t', 'sum').mean()
+        entropy_loss = - reduce(entropies, 'b t na -> b t', 'sum')
 
         total_policy_loss = (
             policy_loss +
             entropy_loss * self.policy_entropy_weight
         )
 
+        # maybe handle variable lengths
+
+        if is_var_len:
+            total_policy_loss = total_policy_loss[mask].mean()
+        else:
+            total_policy_loss = total_policy_loss.mean()
+
         # maybe take policy optimizer step
 
         if exists(policy_optim):
@@ -2316,10 +2332,19 @@ class DynamicsWorldModel(Module):
 
         return_bins = self.reward_encoder(returns)
 
+        value_bins, return_bins, clipped_value_bins = tuple(rearrange(t, 'b t l -> b l t') for t in (value_bins, return_bins, clipped_value_bins))
+
         value_loss_1 = F.cross_entropy(value_bins, return_bins, reduction = 'none')
         value_loss_2 = F.cross_entropy(clipped_value_bins, return_bins, reduction = 'none')
 
-        value_loss = torch.maximum(value_loss_1, value_loss_2).mean()
+        value_loss = torch.maximum(value_loss_1, value_loss_2)
+
+        # maybe variable length
+
+        if is_var_len:
+            value_loss = value_loss[mask].mean()
+        else:
+            value_loss = value_loss.mean()
 
         # maybe take value optimizer step
 

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "dreamer4"
-version = "0.0.76"
+version = "0.0.77"
 description = "Dreamer 4"
 authors = [
     { name = "Phil Wang", email = "lucidrains@gmail.com" }