better causal-ness for split loss calc, and also do masking for NAR-len for it

vall_e/models/base.py

@@ -1445,15 +1445,30 @@ class Base(nn.Module):
 		batch_size = len( inputs )
 
 		for i, batch in enumerate( inputs ):
-			quant_level = quant_levels[i]
-
 			it = 0
-
+			quant_level = quant_levels[i]
 			task_name = None
+
+			causal = (quant_level == 0 and "ar" in self.capabilities) or ("nar" not in self.capabilities)
+			dropout_mask = None
 			for name, input in batch:
-				# do not use resp
+				if name == "dropout_mask":
+					dropout_mask = input
+
+			for name, input in batch:
+				# meta-input, no corresponding token at the moment
+				if name == "task":
+					task_name = input
+					if task_type in ["len", "stt"]:
+						causal = True
+					continue
+				# do not use resp as-is
 				if name == "resp":
-					if self.interleave:
+					if dropout_mask is not None:
+						# if mask use original token, else ignore
+						causal = False
+						target.append( torch.where( dropout_mask, input if input.dim() == 1 else input[:, 0], self.ignore_index ) )
+					elif self.interleave:
 						input = _interleave_sequence_flatten( [ input[:, l] for l in range( input.shape[-1] ) ] )
 					elif task_type in summed_embeddings_task:
 						input = torch.full_like(input[..., 0], self.ignore_index)
@@ -1463,17 +1478,11 @@ class Base(nn.Module):
 				elif name == "prom":
 					proms = [ input ] if isinstance(input, torch.Tensor) else input
 					input = torch.cat( [ prompt_input_to_token( input, quant_level ) for input in proms ] )
-				# meta-input, no corresponding token at the moment
-				elif name == "task":
-					task_name = input
-					continue
 
 				seq_len = input.shape[0]
 
 				logit = logits[i][it:it+seq_len]
 				it += seq_len + 1 # +1 to incorporate the separator
-
-				causal = (quant_level == 0 and "ar" in self.capabilities) or ("nar" not in self.capabilities) or (task_name in ["len", "stt"])				
 				
 				# for the AR, shift sequence so that it predicts the next token
 				#	 (the NAR predicts the next token in place, so it's not necessary to do any modifications for it)