actually use langs from the dataloader

vall_e/models/ar.py

@@ -102,6 +102,7 @@ class AR(Base):
 		text_list: list[Tensor],
 		proms_list: list[Tensor],
 		resps_list: list[Tensor] | None = None,
+		lang_list: list[Tensor] | None = None,
 		max_steps: int = 1000,
 		max_resp_context: int = -1,
 
@@ -128,6 +129,7 @@ class AR(Base):
 				proms_list=proms_list,
 				resps_list=self._unsqueeze_list(resps_list),
 				targ_list=resps_list,
+				lang_list=lang_list,
 				quant_levels=None,
 			)
 

vall_e/models/ar_nar.py

@@ -134,6 +134,7 @@ class AR_NAR(Base):
 					proms_list=proms_list,
 					resps_list=resps_list,
 					targ_list=targ_list,
+					lang_list=lang_list,
 					quant_levels=quant_levels,
 				)
 			# is NAR
@@ -153,6 +154,7 @@ class AR_NAR(Base):
 					text_list=text_list,
 					proms_list=proms_list,
 					resps_list=prev_list,
+					lang_list=lang_list,
 					quant_levels=quant_levels,
 				)
 
@@ -204,7 +206,7 @@ class AR_NAR(Base):
 				text_list=text_list,
 				proms_list=proms_list,
 				resps_list=resps_list,
-				
+				lang_list=lang_list,
 				state=recurrent_state
 			)
 

vall_e/models/base.py

@@ -304,7 +304,7 @@ class Base(nn.Module):
 		batch_size = len(text_list)
 
 		if self.langs_emb is None:
-			langs_list = None
+			lang_list = None
 
 		x_list = self._samplewise_merge_tensors(
 			self.text_emb(text_list),

vall_e/models/nar.py

@@ -74,6 +74,7 @@ class NAR(Base):
 		text_list: list[Tensor],
 		proms_list: list[Tensor],
 		resps_list: list[Tensor],
+		lang_list: list[Tensor] | None = None,
 		max_levels: int = 0,
 		sampling_temperature: float = 0.2,
 		sampling_min_temperature: float = -1.0,
@@ -118,6 +119,7 @@ class NAR(Base):
 				proms_list=proms_list,
 				resps_list=prev_list,
 				targ_list=targ_list,
+				lang_list=lang_list,
 				quant_levels=quant_levels,
 			)
 
@@ -139,6 +141,7 @@ class NAR(Base):
 					text_list=text_list,
 					proms_list=proms_list,
 					resps_list=prev_list,
+					lang_list=lang_list,
 					quant_levels=quant_levels,
 				)
 

vall_e/train.py

@@ -26,7 +26,8 @@ def train_feeder(engine, batch):
 		engine(
 			text_list=batch["text"],
 			proms_list=[prom[:, :engine._cfg.prom_levels] for prom in batch["proms"]], # reduce the input prompt to the target prom level
-			resps_list=batch["resps"]
+			resps_list=batch["resps"],
+			lang_list=batch["lang"],
 		)
 
 		losses = engine.gather_attribute("loss")