Update modeling_gemma3mm.py

modeling_gemma3mm.py

@@ -491,12 +491,13 @@ class Gemma3MMForConditionalGeneration(Gemma3MMPreTrainedModel, GenerationMixin)
 
             if not is_torchdynamo_compiling() and inputs_embeds[special_audio_mask].numel() != masked_audio_features.numel():
                 audio_tokens_in_text = (special_audio_mask).sum(dim=1).sum(dim=0)[0]
-                masked_audio_size = audio_embed_sizes.sum()[0]
-                raise ValueError(
-                    f"Number of images does not match number of special image tokens in the input text. "
-                    f"Got {audio_tokens_in_text} image tokens in the text but {masked_audio_size} "
-                    "tokens from image embeddings."
-                )
+                masked_audio_size = audio_embed_sizes.sum()
+                #raise ValueError(
+                #    f"Number of images does not match number of special image tokens in the input text. "
+                #    f"Got {audio_tokens_in_text} image tokens in the text but {masked_audio_size} "
+                #    "tokens from image embeddings."
+                #)
+                pass
             masked_audio_features = masked_audio_features.to(inputs_embeds.device, inputs_embeds.dtype)
             inputs_embeds = inputs_embeds.masked_scatter(special_audio_mask, masked_audio_features)