Upload hunyuan.py with huggingface_hub

hunyuan.py

@@ -358,7 +358,7 @@ class HunYuanModel(HunYuanPreTrainedModel):
         )
 
 
-class HunYuanForCausalLM(HunYuanPreTrainedModel):
+class HunYuanMoEV1ForCausalLM(HunYuanPreTrainedModel):
     _tied_weights_keys = ["lm_head.weight"]
 
     def __init__(self, config: HunYuanConfig):
@@ -527,7 +527,7 @@ class HunYuanForCausalLM(HunYuanPreTrainedModel):
             if isinstance(past_key_values, Cache):
                 cache_length = past_key_values.get_seq_length()
                 past_length = past_key_values.seen_tokens
-                max_cache_length = past_key_values.get_max_length()
+                max_cache_length = past_key_values.get_max_cache_shape()
             else:
                 cache_length = past_length = past_key_values[0][0].shape[2]
                 max_cache_length = None
@@ -586,7 +586,7 @@ class HunYuanForCausalLM(HunYuanPreTrainedModel):
         return reordered_past
 
 
-class MultimodelHunYuanForCausalLM(HunYuanForCausalLM):
+class MultimodelHunYuanForCausalLM(HunYuanMoEV1ForCausalLM):
     _tied_weights_keys = ["lm_head.weight"]
 
     def __init__(self, config: HunYuanConfig):