fix: jamba cache fails to use torch.nn.module (#32894)

Co-authored-by: Gal Cohen <[email protected]>

src/transformers/models/jamba/modeling_jamba.py

@@ -210,6 +210,7 @@ class HybridMambaAttentionDynamicCache(DynamicCache):
     """
 
     def __init__(self, config, batch_size, dtype=torch.float16, device=None):
+        super().__init__()
         self.dtype = dtype
         self.layers_block_type = config.layers_block_type
         self.has_previous_state = False  # only used by mamba