fix: no need to dtype A in jamba (huggingface#32924)

Co-authored-by: Gal Cohen <galc@ai21.com>

src/transformers/models/jamba/modeling_jamba.py

@@ -630,7 +630,7 @@ def __init__(self, config: JambaConfig, layer_idx):
 
         # S4D real initialization. These are not discretized!
         # The core is to load them, compute the discrete states, then write the updated state. Keeps the memory bounded
-        A = torch.arange(1, self.ssm_state_size + 1, dtype=torch.float32)[None, :]
+        A = torch.arange(1, self.ssm_state_size + 1)[None, :]
         A = A.expand(self.intermediate_size, -1).contiguous()
 
         self.A_log = nn.Parameter(torch.log(A))