Fix Gemma2 4d attention mask (#31674)

Update modeling_gemma2.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

src/transformers/models/gemma2/modeling_gemma2.py

@@ -629,11 +629,13 @@ class Gemma2DecoderLayer(nn.Module):
         if (
             self.config._attn_implementation != "flash_attention_2" and self.is_sliding and attention_mask is not None
         ):  # efficient SDPA and no padding
-            attention_mask = attention_mask * torch.tril(
-                torch.ones_like(attention_mask), diagonal=-self.sliding_window
+            min_dtype = torch.finfo(hidden_states.dtype).min
+            sliding_window_mask = torch.tril(
+                torch.ones_like(attention_mask, dtype=torch.bool), diagonal=-self.sliding_window
             )
-            if attention_mask.shape[1] <= 1:  # when decoding
-                attention_mask = attention_mask[:, -self.sliding_window :]
+            attention_mask = torch.where(sliding_window_mask, min_dtype, attention_mask)
+            if attention_mask.shape[-1] <= 1:  # when decoding
+                attention_mask = attention_mask[:, :, :, -self.sliding_window :]
 
         residual = hidden_states