[ModernBert] Prevent the attention mask from being None in ModernBertForSequenceClassification (#35991)

* [ModernBert] Prevent the attention mask from being None in ModernBertForSequenceClassification

* fix the modular conversion

Author: ashmikuz

src/transformers/models/modernbert/modeling_modernbert.py

@@ -1151,6 +1151,19 @@ def forward(
         return_dict = return_dict if return_dict is not None else self.config.use_return_dict
         self._maybe_set_compile()
 
+        if input_ids is not None:
+            self.warn_if_padding_and_no_attention_mask(input_ids, attention_mask)
+
+        if batch_size is None and seq_len is None:
+            if inputs_embeds is not None:
+                batch_size, seq_len = inputs_embeds.shape[:2]
+            else:
+                batch_size, seq_len = input_ids.shape[:2]
+        device = input_ids.device if input_ids is not None else inputs_embeds.device
+
+        if attention_mask is None:
+            attention_mask = torch.ones((batch_size, seq_len), device=device, dtype=torch.bool)
+
         outputs = self.model(
             input_ids=input_ids,
             attention_mask=attention_mask,

src/transformers/models/modernbert/modular_modernbert.py

@@ -1277,6 +1277,19 @@ def forward(
         return_dict = return_dict if return_dict is not None else self.config.use_return_dict
         self._maybe_set_compile()
 
+        if input_ids is not None:
+            self.warn_if_padding_and_no_attention_mask(input_ids, attention_mask)
+
+        if batch_size is None and seq_len is None:
+            if inputs_embeds is not None:
+                batch_size, seq_len = inputs_embeds.shape[:2]
+            else:
+                batch_size, seq_len = input_ids.shape[:2]
+        device = input_ids.device if input_ids is not None else inputs_embeds.device
+
+        if attention_mask is None:
+            attention_mask = torch.ones((batch_size, seq_len), device=device, dtype=torch.bool)
+
         outputs = self.model(
             input_ids=input_ids,
             attention_mask=attention_mask,