diff --git a/unsloth/models/rl_replacements.py b/unsloth/models/rl_replacements.py
index d8a1c63715..ee57055a00 100644
--- a/unsloth/models/rl_replacements.py
+++ b/unsloth/models/rl_replacements.py
@@ -237,7 +237,7 @@ def grpo_trainer_compute_loss(function_name, function):
             ref_per_token_logps, per_token_logps, input_ids, completion_mask, self.beta, advantages, bsz,
         )
         accumulated_loss, accumulated_completion_length, accumulated_mean_kl = grpo_accumulated_loss(
-            self, input_ids, logits_to_keep, completion_mask, advantages, n_chunks = 1,
+            self, _input_ids, logits_to_keep, completion_mask, advantages, n_chunks = 1,
         )
         print("loss", loss, accumulated_loss)
         print("completion_length", completion_length, accumulated_completion_length)