diff --git a/megatron/rl/rl_utils.py b/megatron/rl/rl_utils.py
index df19145bf2d..8f914dd11ff 100644
--- a/megatron/rl/rl_utils.py
+++ b/megatron/rl/rl_utils.py
@@ -799,13 +799,18 @@ def get_logprobs(model, tokens, position_ids, attention_mask, no_grad=False, pac
                         tp_world_size = get_tensor_model_parallel_world_size()
                         if actual_len % tp_world_size != 0:
                             actual_len = ((actual_len + tp_world_size - 1) // tp_world_size) * tp_world_size
-                            # Update cu_seqlens to match the padded length.
-                            # The last entry of cu_seqlens must equal the tensor's sequence dimension.
-                            # Without this, TE attention/rotary ops see mismatched dimensions.
+                            # We need to add padding tokens to make the sequence divisible by TP.
+                            # if we extend cu_seqlens[-1] to include padding in the
+                            # last real sequence,  this would cause that sequence to attend to
+                            # padding tokens (garbage values), leading to numerical instability/NaN.
+                            # Instead, we add a new "padding sequence" entry to cu_seqlens.
                             if packed_seq_params.cu_seqlens_q[-1].item() != actual_len:
-                                # Clone to avoid modifying cached params
-                                new_cu_seqlens = packed_seq_params.cu_seqlens_q.clone()
-                                new_cu_seqlens[-1] = actual_len
+                                # Add a new entry for the padding pseudo-sequence
+                                # cu_seqlens goes from [0, ..., total_real] to [0, ..., total_real, padded_len]
+                                new_cu_seqlens = torch.cat([
+                                    packed_seq_params.cu_seqlens_q,
+                                    torch.tensor([actual_len], dtype=torch.int32, device=packed_seq_params.cu_seqlens_q.device)
+                                ])
                                 packed_seq_params = PackedSeqParams(
                                     qkv_format=packed_seq_params.qkv_format,
                                     cu_seqlens_q=new_cu_seqlens,