fused_linear_cross_entropy caused bad performance

Nobody has claimed this yet.

### 🐛 Describe the bug

Hello, i'm finetuning Qwen-2.5-VL-3B with trl. When i turn on the liger kernel, it could lead to poor performance while the performance is normal without liger kernel. I later discovered that it was `fused_linear_cross_entropy` that caused the performance difference. However, their loss looks almost the same, and the problem is that this option is crucial for saving GPU memory.

Here is my training args
deepspeed --master_port 25410 src/open_r1/sft.py \
--model_name_or_path .cache/Qwen2.5-VL-3B-Instruct \
--dataset_name data/instructions_dynamic_combine_action.jsonl \
--deepspeed scripts/zero3.json \
--learning_rate 2.0e-5 \
--num_train_epochs 1 \
--packing \
--max_seq_length 64800 \
--per_device_train_batch_size 4 \
--gradient_accumulation_steps 4 \
--dataloader_num_workers 4 \
--gradient_checkpointing True \
--torch_dtype bfloat16 \
--bf16 \
--logging_steps 5 \
--eval_strategy no \
--eval_steps 100 \
--save_strategy no \
--save_steps 6000 \
--attn_implementation flash_attention_2 \
--output_dir result/Qwen2.5-VL_3B_sft_r2r_dynamic_test_fix_liger \
--report_to tensorboard

### Reproduce

_No response_

### Versions

Environment Report:
-------------------
Operating System: Linux-6.8.0-54-generic-x86_64-with-glibc2.39
Python version: 3.10.16
Liger Kernel version: 0.5.10
PyTorch version: 2.7.1+cu126
CUDA version: 12.6
HIP(ROCm) version: Not available
Triton version: 3.3.1
Transformers version: 4.50.3
Trl version: 0.16.0
XPU version: XPU Not Available

Start with the fused_linear_cross_entropy entry point and the training command in src/open_r1/sft.py, using scripts/zero3.json and the reported environment as context. Compare training with and without the option, then establish a reproducible performance difference and identify the affected behavior; no dedicated test or reproduction is provided in the issue.

Written by the indexing model from the issue text.