transformers is meant to be an optional dep but import fails without it

[tyler-romero]

🐛 Describe the bug

Attempting to run from liger_kernel.transformers.fused_linear_cross_entropy import LigerFusedLinearCrossEntropyLoss fails when transformers is not installed. However transformers is an optional dependency.

Many modules in Liger-Kernel are useful outside of the context of transformers (such as LigerFusedLinearCrossEntropyLoss) and so it would be nice to avoid the dependency when it is not needed (even in liger_kernel.transformers package).

Traceback (most recent call last):
  File "...", line 20, in <module>
    from liger_kernel.transformers.fused_linear_cross_entropy import LigerFusedLinearCrossEntropyLoss
  File "/home/tromero/workspace/nanogpt-speedrun/.venv/lib/python3.12/site-packages/liger_kernel/transformers/__init__.py", line 1, in <module>
    from liger_kernel.transformers.auto_model import AutoLigerKernelForCausalLM  # noqa: F401
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/tromero/workspace/.../.venv/lib/python3.12/site-packages/liger_kernel/transformers/auto_model.py", line 3, in <module>
    from transformers import AutoConfig
ModuleNotFoundError: No module named 'transformers'
Traceback (most recent call last):
  File "/home/tromero/workspace/...", line 20, in <module>
    from liger_kernel.transformers.fused_linear_cross_entropy import LigerFusedLinearCrossEntropyLoss
  File "/home/tromero/workspace/.../.venv/lib/python3.12/site-packages/liger_kernel/transformers/__init__.py", line 1, in <module>
    from liger_kernel.transformers.auto_model import AutoLigerKernelForCausalLM  # noqa: F401
    ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/tromero/workspace/.../.venv/lib/python3.12/site-packages/liger_kernel/transformers/auto_model.py", line 3, in <module>
    from transformers import AutoConfig
ModuleNotFoundError: No module named 'transformers'

Reproduce

No response

Versions

Environment Report:

Operating System: Linux-6.8.0-52-generic-x86_64-with-glibc2.35
Python version: 3.12.5
Liger Kernel version: 0.5.3
PyTorch version: 2.6.0+cu124
CUDA version: 12.4
HIP(ROCm) version: Not available
Triton version: 3.2.0
Transformers version: 4.49.0
XPU version: XPU Not Available