[SFT] fix check for AutoLigerKernelForCausalLM (#2874)

* fix check for AutoLigerKernelForCausalLM

* fix case where AutoLigerKernelForCausalLM is not defined

* update min liger version

* formatting

* fix win CI

setup.py

@@ -85,7 +85,7 @@
     "diffusers": ["diffusers>=0.18.0"],
     "judges": ["openai>=1.23.2", "llm-blender>=0.0.2"],
     # liger-kernel depends on triton, which is only available on Linux https://github.com/triton-lang/triton#compatibility
-    "liger": ["liger-kernel>=0.4.0; sys_platform != 'win32'"],
+    "liger": ["liger-kernel>=0.5.3; sys_platform != 'win32'"],
     "mergekit": ["mergekit>=0.0.5.1"],
     "peft": ["peft>=0.8.0"],
     "quantization": ["bitsandbytes"],

trl/trainer/sft_trainer.py

@@ -60,6 +60,8 @@
 
 if is_liger_kernel_available():
     from liger_kernel.transformers import AutoLigerKernelForCausalLM
+else:
+    AutoLigerKernelForCausalLM = None
 
 if is_wandb_available():
     import wandb
@@ -449,7 +451,10 @@ def compute_loss(self, model, inputs, return_outputs=False, num_items_in_batch=N
         )
 
         # Compute token accuracy if we have labels and if the model is not using Liger (no logits)
-        if "labels" in inputs and not self.args.use_liger:
+        use_liger = self.args.use_liger or (
+            AutoLigerKernelForCausalLM is not None and isinstance(model, AutoLigerKernelForCausalLM)
+        )
+        if "labels" in inputs and not use_liger:
             shift_logits = outputs.logits[..., :-1, :].contiguous()
             shift_labels = inputs["labels"][..., 1:].contiguous()