Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[SFT] fix check for AutoLigerKernelForCausalLM #2874

Merged
merged 5 commits into from
Feb 17, 2025
Merged

[SFT] fix check for AutoLigerKernelForCausalLM #2874

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
5 commits
Select commit Hold shift + click to select a range
f0677f3
fix check for AutoLigerKernelForCausalLM
kashif Feb 16, 2025
a5c08ce
fix case where AutoLigerKernelForCausalLM is not defined
kashif Feb 16, 2025
636c26b
update min liger version
kashif Feb 16, 2025
beeeaea
formatting
kashif Feb 16, 2025
c5f283d
fix win CI
kashif Feb 16, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion setup.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -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"],
Expand Down
7 changes: 6 additions & 1 deletion trl/trainer/sft_trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -60,6 +60,8 @@

if is_liger_kernel_available():
from liger_kernel.transformers import AutoLigerKernelForCausalLM
else:
AutoLigerKernelForCausalLM = None

if is_wandb_available():
import wandb
Expand Down Expand Up @@ -440,7 +442,10 @@ 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()

Expand Down
Loading