Getting Started

There are a couple of ways to apply Liger kernels, depending on the level of customization required.

1. Use AutoLigerKernelForCausalLM

Using the AutoLigerKernelForCausalLM is the simplest approach, as you don't have to import a model-specific patching API. If the model type is supported, the modeling code will be automatically patched using the default settings.

Example

from liger_kernel.transformers import AutoLigerKernelForCausalLM

# This AutoModel wrapper class automatically monkey-patches the
# model with the optimized Liger kernels if the model is supported.
model = AutoLigerKernelForCausalLM.from_pretrained("path/to/some/model")

2. Apply Model-Specific Patching APIs

Using the patching APIs, you can swap Hugging Face models with optimized Liger Kernels.

Example

import transformers
from liger_kernel.transformers import apply_liger_kernel_to_llama

# 1a. Adding this line automatically monkey-patches the model with the optimized Liger kernels
apply_liger_kernel_to_llama()

# 1b. You could alternatively specify exactly which kernels are applied
apply_liger_kernel_to_llama(
  rope=True,
  swiglu=True,
  cross_entropy=True,
  fused_linear_cross_entropy=False,
  rms_norm=False
)

# 2. Instantiate patched model
model = transformers.AutoModelForCausalLM("path/to/llama/model")

3. Compose Your Own Model

You can take individual kernels to compose your models.

Example

from liger_kernel.transformers import LigerFusedLinearCrossEntropyLoss
import torch.nn as nn
import torch

model = nn.Linear(128, 256).cuda()

# fuses linear + cross entropy layers together and performs chunk-by-chunk computation to reduce memory
loss_fn = LigerFusedLinearCrossEntropyLoss()

input = torch.randn(4, 128, requires_grad=True, device="cuda")
target = torch.randint(256, (4, ), device="cuda")

loss = loss_fn(model.weight, input, target)
loss.backward()