423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2026-01-13 09:30:34 +08:00
Code Issues Packages Projects Releases Wiki Activity

[v1] add sft (#9752)

Browse Source
This commit is contained in:
Yaowei Zheng
2026-01-12 03:15:01 +08:00
committed by GitHub
parent 4d3621e3d3
commit 958b9c3468
29 changed files with 439 additions and 305 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
tests_v1/plugins/model_plugins/test_kernel_plugin.py
View File

@@ -43,7 +43,8 @@ def test_apply_kernel(mock_get_accelerator: MagicMock):
reload_kernels()
from llamafactory.v1.plugins.model_plugins.kernels.interface import apply_default_kernels
model = AutoModelForCausalLM.from_pretrained("llamafactory/tiny-random-qwen3")
# NOTE: use a special model to avoid contamination by other tests
model = AutoModelForCausalLM.from_pretrained("llamafactory/tiny-random-qwen2.5")
original_rmsnorm_forward = model.model.layers[0].input_layernorm.forward
original_swiglu_forward = model.model.layers[0].mlp.forward
model = apply_default_kernels(model=model, include_kernels="npu_fused_rmsnorm")
@@ -62,7 +63,8 @@ def test_apply_all_kernels(mock_get_accelerator: MagicMock):
reload_kernels()
from llamafactory.v1.plugins.model_plugins.kernels.interface import apply_default_kernels
model = AutoModelForCausalLM.from_pretrained("llamafactory/tiny-random-qwen3")
# NOTE: use a special model to avoid contamination by other tests
model = AutoModelForCausalLM.from_pretrained("llamafactory/tiny-random-qwen2.5")
original_rmsnorm_forward = model.model.layers[0].input_layernorm.forward
original_swiglu_forward = model.model.layers[0].mlp.forward