423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2026-03-01 17:26:03 +08:00
Code Issues Packages Projects Releases Wiki Activity

[ci] disable pip cache for ci (#9654)

Browse Source
This commit is contained in:
Yaowei Zheng
2025-12-23 18:37:40 +08:00
committed by GitHub
parent 1c8a42d2f8
commit 84485406b7
5 changed files with 19 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
src/llamafactory/model/model_utils/moe.py
View File

@@ -82,6 +82,11 @@ def add_z3_leaf_module(model: "PreTrainedModel") -> None:
_set_z3_leaf_modules(model, [Glm4vMoeTextMoE])
if model_type == "gpt_oss":
from transformers.models.gpt_oss.modeling_gpt_oss import GptOssMLP
_set_z3_leaf_modules(model, [GptOssMLP])
if model_type == "jamba":
from transformers.models.jamba.modeling_jamba import JambaSparseMoeBlock
@@ -129,13 +134,9 @@ def add_z3_leaf_module(model: "PreTrainedModel") -> None:
if model_type in ("qwen3_omni_moe", "qwen3_omni_moe_thinker"):
from transformers.models.qwen3_omni_moe.modeling_qwen3_omni_moe import Qwen3OmniMoeThinkerTextSparseMoeBlock
_set_z3_leaf_modules(model, [Qwen3OmniMoeThinkerTextSparseMoeBlock])
if model_type == "gpt_oss":
from transformers.models.gpt_oss.modeling_gpt_oss import GptOssMLP
_set_z3_leaf_modules(model, [GptOssMLP])
def configure_moe(config: "PretrainedConfig", model_args: "ModelArguments", is_trainable: bool) -> None:
if not is_trainable or not model_args.moe_aux_loss_coef: