423A35C7/LLaMA-Factory
1
0
Fork 0
mirror of https://github.com/hiyouga/LLaMA-Factory.git synced 2025-12-18 04:40:35 +08:00
Code Issues Packages Projects Releases Wiki Activity

[model] switch to gptqmodel (#8108)

Browse Source
This commit is contained in:
hoshi-hiyouga
2025-05-19 22:25:40 +08:00
committed by GitHub
parent bc7f00f2c7
commit 45030ff803
9 changed files with 78 additions and 62 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/llamafactory/model/model_utils/attention.py
View File

@@ -29,10 +29,8 @@ if TYPE_CHECKING:
logger = logging.get_logger(__name__)
def configure_attn_implementation(
config: "PretrainedConfig", model_args: "ModelArguments", is_trainable: bool
) -> None:
if getattr(config, "model_type", None) == "gemma2" and is_trainable:
def configure_attn_implementation(config: "PretrainedConfig", model_args: "ModelArguments") -> None:
if getattr(config, "model_type", None) == "gemma2":
if model_args.flash_attn == AttentionFunction.AUTO or model_args.flash_attn == AttentionFunction.FA2:
if is_flash_attn_2_available():
if model_args.flash_attn != AttentionFunction.FA2: