basic llama multipack
This commit is contained in:
@@ -10,6 +10,7 @@ from axolotl.monkeypatch.mixtral import patch_mixtral_moe_forward_zero3
|
|||||||
from axolotl.monkeypatch.utils import get_unpad_data
|
from axolotl.monkeypatch.utils import get_unpad_data
|
||||||
|
|
||||||
SUPPORTED_MULTIPACK_MODEL_TYPES = [
|
SUPPORTED_MULTIPACK_MODEL_TYPES = [
|
||||||
|
"llama",
|
||||||
"mixtral",
|
"mixtral",
|
||||||
"qwen2",
|
"qwen2",
|
||||||
"qwen2_moe",
|
"qwen2_moe",
|
||||||
@@ -29,6 +30,10 @@ def patch_for_multipack(model_type, model_name=None):
|
|||||||
)
|
)
|
||||||
if is_deepspeed_zero3_enabled():
|
if is_deepspeed_zero3_enabled():
|
||||||
patch_mixtral_moe_forward_zero3()
|
patch_mixtral_moe_forward_zero3()
|
||||||
|
elif model_type == "llama":
|
||||||
|
transformers.models.llama.modeling_llama._get_unpad_data = ( # pylint: disable=protected-access
|
||||||
|
get_unpad_data
|
||||||
|
)
|
||||||
elif model_type == "qwen2":
|
elif model_type == "qwen2":
|
||||||
transformers.models.qwen2.modeling_qwen2._get_unpad_data = ( # pylint: disable=protected-access
|
transformers.models.qwen2.modeling_qwen2._get_unpad_data = ( # pylint: disable=protected-access
|
||||||
get_unpad_data
|
get_unpad_data
|
||||||
|
|||||||
Reference in New Issue
Block a user