Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Menu
Open sidebar
evt_fugx1
dcu_megatron
Commits
950d42b4
Commit
950d42b4
authored
Mar 26, 2025
by
dongcl
Browse files
megatron patch
parent
4e2de453
Changes
1
Hide whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
3 additions
and
2 deletions
+3
-2
dcu_megatron/adaptor/megatron_adaptor.py
dcu_megatron/adaptor/megatron_adaptor.py
+3
-2
No files found.
dcu_megatron/adaptor/megatron_adaptor.py
View file @
950d42b4
...
@@ -110,7 +110,7 @@ class CoreAdaptation(MegatronAdaptationABC):
...
@@ -110,7 +110,7 @@ class CoreAdaptation(MegatronAdaptationABC):
def
patch_core_transformers
(
self
):
def
patch_core_transformers
(
self
):
from
..core
import
transformer_block_init_wrapper
,
transformer_block_forward
from
..core
import
transformer_block_init_wrapper
,
transformer_block_forward
from
..core.transformer.transformer_config
import
TransformerConfig
,
MLATransformerConfig
from
..core.transformer.transformer_config
import
TransformerConfig
,
MLATransformerConfig
# Transformer block
# Transformer block
MegatronAdaptation
.
register
(
'megatron.core.transformer.transformer_block.TransformerBlock.__init__'
,
MegatronAdaptation
.
register
(
'megatron.core.transformer.transformer_block.TransformerBlock.__init__'
,
transformer_block_init_wrapper
)
transformer_block_init_wrapper
)
...
@@ -141,7 +141,8 @@ class CoreAdaptation(MegatronAdaptationABC):
...
@@ -141,7 +141,8 @@ class CoreAdaptation(MegatronAdaptationABC):
def
patch_core_extentions
(
self
):
def
patch_core_extentions
(
self
):
import
transformer_engine
as
te
import
transformer_engine
as
te
from
..core.extensions.transformer_engine
import
te_dot_product_attention_init
,
TEGroupedLinear
from
..core.extensions.transformer_engine
import
te_dot_product_attention_init
from
megatron.core.extensions.transformer_engine
import
TEGroupedLinear
MegatronAdaptation
.
register
(
'megatron.core.extensions.transformer_engine.TEDotProductAttention.__init__'
,
MegatronAdaptation
.
register
(
'megatron.core.extensions.transformer_engine.TEDotProductAttention.__init__'
,
te_dot_product_attention_init
)
te_dot_product_attention_init
)
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment