louis_lifu
/
AReaL
mirror of https://github.com/inclusionAI/AReaL
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

PullRequest: 31 Fix typo in topology name 

Merge branch fw/topo of git@code.alipay.com:inclusionAI/AReaL.git into main
https://code.alipay.com/inclusionAI/AReaL/pull_requests/31

Signed-off-by: 晓雷 <meizhiyu.mzy@antgroup.com>


* change non-training topo order
* .
* fix the dataloading bug during recover
* fix typo
* fix typo
This commit is contained in:
博惟 2025-03-12 18:31:18 +08:00
parent 56e4dd1f5d
commit 234e3dd3a0
3 changed files with 9 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
realhf/base/testing.py
View File

@ -24,7 +24,7 @@ from realhf.base import constants, gpu_utils, logging, name_resolve, names, topo
from realhf.base.topology import (
DataPipeModelParallelTopology,
ParallelGrid,
PipeModelDataParallelTopology,
PipeDataModelParallelTopology,
)
logger = logging.getLogger("testing")
@ -222,7 +222,7 @@ def init_global_constants(
if topo is None:
if is_train:
topo = PipeModelDataParallelTopology(
topo = PipeDataModelParallelTopology(
num_dp=num_dp,
num_mp=num_mp,
num_pp=num_pp,

10
tests/comm/test_data_transfer.py
View File

@ -19,7 +19,7 @@ from realhf.api.core.data_api import SequenceSample
from realhf.base import constants, testing, topology
from realhf.base.testing import (
LocalMultiProcessTest,
PipeModelDataParallelTopology,
PipeDataModelParallelTopology,
init_global_constants,
)
from realhf.system.data_manager import DataManager
@ -39,7 +39,7 @@ def get_data_manager(
from_world_size = from_num_dp * from_num_mp * from_num_pp
to_world_size = to_num_dp * to_num_mp * to_num_pp
from_topo = topology.PipeModelDataParallelTopology(
from_topo = topology.PipeDataModelParallelTopology(
num_dp=from_num_dp,
num_mp=from_num_mp,
num_pp=from_num_pp,
@ -48,7 +48,7 @@ def get_data_manager(
max_prompt_len=None,
gradient_accumulation_fusion=False,
)
to_topo = topology.PipeModelDataParallelTopology(
to_topo = topology.PipeDataModelParallelTopology(
num_dp=to_num_dp,
num_mp=to_num_mp,
num_pp=to_num_pp,
@ -143,7 +143,7 @@ def _test_data_transfer(
):
from_model_name = ModelName("data_transfer_test", 0)
from_topo = PipeModelDataParallelTopology(
from_topo = PipeDataModelParallelTopology(
num_pp=from_pp_dp_mp[0],
num_mp=from_pp_dp_mp[-1],
num_dp=from_pp_dp_mp[1],
@ -152,7 +152,7 @@ def _test_data_transfer(
gradient_accumulation_fusion=True,
)
to_model_name = ModelName("data_transfer_test", 1)
to_topo = PipeModelDataParallelTopology(
to_topo = PipeDataModelParallelTopology(
num_pp=to_pp_dp_mp[0],
num_mp=to_pp_dp_mp[-1],
num_dp=to_pp_dp_mp[1],

4
tests/comm/test_param_realloc.py
View File

@ -139,7 +139,7 @@ def setup_constants_and_param_realloc(
from_world_size = from_num_dp * from_num_mp * from_num_pp
to_world_size = to_num_dp * to_num_mp * to_num_pp
from_topo = topology.PipeModelDataParallelTopology(
from_topo = topology.PipeDataModelParallelTopology(
num_dp=from_num_dp,
num_mp=from_num_mp,
num_pp=from_num_pp,
@ -148,7 +148,7 @@ def setup_constants_and_param_realloc(
max_prompt_len=None,
gradient_accumulation_fusion=False,
)
to_topo = topology.PipeModelDataParallelTopology(
to_topo = topology.PipeDataModelParallelTopology(
num_dp=to_num_dp,
num_mp=to_num_mp,
num_pp=to_num_pp,