Skip to content

update ep cookbook #40

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
kevssim merged 1 commit into from
Feb 6, 2026
+11 −20
Merged

update ep cookbook #40

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
26 changes: 6 additions & 20 deletions cookbook/sft/ep_fsdp_qwen3_moe.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,12 +1,11 @@
# Copyright (c) ModelScope Contributors. All rights reserved.
import os

import numpy as np
import torch.distributed as dist
from transformers import AutoConfig

import twinkle
from twinkle import DeviceGroup, DeviceMesh, Platform, get_device_placement, get_logger
from twinkle import DeviceMesh, Platform, get_device_placement, get_logger
from twinkle.dataloader import DataLoader
from twinkle.dataset import Dataset, DatasetMeta
from twinkle.model import TransformersModel
Expand All @@ -19,29 +18,18 @@
'QWEN3_DATASET_ID', '/path/to/alpaca/dataset')
TEMPLATE_ID = os.environ.get('QWEN3_TEMPLATE_ID', 'Template')
PROCESSOR_ID = os.environ.get('QWEN3_PROCESSOR_ID', 'AlpacaProcessor')
REMOTE_GROUP = 'model'
NUM_LAYERS = int(os.environ.get('QWEN3_NUM_LAYERS', '1'))

device_group = [
DeviceGroup(
name=REMOTE_GROUP,
ranks=[0, 1, 2, 3],
device_type=Platform.get_platform().device_prefix(),
)
]

# 4 GPUs: dp=2, ep=2
device_mesh = DeviceMesh(
device_mesh = DeviceMesh.from_sizes(
device_type=Platform.get_platform().device_prefix(),
mesh=np.array([[0, 1], [2, 3]]),
mesh_dim_names=('dp', 'ep'),
dp_size=2,
ep_size=2,
)

os.environ.setdefault("RAY_DEDUP_LOGS", "0")
twinkle.initialize(
mode='ray',
mode='local',
nproc_per_node=4,
groups=device_group,
global_device_mesh=device_mesh,
)

Expand Down Expand Up @@ -78,15 +66,13 @@ def train():
dataloader = DataLoader(
dataset=dataset,
batch_size=4,
remote_group=REMOTE_GROUP,
device_mesh=device_mesh,
)

grad_accum_steps = 4
model = TransformersModel(
model_id=MODEL_ID,
config=config,
remote_group=REMOTE_GROUP,
device_mesh=device_mesh,
fsdp_config={
"expert_parallel": {
Expand Down Expand Up @@ -115,7 +101,7 @@ def train():
metric = metric()
logger.info(
f'Current is step {step // grad_accum_steps}, metric: {metric}')
if step % 50 == 0:
if step > 1 and step % 50 == 0:
model.save('./output')


Expand Down
5 changes: 5 additions & 0 deletions src/twinkle/utils/platform.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -86,6 +86,11 @@ def from_sizes(*, world_size: int = 1, dp_size: int = 1, fsdp_size: int = None,
mesh_dim_sizes.append(dp_size)
else:
mesh_dim_sizes.append(-1)
if ep_size is not None:
mesh_dim_sizes.append(ep_size)
mesh_dim_names.append("ep")
if origin_world_size == 1:
world_size *= ep_size
Comment on lines +89 to +93
Copy link
Copy Markdown
Contributor

@gemini-code-assist gemini-code-assist Bot Feb 6, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

To improve robustness, it's a good practice to validate input parameters. The ep_size should be a positive integer. Consider adding a check to ensure ep_size > 0 to prevent potential runtime errors during mesh creation if an invalid value is passed.

Suggested change
if ep_size is not None:
mesh_dim_sizes.append(ep_size)
mesh_dim_names.append("ep")
if origin_world_size == 1:
world_size *= ep_size
if ep_size is not None:
if ep_size <= 0:
raise ValueError(f'ep_size must be positive, but got {ep_size}')
mesh_dim_sizes.append(ep_size)
mesh_dim_names.append("ep")
if origin_world_size == 1:
world_size *= ep_size

if cp_size is not None:
mesh_dim_sizes.append(cp_size)
mesh_dim_names.append("cp")
Expand Down