question on training on multi gpus

i have installed the xtuner sucessfully and sft several times by following cmd:

```
xtuner train /home/fusionai/project/internllm_demo/llama3/llama3-ft/configs/ztf_llava_llama3_8b_instruct_qlora_clip_vit_large_p14_336_e1_gpu1_stf_1k_repeated_4k_codeflow.py \
--work-dir /home/fusionai/project/internllm_demo/llama3/llama3-ft/train/llava_train_20240528_sft_stf_1k_repeated_4k_codeflow \
--deepspeed deepspeed_zero2
```
that's ok , and nothing wrong!

becase i need to execute manay experiments and want to train serveral configs on 2 A100.
that's to say, i want to train 4 experiments on 2 x A100 at the same time,
i follow the issues on multi gpus trains [issue 336](https://github.com/InternLM/xtuner/issues/366).

```
# excuete on node 0
NPROC_PER_NODE=2 NNODES=4 PORT=29600 ADDR="0.0.0.0" NODE_RANK=0 \
xtuner train /home/fusionai/project/internllm_demo/llama3/llama3-ft/configs/ztf_llava_llama3_8b_instruct_qlora_clip_vit_large_p14_336_e1_gpu1_stf_1k_repeated_4k_codeflow.py \
--work-dir /home/fusionai/project/internllm_demo/llama3/llama3-ft/train/llava_train_20240528_sft_stf_1k_repeated_4k_codeflow \
--deepspeed deepspeed_zero2
```

there is no response for a long time and stucked in followings :

```
(llama3) fusionai@train68:~/project/internllm_demo/llama3/llama3-ft/gstat_log$ NPROC_PER_NODE=2 NNODES=4 PORT=29600 ADDR="0.0.0.0" NODE_RANK=0
xtuner train /home/fusionai/project/internllm_demo/llama3/llama3-ft/configs/ztf_llava_llama3_8b_instruct_qlora_clip_vit_large_p14_336_e1_gpu1_s
--work-dir /home/fusionai/project/internllm_demo/llama3/llama3-ft/train/llava_train_20240528_sft_stf_1k_repeated_4k_codeflow \
--deepspeed deepspeed_zero2
[2024-05-28 14:46:05,932] [INFO] [real_accelerator.py:203:get_accelerator] Setting ds_accelerator to cuda (auto detect)
[WARNING] async_io requires the dev libaio .so object and headers but these were not found.
[WARNING] async_io: please install the libaio-dev package with apt
[WARNING] If libaio is already installed (perhaps from source), try setting the CFLAGS and LDFLAGS environment variables to where it can be f
[WARNING] Please specify the CUTLASS repo directory as environment variable $CUTLASS_PATH
[WARNING] sparse_attn requires a torch version >= 1.5 and < 2.0 but detected 2.1
[WARNING] using untested triton version (2.1.0), only 1.0.0 is known to be compatible
[2024-05-28 14:46:09,280] torch.distributed.run: [WARNING]
[2024-05-28 14:46:09,280] torch.distributed.run: [WARNING] *****************************************
[2024-05-28 14:46:09,280] torch.distributed.run: [WARNING] Setting OMP_NUM_THREADS environment variable for each process to be 1 in default, to please further tune the variable for optimal performance in your application as needed.
[2024-05-28 14:46:09,280] torch.distributed.run: [WARNING] *****************************************
```

and the config can be listed as followwing:
```
# Copyright (c) OpenMMLab. All rights reserved.
import torch
from mmengine.hooks import (CheckpointHook, DistSamplerSeedHook, IterTimerHook,
LoggerHook, ParamSchedulerHook)
from mmengine.optim import AmpOptimWrapper, CosineAnnealingLR, LinearLR
from peft import LoraConfig
from torch.optim import AdamW
from transformers import (AutoModelForCausalLM, AutoTokenizer,
BitsAndBytesConfig, CLIPImageProcessor,
CLIPVisionModel)

from xtuner.dataset import LLaVADataset
from xtuner.dataset.collate_fns import default_collate_fn
from xtuner.dataset.map_fns import llava_map_fn, template_map_fn_factory
from xtuner.dataset.samplers import LengthGroupedSampler
from xtuner.engine.hooks import DatasetInfoHook, EvaluateChatHook
from xtuner.engine.runner import TrainLoop
from xtuner.model import LLaVAModel
from xtuner.utils import PROMPT_TEMPLATE

#######################################################################
# PART 1 Settings #
#######################################################################
# Model
llm_name_or_path = '/home/fusionai/.cache/modelscope/hub/LLM-Research/Meta-Llama-3-8B-Instruct'
visual_encoder_name_or_path = '/home/fusionai/.cache/modelscope/hub/AI-ModelScope/clip-vit-large-patch14-336'
# Specify the pretrained pth
pretrained_pth = '/home/fusionai/project/internllm_demo/llama3/pretrained-model/llama3-llava-iter_2181.pth' # noqa: E501

# Data
data_root = '/home/fusionai/project/datasets/llama3_datasets/llama3_repeated1k_codeflow4.2k/'
data_path = data_root + 'qa_repeated1k_codeflow4.2k.json'
image_folder = data_root
prompt_template = PROMPT_TEMPLATE.llama3_chat
max_length = int(2048 - (336 / 14)**2)

# Scheduler & Optimizer
batch_size = 1 # per_device
accumulative_counts = 16
dataloader_num_workers = 0
max_epochs = 1
optim_type = AdamW
lr = 2e-4
betas = (0.9, 0.999)
weight_decay = 0
max_norm = 1 # grad clip
warmup_ratio = 0.03

# Save
save_steps = 2000
save_total_limit = 3 # Maximum checkpoints to keep (-1 means unlimited)

# Evaluate the generation performance during the training
evaluation_freq = 2000
SYSTEM = ''
evaluation_images = '/home/fusionai/project/datasets/llama3_datasets/llama3_repeated/imgs/test0001.png'
evaluation_inputs = ['描述一下图中逻辑关系。', '图中包含哪些信号或设备？']

#######################################################################
# PART 2 Model & Tokenizer & Image Processor #
#######################################################################
tokenizer = dict(
type=AutoTokenizer.from_pretrained,
pretrained_model_name_or_path=llm_name_or_path,
trust_remote_code=True,
padding_side='right')

image_processor = dict(
type=CLIPImageProcessor.from_pretrained,
pretrained_model_name_or_path=visual_encoder_name_or_path,
trust_remote_code=True)

model = dict(
type=LLaVAModel,
freeze_llm=True,
freeze_visual_encoder=True,
pretrained_pth=pretrained_pth,
llm=dict(
type=AutoModelForCausalLM.from_pretrained,
pretrained_model_name_or_path=llm_name_or_path,
trust_remote_code=True,
torch_dtype=torch.float16,
quantization_config=dict(
type=BitsAndBytesConfig,
load_in_4bit=True,
load_in_8bit=False,
llm_int8_threshold=6.0,
llm_int8_has_fp16_weight=False,
bnb_4bit_compute_dtype=torch.float16,
bnb_4bit_use_double_quant=True,
bnb_4bit_quant_type='nf4')),
llm_lora=dict(
type=LoraConfig,
r=64,
lora_alpha=16,
lora_dropout=0.05,
bias='none',
task_type='CAUSAL_LM'),
visual_encoder=dict(
type=CLIPVisionModel.from_pretrained,
pretrained_model_name_or_path=visual_encoder_name_or_path))

#######################################################################
# PART 3 Dataset & Dataloader #
#######################################################################
llava_dataset = dict(
type=LLaVADataset,
data_path=data_path,
image_folder=image_folder,
tokenizer=tokenizer,
image_processor=image_processor,
dataset_map_fn=llava_map_fn,
template_map_fn=dict(
type=template_map_fn_factory, template=prompt_template),
max_length=max_length,
pad_image_to_square=True)

train_dataloader = dict(
batch_size=batch_size,
num_workers=dataloader_num_workers,
dataset=llava_dataset,
sampler=dict(
type=LengthGroupedSampler,
length_property='modality_length',
per_device_batch_size=batch_size * accumulative_counts),
collate_fn=dict(type=default_collate_fn))

#######################################################################
# PART 4 Scheduler & Optimizer #
#######################################################################
# optimizer
optim_wrapper = dict(
type=AmpOptimWrapper,
optimizer=dict(
type=optim_type, lr=lr, betas=betas, weight_decay=weight_decay),
clip_grad=dict(max_norm=max_norm, error_if_nonfinite=False),
accumulative_counts=accumulative_counts,
loss_scale='dynamic',
dtype='float16')

# learning policy
# More information: https://github.com/open-mmlab/mmengine/blob/main/docs/en/tutorials/param_scheduler.md # noqa: E501
param_scheduler = [
dict(
type=LinearLR,
start_factor=1e-5,
by_epoch=True,
begin=0,
end=warmup_ratio * max_epochs,
convert_to_iter_based=True),
dict(
type=CosineAnnealingLR,
eta_min=0.0,
by_epoch=True,
begin=warmup_ratio * max_epochs,
end=max_epochs,
convert_to_iter_based=True)
]

# train, val, test setting
train_cfg = dict(type=TrainLoop, max_epochs=max_epochs)

#######################################################################
# PART 5 Runtime #
#######################################################################
# Log the dialogue periodically during the training process, optional
custom_hooks = [
dict(type=DatasetInfoHook, tokenizer=tokenizer),
dict(
type=EvaluateChatHook,
tokenizer=tokenizer,
image_processor=image_processor,
every_n_iters=evaluation_freq,
evaluation_inputs=evaluation_inputs,
evaluation_images=evaluation_images,
system=SYSTEM,
prompt_template=prompt_template)
]

# configure default hooks
default_hooks = dict(
# record the time of every iteration.
timer=dict(type=IterTimerHook),
# print log every 10 iterations.
logger=dict(type=LoggerHook, log_metric_by_epoch=False, interval=10),
# enable the parameter scheduler.
param_scheduler=dict(type=ParamSchedulerHook),
# save checkpoint per `save_steps`.
checkpoint=dict(
type=CheckpointHook,
by_epoch=False,
interval=save_steps,
max_keep_ckpts=save_total_limit),
# set sampler seed in distributed evrionment.
sampler_seed=dict(type=DistSamplerSeedHook),
)

# configure environment
env_cfg = dict(
# whether to enable cudnn benchmark
cudnn_benchmark=False,
# set multi process parameters
mp_cfg=dict(mp_start_method='fork', opencv_num_threads=0),
# set distributed parameters
dist_cfg=dict(backend='nccl'),
)

# set visualizer
visualizer = None

# set log level
log_level = 'INFO'

# load from which checkpoint
load_from = None

# whether to resume training from the loaded checkpoint
resume = False

# Defaults to use random seed and disable `deterministic`
randomness = dict(seed=None, deterministic=False)

# set log processor
log_processor = dict(by_epoch=False)

```

is there anything wrong on multi gpus training?
look forward to your anwser.