关于Qwen1.5 32B-Chat 训练的问题

很奇怪的问题是，在8*A100（80G）上无论我如何设置max-seq，从16000降到200，始终都会OOM。
如下是我的命令和配置：

NPROC_PER_NODE=8 nohup xtuner train qwen1_5_32b_chat --deepspeed deepspeed_zero3 > instruct.out 2>&1 &

#######################################################################
# PART 1 Settings #
#######################################################################

pretrained_model_name_or_path = '/Qwen1.5-32B-Chat'
use_varlen_attn = False

data_files = ['test.json']
prompt_template = PROMPT_TEMPLATE.qwen_chat
max_length = 200
pack_to_max_length = False

sequence_parallel_size = 1

batch_size = 1 # per_device

accumulative_counts = 1
accumulative_counts *= sequence_parallel_size
dataloader_num_workers = 0
max_epochs = 3
optim_type = AdamW
lr = 1e-5
betas = (0.9, 0.999)
weight_decay = 0
max_norm = 1 # grad clip
warmup_ratio = 0.03

save_steps = 500
save_total_limit = 1 # Maximum checkpoints to keep (-1 means unlimited)

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

model = dict(
type=SupervisedFinetune,
use_varlen_attn=use_varlen_attn,
llm=dict(
type=AutoModelForCausalLM.from_pretrained,
pretrained_model_name_or_path=pretrained_model_name_or_path,
trust_remote_code=True,
torch_dtype=torch.float16))

#######################################################################
# PART 3 Dataset & Dataloader #
#######################################################################
sampler = SequenceParallelSampler \
if sequence_parallel_size > 1 else DefaultSampler

train_dataset = dict(
type=process_hf_dataset,
use_varlen_attn=use_varlen_attn,
dataset=dict(type=load_dataset, path='json', data_files=data_files),
tokenizer=tokenizer,
max_length=max_length,
dataset_map_fn=None,
template_map_fn=dict(
type=template_map_fn_factory, template=prompt_template),
remove_unused_columns=True,
shuffle_before_pack=True,
pack_to_max_length=pack_to_max_length)

train_dataloader = dict(
batch_size=batch_size,
num_workers=dataloader_num_workers,
dataset=train_dataset,
sampler=dict(type=sampler, shuffle=True),
collate_fn=dict(type=default_collate_fn, use_varlen_attn=use_varlen_attn))

#######################################################################
# PART 4 Scheduler & 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')

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_cfg = dict(type=TrainLoop, max_epochs=max_epochs)

#######################################################################
# PART 5 Runtime #
#######################################################################
custom_hooks = [
dict(type=DatasetInfoHook, tokenizer=tokenizer),
dict(type=ThroughputHook)
]

if use_varlen_attn:
custom_hooks += [dict(type=VarlenAttnArgsToMessageHubHook)]

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),
sampler_seed=dict(type=DistSamplerSeedHook),
)

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'),
)

visualizer = None

log_level = 'INFO'

load_from = None

resume = True

randomness = dict(seed=None, deterministic=False)

log_processor = dict(by_epoch=False)