请提供下运行方式，以及Paddle版本，cuda和cudnn版本等详细信息，另外可以把训练日志也上传一下

运行方式：python ./tools/train.py -c /home/compat/chaofanl/PaddleClas/ResNet50-new.yaml
paddlepaddle ：f55b387df0f473574f82c83da0c4c821829f35a7 （Date: Tue Apr 25 21:02:37 2023）
paddleclas ：8ed2060033bfccef58c6e07a96ab4181f72fa7c5 （Date: Fri Jul 7 07:45:52 2023）
我使用的是intel-gpu

train.log

yaml文件如下
Global:
checkpoints: null
pretrained_model: null
output_dir: ./output-new/
device: intel_gpu
save_interval: 1
eval_during_train: False
eval_interval: 1
epochs: 120
print_batch_step: 1
use_visualdl: True
image_channel: &image_channel 3
image_shape: [*image_channel, 224, 224]
save_inference_dir: ./inference
to_static: False

Arch:
name: ResNet50
class_num: 1000
input_image_channel: *image_channel
data_format: "NHWC"
Loss:
Train:

• CELoss:
  weight: 1.0
  Eval:
• CELoss:
  weight: 1.0
  Optimizer:
  name: SGD
  lr:
  name: Piecewise
  learning_rate: 0.1
  decay_epochs: [30, 60, 90]
  values: [0.1, 0.01, 0.001, 0.0001]
  regularizer:
  name: 'L2'
  coeff: 0.0001

DataLoader:
Train:
dataset:
name: ImageNetDataset
image_root: /home_notmod/dataset_broad/dataset/imagenet/img_raw
cls_label_path: /home_notmod/dataset_broad/dataset/imagenet/img_raw/train_list.txt
transform_ops:

• DecodeImage:
  to_rgb: True
  channel_first: False
• RandCropImage:
  size: 224
• RandFlipImage:
  flip_code: 1
• NormalizeImage:
  scale: 1.0/255.0
  mean: [0.485, 0.456, 0.406]
  std: [0.229, 0.224, 0.225]
  order: ''
  channel_num: *image_channel

sampler:
  name: DistributedBatchSampler
  batch_size: 256
  drop_last: False
  shuffle: True
loader:
  num_workers: 4
  use_shared_memory: True

Eval:
dataset:
name: ImageNetDataset
image_root: /home_notmod/dataset_broad/dataset/imagenet/img_raw
cls_label_path: /home_notmod/dataset_broad/dataset/imagenet/img_raw/val_list.txt
transform_ops:

• DecodeImage:
  to_rgb: True
  channel_first: False
• ResizeImage:
  resize_short: 256
• CropImage:
  size: 224
• NormalizeImage:
  scale: 1.0/255.0
  mean: [0.485, 0.456, 0.406]
  std: [0.229, 0.224, 0.225]
  order: ''
  channel_num: *image_channel
  sampler:
  name: DistributedBatchSampler
  batch_size: 64
  drop_last: False
  shuffle: False
  loader:
  num_workers: 4
  use_shared_memory: True

Infer:
infer_imgs: docs/images/inference_deployment/whl_demo.jpg
batch_size: 10
transforms:

• DecodeImage:
  to_rgb: True
  channel_first: False
• ResizeImage:
  resize_short: 256
• CropImage:
  size: 224
• NormalizeImage:
  scale: 1.0/255.0
  mean: [0.485, 0.456, 0.406]
  std: [0.229, 0.224, 0.225]
  order: ''
  channel_num: *image_channel
• ToCHWImage:
  PostProcess:
  name: Topk
  topk: 5
  class_id_map_file: ppcls/utils/imagenet1k_label_list.txt

Metric:
Train:

• TopkAcc:
  topk: [1, 5]
  Eval:
• TopkAcc:
  topk: [1, 5]

我当初也是有这个么问题，用paddle.vision.model 的模型训练没法收敛，自己随便搭个四五层的都能收敛。。。你发现问题没

我当初也是有这个么问题，用paddle.vision.model 的模型训练没法收敛，自己随便搭个四五层的都能收敛。。。你发现问题没

你后来怎么解决的问题?

我当初也是有这个么问题，用paddle.vision.model 的模型训练没法收敛，自己随便搭个四五层的都能收敛。。。你发现问题没

你后来怎么解决的问题?
后来用paddlecls套件训练的，套件训练没问题，套件用paddle的版本比较低。具体怎么回事我也不清楚

能否使用PaddleClas套件的release2.5分支，同时用Paddle最新的release版本试试看是否还存在这个问题？