训练时Acc会一直为0吗？

[zch-ssc]

使用casia数据集训练，无预训模型。 使用config_ms1m_res50.yaml配置项acc会一直为0吗？

model params

backbone_type: resnet_v2_50 loss_type: arcface out_type: E

image_size: 112 embd_size: 512 class_num: 10572

hyper params

bn_decay: 0.9 keep_prob: 0.4 weight_decay: !!float 5e-4 logits_scale: 64.0 logits_margin: 0.5 momentum: 0.9

run params

val_bn_train: False augment_flag: True augment_margin: 16

gpu_num: 1 batch_size: 256 epoch_num: 20 step_per_epoch: 10000 val_freq: 2000

lr_steps: [40000, 60000, 80000] lr_values: [0.004, 0.002, 0.0012, 0.0004]

paths

pretrained_model: ''

train_data: ['/opt/gpu/z/InsightFace-tensorflow-master/data/casia.tfrecord']

[knaffe]

I use my the stanford dogs dataset to fine tuning , but get acc always 0 until 10000 steps. And the acc are nearly 1.0 or 0.99. What is the suitable config for new dataset?

[YTEP-ZHI]

I also encountered this problem, when fine tuning on my own datasets, the acc is always 0, did u solved this problem?

[oldwolf204]

我也遇到了同样的问题，楼上的有解决方案了吗？

[knaffe]

训练久一点，或者更换其他的参数试试。