RuiyingLu / HVQ-Trans

Code for "Hierarchical Vector Quantized Transformer for Multi-class Unsupervised Anomaly Detection" published in NeurIPS 2023
29 stars 2 forks source link

作者您好!sampler = DistributedSampler(dataset).set_epoch(epoch) 请问分布式训练的时候,是否对训练集数据顺序打乱了呢 #5

Open ZHE-SAPI opened 1 week ago

ZHE-SAPI commented 1 week ago

因为根据代码 if distributed: sampler = DistributedSampler(dataset) # 似乎没有设置乱序,shuffle应该是默认为false else: sampler = RandomSampler(dataset)

同时谢谢您的杰出工作!

RuiyingLu commented 1 week ago

您好,谢谢关注与提醒。我们的训练与测试setting都是follow uniad。另外,训练均为正常样本,是否shuffle对结果影响不大

---- 回复的原邮件 ---- | 发件人 | @.> | | 发送日期 | 2024年10月24日 03:59 | | 收件人 | RuiyingLu/HVQ-Trans @.> | | 抄送人 | Subscribed @.***> | | 主题 | [RuiyingLu/HVQ-Trans] 作者您好!sampler = DistributedSampler(dataset).set_epoch(epoch) 请问分布式训练的时候,是否对训练集数据顺序打乱了呢 (Issue #5) |

因为根据代码 if distributed: sampler = DistributedSampler(dataset) # 似乎没有设置乱序,shuffle应该是默认为false else: sampler = RandomSampler(dataset)

同时谢谢您的杰出工作!

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you are subscribed to this thread.Message ID: @.***>