From ef07d9412dd45f63d3be3ae5bf3cdbac47c5c0ca Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=9D=8E=E8=8A=B3=E5=B7=9E?= <744976956@qq.com> Date: Fri, 25 Oct 2024 15:21:41 +0800 Subject: [PATCH] =?UTF-8?q?=E6=9B=B4=E6=96=B0=2010=E6=9C=8825=E6=97=A5?= =?UTF-8?q?=E6=9C=AC=E5=91=A8=E7=A7=91=E7=A0=94=E5=B7=A5=E4=BD=9C=E8=BF=9B?= =?UTF-8?q?=E5=B1=95.md?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- 10月25日本周科研工作进展.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/10月25日本周科研工作进展.md b/10月25日本周科研工作进展.md index fd38e71..fcbd530 100644 --- a/10月25日本周科研工作进展.md +++ b/10月25日本周科研工作进展.md @@ -39,6 +39,6 @@ loss_proto /= len(args.num_cluster) ### 解决办法与下周安排: -1.将实验视为loss相互竞争的多任务学习,考虑采用梯度标准化多目标优化(GradNorm: Gradient Normalization for Adaptive Loss Balancing in Deep Multitask Networks)或多目标优化(Multi-task learning as multi-objective optimization)的方式修改代码。 +1.将实验视为loss相互竞争的多任务学习,考虑采用梯度标准化(GradNorm: Gradient Normalization for Adaptive Loss Balancing in Deep Multitask Networks)或多目标优化(Multi-task learning as multi-objective optimization)的方式修改代码。 2.增加下游任务实验,观察不同结果的模型在图像分类/目标检测/聚类评估的任务中的发挥。 \ No newline at end of file