-
我在GPU多卡下的模型预测结果有问题。
我在模型训练的时候每隔n个batch会用测试数据评估一下,假设这个时候模型的预测值为A;
然后在模型训练完成之后我想重新加载模型再次用同样的测试数据评估,然后这个时候模型的预测值为B;
但是我发现
GPU单卡的情况下B==A;GPU多卡的情况下B!=A。
而且很奇怪的是,多卡的情况下,一个batch里面前半部分的预测结果是对的,而后半部分的预测结果…
-
At the moment the docs give examples of how to do inference, but then a novice user has no way to know how to analyse their data. I suggest it might be good to have a (very short) `analysis.rst` file …
-
## ❓ Questions & Help
There is a multi gpus implementation of GraphSAGE in DGL:
https://github.com/dmlc/dgl/blob/master/examples/pytorch/graphsage/train_sampling_multi_gpu.py
The example used D…
-
Hi, I am trying to get the feature map of PointGNN, and I modified the method `MultiLayerFastLocalGraphModelV2.predict()` in `models/models.py`. In the return value, I only added `return logits, box_e…
-
版本、环境信息:
1)PaddlePaddle版本:1.6.2
2)python:3.6.3
训练信息
1)单机单卡
```python
total_loss = self.alpha * losses[0] + self.beta * losses[1] + self.gamma * losses[2]
logging.info("Shape of tot…
-
- 版本、环境信息:
1)PaddlePaddle版本:1.6.2
2)python:3.6.3
- 训练信息
1)单机单卡
- 问题描述
```python
image_input = fluid.layers.data(name="image",
shape=[512,
…
-
As there seem to be different opinions on workflow, especially merges, we need to discuss that in an own issue instead of the now closed pull request #58.
Requirements/demands from the Ontology hac…
-
1. 数据处理部分
- 执行` python preprocess.py` 问题:
```bash
Namespace(dataset='sample')
-- Starting @ 2019-03-28 05:49:51.411397s
Traceback (most recent call last):
File "preprocess.py", lin…
-
## 🐛 Bug
## To Reproduce
Steps to reproduce the behavior:
```
import torch
from torch_geometric.nn import EdgeConv
import torch.nn as nn
import torch.nn.functional as F
from torch_…
-
# Next paper candidates
Let's propose papers to study next! All papers mentioned in the comments of this issue will be listed in the next vote.
## Last session runner-up(s)
- [Learning and Evalua…