-
Hi I used your implementation of wrapper of Conv2D function as :
```
with argscope([Conv2D, MaxPooling, AvgPooling, GlobalAvgPooling, BatchNorm], data_format='NCHW'), \
args…
-
### Description
I want to use transformer to reproduce the result of EN-DE 4.5M dataset which is used in the paper "attention is all you need". But I can't find any guideline.
What I need:
…
-
文档对于`trainer_count`的介绍不多,只在
[本地训练与预测](http://paddlepaddle.org/docs/develop/documentation/zh/faq/local/index_cn.html#id2) 里说这个是本地多线程的参数,可以理解这个就是训练线程数吗?
能否估算这个参数对内存(CPU训练下)的影响吗,我现在用4个count,占14G;1个c…
-
i was successful in testing the trained model by getting the trained weights you uploaded in Dropbox. However, I want to retrain the model on new training data.
I added one new image for the exist…
-
背景
------
目前在做句子打分(排序)模型,希望在开发集上算一个NDCG分数。
当前paddle在训练过程中做评价的方法
-------------------------------------
我查看了一些示例代码,感觉基本是这么做的:
```Python
trainer = paddle.trainer.SGD(
cost=model.cos…
-
Hi,
you have a mistake in your detector.py file.
line 7 recognizer.load('trainner/trainner.yml')
but in your README file je trainer, only one **N**
petus updated
7 years ago
-
Description:
When I try to train mount, for example, the windows with the chat with the npc wont open.
Current behaviour:
When I try to train mount, for example, the windows with the chat wit…
-
```text
@ubuntu:~/paddle/paddle/capi/examples/model_inference/sparse_binary$ ./sparse_binary
I0702 19:58:13.016194 6561 Util.cpp:166] commandline: --use_gpu=False
I0702 19:58:13.025660 6561 Grad…
-
相同配置下,单机运行是能正常打印指标的,但观察到集群训练的训练日志里,指标异常如下:
> I1213 14:44:38.848351 30749 CostLayer.cpp:337] calc pos/neg: 0 pos= 0 neg= 0
> I1213 14:44:38.848542 30749 CostLayer.cpp:337] calc pos/neg: 0 pos= 0 ne…
-
I am having some crash 2 weeks ago, so i decided to compile it in debug mode to get the crash log, but in my debug crash it doesnt appear nothing interesting (at leat for me).
Did I change something…