-
I want to collect stats for gradients.
I put stat collection code at couple places (method 1 and method 2 below). The stats coming from these two methods are very different. **Any idea why and whic…
-
-
Hi,
I'm training an image2image model.
Since my (8) CPUs load is pretty low and my (4) GPUs utilization is ~70% I'm trying to find where the bottleneck is.
The main Python process uses 3 diffe…
yg320 updated
6 years ago
-
Could not reopen the issue, please see here for more context: https://github.com/ppwwyyxx/tensorpack/issues/27#issuecomment-249977898
> The sign-vs-unsign problem is more relevant in FPGA. But as we …
-
In reference to #29.
I am also interested in implementing a multi-task learning model using tensorpack - similar to the "alternating training" example in https://jg8610.github.io/Multi-Task/. In thi…
-
I find that all layers only store quantized weights.
It's different from Dorefa-net, which also store origin weights and update them.
Will this affect the result?
-
Hi,
I can see where fw and fa being used, but where is fg being used?
-
Hi, 您好:
看了您的论文以及源码,获益颇多,在此有几个问题想请教一下,还请帮忙解惑~
1. 权重量化为低比特,input 以及features map是float32,是否有什么方法也可以做到量化?我想过直接映射到8bit,但是与-1这样的权重乘积会出错,除非把权重量化到8bit,然后扩成16bit计算,防止溢出,效率会下降,而且每层还需要取max和min。…