ScutterKey

I also encounter the total_loss explosion when try the newest commit. I implement a caffe version of mask-rcnn and also encounter the same problem. here is the newest commit loss:...

https://github.com/caffe2/models/issues/5 could anyone help me? ``` from caffe2.proto import caffe2_pb2 from caffe2.python import core, workspace, models import numpy as np net_def = caffe2_pb2.NetDef() with open('./predict_net.pb') as f: net_def.ParseFromString(f.read()) device_opts =...

@zhangguanqun , Maybe i found a solution, here is my code ``` import numpy as np #import caffe2.python._import_c_extension as C from caffe2.proto import caffe2_pb2 import os, time import caffe2.python._import_c_extension as...

Oh, I find the lite_ios_build_armv7 in tools/anakin-lite. -- However, it build failed with this error /Users/ky/workspaces/comments/Anakin/saber/lite/funcs/neon/saber_eltwise_act.cpp:647:63: error: cannot initialize a parameter of type 'float *' with an lvalue of type...

Could you show me a project (using anakin lite) to help me with anakin lite API ?

@Superjomn @cs2be @varunarora @reyoung 请问你们能解释一下什么是AOT模式吗？和普通模式有什么区别。Anakin支持8位计算吗？。。有quantise的工具吗？感觉人都去哪了

I think ``` if (srcIdxOff+1 > src_count - 1 || srcIdxOff+1 < 0) continue; ```

@Zehaos 你的解释感觉Caffe已经实现了，这里是Caffe的源码 ``` template void BaseConvolutionLayer::forward_cpu_gemm(const Dtype* input, const Dtype* weights, Dtype* output, bool skip_im2col) { const Dtype* col_buff = input; if (!is_1x1_) { if (!skip_im2col) { conv_im2col_cpu(input, col_buffer_.mutable_cpu_data()); }...

@jy001227801 这个LAYER 只写了CUDA的，CPU的实现貌似还是原来的Caffe conv+group

我的想法不知道对不对？我看了下源码，貌似TF的速度快 用了多线程 work_sharder + eigen?