Hi, tf hackers. Is there any method that I can convert DT_FLOAT to DT_HALF/DF_BFLOAT16 in my graph? I am thinking of accerlate the performance for Forward path only.
Hello, I've proposed a project for chromosomes segmentation : https://github.com/jeanpat/DeepFISH . Up to now there are datasets (82146 images+ground-truth labels) and python notebooks (way to generate datasets, scikit-image based). The notebooks could be better: the size of the datasets are limited by the amount of RAM. The datasets are waiting to train a classifier.
@jeanpat what kind of GPU card did you use?
A small nvdia gpu: lspci -k | grep -A 2 -E "(VGA|3D)"
00:02.0 VGA compatible controller: Intel Corporation 4th Gen Core Processor Integrated Graphics Controller (rev 06)
Subsystem: CLEVO/KAPOK Computer Device 5000
Kernel driver in use: i915--
01:00.0 3D controller: NVIDIA Corporation GK208M [GeForce GT 740M] (rev ff)
Kernel modules: nouveau, nvidia_drm, nvidia
03:00.0 Network controller: Intel Corporation Wireless 7260 (rev 73)
@leoncamel up to now, I don't know if I can train a cnn with the proposed data.