yahoo / CaffeOnSpark

Distributed deep learning on Hadoop and Spark clusters.
Apache License 2.0
1.27k stars 358 forks source link

err “java.lang.UnsupportedOperationException: empty.reduceLeft” #297

Open Zzmc opened 6 years ago

Zzmc commented 6 years ago

hello, After i run 100 iters, there is an err "java.lang.UnsupportedOperationException: empty.reduceLeft", it look like somthing wrong with my DataFrame data, but the same data can work well in another cluster. The different with two clusters is that: Working well cluster:
os:ubuntu | num_cpus/per_computer:1| devices/per_computer:3(GTX1080)| total_divices:9 Wrong cluster: os:centos| num_cpus/per_computer:2| devices/per_computer:8(GTX1080)| total_devices:16 can you give me some suggestion? Thanks.