sheaconlon / serverless_learn Goto Github PK

Given our tight timeframe, it may be more productive to switch to a programming language we're all very familiar with and is easy to use, like Python. Then we can port it over to C++ afterwards. Sure, we'll take a performance hit (maybe not, as numpy is very efficient), but at least we'll more likely have something up.

What do you guys think @linyxy @sheaconlon? I can go ahead and start the porting tomorrow, before we meet

It looks like e1be9ac is incomplete, with lots of make errors occurring.

ToDos (from examining make errors):

• Add clear method to PeerList
• Add push_back method to PeerList
• Declare model_state and old_state identifiers
• Include using serverless_learn::Update; in worker.cc file

Here's a snapshot of some of the make errors