Andy-P

15:10:38.984 [main] INFO  org.nd4j.nativeblas.NativeOps - Number of threads used for NativeOps: 32
15:10:48.878 [main] INFO  org.nd4j.nativeblas.Nd4jBlas - Number of threads used for BLAS: 4
15:11:58.989 [main] DEBUG org.reflections.Reflections - going to scan these urls:
jar:file:/simplex/simulator/arcadia-dealing-simulator.jar!/
15:11:59.760 [main] INFO  org.reflections.Reflections - Reflections took 766 ms to scan 1 urls, producing 29 keys and 172 values
15:12:05.823 [main] DEBUG o.n.j.a.c.impl.BasicContextPool - Creating new stream for thread: [1], device: [0]...
15:12:05.826 [main] DEBUG o.n.j.a.c.impl.BasicContextPool - Creating new stream for thread: [1], device: [0]...
15:12:05.826 [main] DEBUG o.n.j.a.c.impl.BasicContextPool - Creating new stream for thread: [1], device: [0]...
15:12:05.827 [main] DEBUG o.n.j.a.c.impl.BasicContextPool - Creating new stream for thread: [1], device: [0]...
15:12:05.827 [main] DEBUG o.n.j.a.c.impl.BasicContextPool - Creating new stream for thread: [1], device: [0]...

Andy-P

SLF4J: Failed to load class "org.slf4j.impl.StaticLoggerBinder".
SLF4J: Defaulting to no-operation (NOP) logger implementation
SLF4J: See http://www.slf4j.org/codes.html#StaticLoggerBinder for further details.
CudaGridExecutioner
Build model....
Beginning training

Epoch = 0
Batch 000 [150, 3] [150, 3, 32, 16] 
CUDA error at /projects/skymind/libnd4j/blas/cuda/NativeOps.cu:4969 code=77(<unknown>) "result" 

Andy-P

Multiple annotations found at this line:
	- Missing artifact concurrent:concurrent:jar:1.3.4
	- Missing artifact jna:jna:jar:1.0
	- Missing artifact commons-net:commons-net:jar:2.2
	- Missing artifact com.github.lucarosellini.rJava:JRI:jar:0.9
	- Missing artifact org.apache.commons:commons-lang3:jar:3.2
	- Missing artifact poi-ooxml:poi-ooxml:jar:3.7
	- Missing artifact log4j:log4j:jar:1.2.14
	- Missing artifact simplex.galaxy.l3j:l3j:jar:1.1.80
	- Missing artifact mail:mail:jar:1.0.0

Andy-P

""" Trains an agent with (stochastic) Policy Gradients on Pong. Uses OpenAI Gym. """
import numpy as np
import cPickle as pickle
import gym

# hyperparameters
H = 200 # number of hidden layer neurons
batch_size = 10 # every how many episodes to do a param update?
learning_rate = 1e-4
gamma = 0.99 # discount factor for reward

Andy-P

function generate_batch(batch_size, num_skips, skip_window, data_index)
    @assert batch_size % num_skips == 0
    @assert num_skips <= 2 * skip_window
    batch  = zeros(Int32, (batch_size)) 
    labels = zeros(Int32, (batch_size,1))
    span = 2 * skip_window + 1 
    buffer = Int[] 
    
    for _ in 1:span
        push!(buffer,data[data_index])

Andy-P

using Dates

# set_default_plot_size(20cm, 10cm)
# simple Integrate and Fire Synapse Mode
type Synapse 
    Ω::Real # membrane resistance [MΩ]
    τ::Real # membrane time constant [ms]
    rV::Real # resting membrane potential [mV]
    thV::Real # spike threshold [mV]
    sV::Real # spike voltage [mV]

Andy-P

# This code is part of a presentation on streaming analytics in Julia
# It was inspired by a number of individuals and makes use of some of their ideas
#     1. FastML.com got me thinking about inline processing after 
#        reading his great Vowpal Wabbit posts
#     2. John Lanford and his fantastic Vowpal Wabbit library. 
#        Check out his NYU video course to learn more (see below)
#     3. John Myles White's presentation on online SDG and his StreamStats.jl library 
# Thank you all!