nagadomi

# pip3 install websocket-client
import websocket
from multiprocessing import Process, Value, Lock, Event
from datetime import datetime
import dateutil.parser
import json
from time import sleep

class SFDInfo():
    def __init__(self):

nagadomi

import websocket
import json
from pprint import pprint

def on_message(ws, message):
    message = json.loads(message)
    pprint(message)

def on_error(ws, error):
    print(error)

nagadomi

# pip3 install tornado pubnub python-dateutil
# python3 -u sfd.py | tee sfd.log
from multiprocessing import Process, Value, Lock, Event
from datetime import datetime
import dateutil.parser
from pubnub.callbacks import SubscribeCallback
from pubnub.enums import PNStatusCategory
from pubnub.pnconfiguration import PNConfiguration
from pubnub.pubnub_tornado import PubNubTornado
from pubnub.pnconfiguration import PNReconnectionPolicy

nagadomi

/*
 * L4D2特殊感染者BOTのプレイスタイルをカスタマイズする実験的なプラグイン
 *
 * 出現する特殊感染者BOTの2/3のだけ改変します
 *
 * キー入力をシミュレートすることでBOTを操作するため
 * システムはデフォルトのまま!!
 *
 *   addons/sourcemod/scripting
 * において

nagadomi

require 'nn'

local N = 1000
local FEAT = 64
local S = 2

function nn.SpatialConvolution:reset()
   self.weight:uniform(0, 1)
   self.bias:zero()
end

nagadomi

require 'cutorch'
require 'cunn'
require 'cudnn'
require 'sys'

-- WINOGRAD benchmark
-- required: cuDNN v5, cudnn.torch R5 branch

function create_model(ch)
   local model = nn.Sequential()

nagadomi

require 'cutorch'
require 'cunn'
require 'cudnn'
require 'sys'

-- WINOGRAD benchmark
-- required: cuDNN v5, cudnn.torch R5 branch

function create_model(ch) -- simple 3x3 conv model
   local model = nn.Sequential()

nagadomi

// __attribute__((aligned(16))) float input[n], weight[n];
float dot_sse(float *input, float *weight, int n)
{
	__attribute__((aligned(16))) float mm[4] = {0};
	__m128 x, u;
	int pk_lp = (n & 0xfffffffc);
	int i;
	float sum = 0;
	u = _mm_setzero_ps();
	for (i = 0; i < pk_lp; i += 4) {

nagadomi

name: "srcnn"
layer {
  name: "input"
  type: "Input"
  top: "input"
  input_param { shape: { dim: 1 dim: 3 dim: 32 dim: 32 } }
}
layer {
  name: "conv1_layer"
  type: "Convolution"

nagadomi

require 'nn'

torch.setdefaulttensortype("torch.FloatTensor")

upconv = nn.SpatialFullConvolution(1, 1, 2, 2, 2, 2, 0, 0)
input = torch.Tensor({{1, 10},{100, 1000}}):reshape(1, 1, 2, 2) -- (batch_size, input_dim, height, width)
weight = torch.Tensor({{1, 2}, {3, 4}}) -- 1x1x2x2 filter
bias = torch.Tensor({0.5})
upconv.weight:copy(weight)
upconv.bias:copy(bias)