async def part2():
memory = intcode.init(data.strip().split(','))
perf()
mx = 0
for phases in permutations(range(5, 10), 5):
sig = await run(memory, phases)
mx = max(mx, sig)
print(f'part 2: {mx}')
# part 2: 2645740
perf()
async def main():
memory = data.strip().split(',')
memory = {i: int(memory[i]) for i in range(len(memory))}
memory['rb'] = 0
perf()
part_a = await run(memory, 1)
print(f'part 1: {part_a}')
# part 1: 1 -> 2316632620
perf()
part_b = await run(memory, 2)
print(f'part 2: {part_b}')
# part 2: 2 -> 78869
perf()
return part_a, part_b
border = 1 - border
for a, b in [(0, 0), (0, 1), (1, 0), (1, 1)]:
cm[a][b] += torch.sum((z == a).float() * (y == b).float())
cm1[a][b] += torch.sum(
(z == a).float() * (y == b).float() * border)
globalcm += cm
globalcm1 += cm1
if len(os.listdir("build")) < 100:
nextI = len(os.listdir("build"))
debug = util.torchTOpil(globalresize(x))
debug = PIL.Image.fromarray(numpy.uint8(debug))
debug.save("build/" + str(nextI) + "_x.png")
debug = (2.0 * y - 1) * border * 127 + 127
debug = debug.cpu().numpy()
debug = PIL.Image.fromarray(numpy.uint8(debug))
debug.save("build/" + str(nextI) + "_y.png")
debug = z.cpu().numpy() * 255
debug = PIL.Image.fromarray(numpy.uint8(debug))
debug.save("build/" + str(nextI) + "_z.png")
print("perf0=", util.perf(cm))
print("perf1=", util.perf(cm1))
print("bords=", util.perf(cm - cm1))
print("global result")
print("perf0=", util.perf(globalcm))
print("perf1=", util.perf(globalcm1))
print("bords=", util.perf(globalcm - globalcm1))
for a, b in [(0, 0), (0, 1), (1, 0), (1, 1)]:
stats[a][b] += torch.sum(
(z == a).float() * (y == b).float() * (1 - border))
if i < 10:
print(i, "/", nbbatchs, printloss)
if i < 1000 and i % 100 == 99:
print(i, "/", nbbatchs, printloss / 100)
printloss = torch.zeros(1).cuda()
if i >= 1000 and i % 300 == 299:
print(i, "/", nbbatchs, printloss / 300)
printloss = torch.zeros(1).cuda()
if i % 1000 == 999:
torch.save(net, "build/model.pth")
print(i, "perf", util.perf(stats))
if util.perf(stats)[0] > 96:
print("training stops after reaching high training accuracy")
os._exit(0)
else:
stats = torch.zeros(2, 2).cuda()
if i > nbbatchs * 0.1:
loss = loss * 0.5
if i > nbbatchs * 0.2:
loss = loss * 0.5
if i > nbbatchs * 0.5:
loss = loss * 0.5
if i > nbbatchs * 0.8:
loss = loss * 0.5