notwa/optim
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

.

Browse source
This commit is contained in:
Connor Olding 2017-02-18 18:43:42 -08:00
parent 7e5bb731da
commit a8952bebf5
2 changed files with 31 additions and 11 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

38
optim_nn.py
View file

@ -14,8 +14,15 @@ import sys
def lament(*args, **kwargs): def lament(*args, **kwargs):
print(*args, file=sys.stderr, **kwargs) print(*args, file=sys.stderr, **kwargs)
def log(left, right): _log_was_update = False
lament("{:>20}: {}".format(left, right)) def log(left, right, update=False):
s = "\x1B[1m {:>20}:\x1B[0m {}".format(left, right)
global _log_was_update
if update and _log_was_update:
lament('\x1B[F' + s)
else:
lament(s)
_log_was_update = update
class Dummy: class Dummy:
pass pass
@ -575,7 +582,10 @@ def run(program, args=None):
ritual = 'default', ritual = 'default',
restart_optim = False, # restarts also reset internal state of optimizer restart_optim = False, # restarts also reset internal state of optimizer
warmup = True, # train a couple epochs on gaussian noise and reset warmup = True, # train a couple epochs on gaussian noise and reset
# logging/output
log10_loss = True, # personally, i'm sick of looking linear loss values! log10_loss = True, # personally, i'm sick of looking linear loss values!
#fancy_logs = True, # unimplemented
problem = 3, problem = 3,
compare = ( compare = (
@ -629,12 +639,20 @@ def run(program, args=None):
def print_error(name, inputs, outputs, comparison=None): def print_error(name, inputs, outputs, comparison=None):
predicted = model.forward(inputs) predicted = model.forward(inputs)
err = ritual.measure(predicted, outputs) err = ritual.measure(predicted, outputs)
log(name + " loss", "{:12.6e}".format(err))
if config.log10_loss: if config.log10_loss:
log(name + " log10-loss", "{:+6.3f}".format(np.log10(err))) print(name, "{:12.6e}".format(err))
elif comparison: if comparison:
fmt = "10**({:+7.4f}) times" err10 = np.log10(err)
log("improvement", fmt.format(np.log10(comparison / err))) cmp10 = np.log10(comparison)
color = '\x1B[31m' if err10 > cmp10 else '\x1B[32m'
log(name + " log10-loss", "{:+6.3f} {}({:+6.3f})\x1B[0m".format(err10, color, err10 - cmp10))
else:
log(name + " log10-loss", "{:+6.3f}".format(err, np.log10(err)))
else:
log(name + " loss", "{:12.6e}".format(err))
if comparison:
fmt = "10**({:+7.4f}) times"
log("improvement", fmt.format(np.log10(comparison / err)))
return err return err
train_err = print_error("train", train_err = print_error("train",
@ -696,10 +714,12 @@ def run(program, args=None):
if config.log10_loss: if config.log10_loss:
fmt = "epoch {:4.0f}, rate {:10.8f}, log10-loss {:+6.3f}" fmt = "epoch {:4.0f}, rate {:10.8f}, log10-loss {:+6.3f}"
log("info", fmt.format(learner.epoch + 1, learner.rate, np.log10(avg_loss))) log("info", fmt.format(learner.epoch + 1, learner.rate, np.log10(avg_loss)),
update=True)
else: else:
fmt = "epoch {:4.0f}, rate {:10.8f}, loss {:12.6e}" fmt = "epoch {:4.0f}, rate {:10.8f}, loss {:12.6e}"
log("info", fmt.format(learner.epoch + 1, learner.rate, avg_loss)) log("info", fmt.format(learner.epoch + 1, learner.rate, avg_loss),
update=True)
measure_error() measure_error()

4
optim_nn_core.py
View file

@ -396,7 +396,7 @@ class Sum(Layer):
class Sigmoid(Layer): # aka Logistic class Sigmoid(Layer): # aka Logistic
def F(self, X): def F(self, X):
self.sig = sigmoid(X) self.sig = sigmoid(X)
return X * self.sig return self.sig
def dF(self, dY): def dF(self, dY):
return dY * self.sig * (1 - self.sig) return dY * self.sig * (1 - self.sig)
@ -404,7 +404,7 @@ class Sigmoid(Layer): # aka Logistic
class Tanh(Layer): class Tanh(Layer):
def F(self, X): def F(self, X):
self.sig = np.tanh(X) self.sig = np.tanh(X)
return X * self.sig return self.sig
def dF(self, dY): def dF(self, dY):
return dY * (1 - self.sig * self.sig) return dY * (1 - self.sig * self.sig)