def new_update(x, new_x):
if is_one_of(x, params) and self._do_layer_adaptation(x):
dx = new_x - x
lr_t = K.clip(self.learning_rate, K.epsilon(), 1e10)
x_norm = tf.norm(x)
g_norm = tf.norm(dx / lr_t)
ratio = K.switch(
x_norm > 0.0,
K.switch(g_norm > K.epsilon(), x_norm / g_norm, 1.0),
1.0)
new_x = x + dx * ratio
return old_update(x, new_x)
def new_update(x, new_x):
if x is var and self._do_layer_adaptation(x):
dx = new_x - x
lr_t = self._decayed_lr(x.dtype.base_dtype)
lr_t = K.clip(lr_t, K.epsilon(), 1e10)
x_norm = tf.norm(x)
g_norm = tf.norm(dx / lr_t)
ratio = K.switch(
x_norm > 0.0,
K.switch(g_norm > K.epsilon(), x_norm / g_norm, 1.0),
1.0)
new_x = x + dx * ratio
return old_update(x, new_x)
def _resource_apply(self, grad, var, indices=None):
op = super(NewOptimizer, self)._resource_apply(grad, var, indices)
k, alpha = self.steps_per_slow_update, self.slow_step_size
cond = K.equal(self.iterations % k, 0)
slow_var = self.get_slot(var, 'slow_var')
slow_var_t = slow_var + alpha * (var - slow_var)
with tf.control_dependencies([op]):
slow_update = K.update(slow_var,
K.switch(cond, slow_var_t, slow_var))
with tf.control_dependencies([slow_update]):
copy_update = K.update(var, K.switch(cond, slow_var, var))
return copy_update
def compute_position_ids(self, inputs):
"""T5的相对位置分桶（直接翻译自官方T5源码）
"""
q, v = inputs
# 计算位置差
q_idxs = K.arange(0, K.shape(q)[1], dtype='int32')
q_idxs = K.expand_dims(q_idxs, 1)
v_idxs = K.arange(0, K.shape(v)[1], dtype='int32')
v_idxs = K.expand_dims(v_idxs, 0)
pos_ids = v_idxs - q_idxs
# 后处理操作
num_buckets, max_distance = self.input_dim, self.max_distance
ret = 0
n = -pos_ids
if self.bidirectional:
num_buckets //= 2
ret += K.cast(K.less(n, 0), 'int32') * num_buckets
n = K.abs(n)
else:
n = K.maximum(n, 0)
# now n is in the range [0, inf)
max_exact = num_buckets // 2
is_small = K.less(n, max_exact)
val_if_large = max_exact + K.cast(
K.log(K.cast(n, K.floatx()) / max_exact) /
np.log(max_distance / max_exact) * (num_buckets - max_exact),
'int32',
)
val_if_large = K.minimum(val_if_large, num_buckets - 1)
ret += K.switch(is_small, n, val_if_large)
return ret
def _resource_apply(self, grad, var, indices=None):
# 更新判据
cond = K.equal(self.iterations % self.grad_accum_steps, 0)
# 获取梯度
ag = self.get_slot(var, 'ag')
old_update = K.update
def new_update(x, new_x):
new_x = K.switch(cond, new_x, x)
return old_update(x, new_x)
K.update = new_update
ag_t = ag / self.grad_accum_steps
op = super(NewOptimizer, self)._resource_apply(ag_t, var)
K.update = old_update
# 累积梯度
with tf.control_dependencies([op]):
ag_t = K.switch(cond, K.zeros_like(ag), ag)
with tf.control_dependencies([K.update(ag, ag_t)]):
if indices is None:
ag_t = K.update(ag, ag + grad)
else:
ag_t = self._resource_scatter_add(ag, indices, grad)
return ag_t
def get_updates(self, loss, params):
updates = super(NewOptimizer, self).get_updates(loss, params)
k, alpha = self.steps_per_slow_update, self.slow_step_size
cond = K.equal(self.iterations % k, 0)
slow_vars = [
K.zeros(K.int_shape(p),
dtype=K.dtype(p),
name='slow_var_%s' % i) for i, p in enumerate(params)
]
with tf.control_dependencies(updates):
slow_updates = [
K.update(q, K.switch(cond, q + alpha * (p - q), q))
for p, q in zip(params, slow_vars)
]
with tf.control_dependencies(slow_updates):
copy_updates = [
K.update(p, K.switch(cond, q, p))
for p, q in zip(params, slow_vars)
]
return copy_updates
def new_update(x, new_x):
new_x = K.switch(cond, new_x, x)
return old_update(x, new_x)
