deterministic_synthesis=False,
forbidden_tokens=None,
forced_biases=None,
):
to_generate = (ar_mask.sum(0) > 0).nonzero()
deterministic_synthesis=False,
forbidden_tokens=None,
forced_biases=None,
):
to_generate = (ar_mask.sum(0) > 0).nonzero()
if to_generate.min() > 0:
self(
BracketedSequence(input, 0, to_generate.min())
) # Needed to initialize the model's cache
for s in range(to_generate.min(), to_generate.max() + 1):
output = self(BracketedSequence(input, s, 1)).x
if to_generate.min() > 0:
self(
BracketedSequence(input, 0, to_generate.min())
) # Needed to initialize the model's cache
for s in range(to_generate.min(), to_generate.max() + 1):
output = self(BracketedSequence(input, s, 1)).x
if forbidden_tokens is not None:
logits = logits.masked_fill(forbidden_tokens, float("-inf"))
if forbidden_tokens is not None:
logits = logits.masked_fill(forbidden_tokens, float("-inf"))
if forced_biases is not None:
logits = logits + forced_biases[None, :]
if forced_biases is not None:
logits = logits + forced_biases[None, :]
input[:, s] = ar_mask[:, s] * t_next + (1 - ar_mask[:, s]) * input[:, s]
def record_attention(self, v=True):
input[:, s] = ar_mask[:, s] * t_next + (1 - ar_mask[:, s]) * input[:, s]
def record_attention(self, v=True):