def printKclosest(arr, n, x, k):
# Make a max heap of difference with
# first k elements.
pq = PriorityQueue()
for i in range(k):
pq.put((-abs(arr[i] - x), i))
# Now process remaining elements
for i in range(k, n):
diff = abs(arr[i] - x)
p, pi = pq.get()
curr = -p
# If difference with current
# element is more than root,
# then put it back.
if diff > curr:
pq.put((-curr, pi))
continue
else:
# Else remove root and insert
pq.put((-diff, i))
# Print contents of heap.
curr = 0
while (not pq.empty()):
p, q = pq.get()
curr += abs(val - arr[q])
last_ans[val] = curr
async def _dequeue_entry(self,
queue: asyncio.PriorityQueue,
timeout: int = 5):
while not self.is_shutting_down:
wrapper: DateTimePriorityWrapper = await asyncio.wait_for(
queue.get(), timeout=timeout)
if not wrapper.can_process_now():
await queue.put(wrapper)
wait_for = max(wrapper.scheduled_in_seconds() * .5, 1)
log.debug(
"it's not yet time to use this request, waiting for %d",
wait_for)
await asyncio.sleep(wait_for)
continue
return wrapper.item
raise ShuttingDown()
def beam_decode(target_tensor, decoder_hiddens, encoder_outputs=None):
'''
:param target_tensor: target indexes tensor of shape [B, T] where B is the batch size and T is the maximum length of the output sentence
:param decoder_hidden: input tensor of shape [1, B, H] for start of the decoding
:param encoder_outputs: if you are using attention mechanism you can pass encoder outputs, [T, B, H] where T is the maximum length of input sentence
:return: decoded_batch
'''
beam_width = 10
topk = 1 # how many sentence do you want to generate
decoded_batch = []
# decoding goes sentence by sentence
for idx in range(target_tensor.size(0)):
if isinstance(decoder_hiddens, tuple): # LSTM case
decoder_hidden = (decoder_hiddens[0][:, idx, :].unsqueeze(0),
decoder_hiddens[1][:, idx, :].unsqueeze(0))
else:
decoder_hidden = decoder_hiddens[:, idx, :].unsqueeze(0)
encoder_output = encoder_outputs[:, idx, :].unsqueeze(1)
# Start with the start of the sentence token
decoder_input = torch.LongTensor([[SOS_token]], device=device)
# Number of sentence to generate
endnodes = []
number_required = min((topk + 1), topk - len(endnodes))
# starting node - hidden vector, previous node, word id, logp, length
node = BeamSearchNode(decoder_hidden, None, decoder_input, 0, 1)
nodes = PriorityQueue()
# start the queue
nodes.put((-node.eval(), node))
qsize = 1
# start beam search
while True:
# give up when decoding takes too long
if qsize > 2000: break
# fetch the best node
score, n = nodes.get()
decoder_input = n.wordid
decoder_hidden = n.h
if n.wordid.item() == EOS_token and n.prevNode != None:
endnodes.append((score, n))
# if we reached maximum # of sentences required
if len(endnodes) >= number_required:
break
else:
continue
# decode for one step using decoder
decoder_output, decoder_hidden = decoder(decoder_input,
decoder_hidden,
encoder_output)
# PUT HERE REAL BEAM SEARCH OF TOP
log_prob, indexes = torch.topk(decoder_output, beam_width)
nextnodes = []
for new_k in range(beam_width):
decoded_t = indexes[0][new_k].view(1, -1)
log_p = log_prob[0][new_k].item()
node = BeamSearchNode(decoder_hidden, n, decoded_t,
n.logp + log_p, n.leng + 1)
score = -node.eval()
nextnodes.append((score, node))
# put them into queue
for i in range(len(nextnodes)):
score, nn = nextnodes[i]
nodes.put((score, nn))
# increase qsize
qsize += len(nextnodes) - 1
# choose nbest paths, back trace them
if len(endnodes) == 0:
endnodes = [nodes.get() for _ in range(topk)]
utterances = []
for score, n in sorted(endnodes, key=operator.itemgetter(0)):
utterance = []
utterance.append(n.wordid)
# back trace
while n.prevNode != None:
n = n.prevNode
utterance.append(n.wordid)
utterance = utterance[::-1]
utterances.append(utterance)
decoded_batch.append(utterances)
return decoded_batch
