+ train_sequences = expr.generate_sequences(
+ nb_train_samples,
+ nb_variables=nb_variables,
+ length=sequence_length,
+ # length=2 * sequence_length,
+ # randomize_length=True,
+ )
+ test_sequences = expr.generate_sequences(
+ nb_test_samples,
+ nb_variables=nb_variables,
+ length=sequence_length,
+ )
+ self.char2id = dict(
+ [
+ (c, n)
+ for n, c in enumerate(
+ set("#" + "".join(train_sequences + test_sequences))
+ )
+ ]
+ )
+ self.id2char = dict([(n, c) for c, n in self.char2id.items()])
+
+ self.filler, self.space = self.char2id["#"], self.char2id[" "]
+
+ len_max = max([len(x) for x in train_sequences])
+ self.train_input = torch.cat(
+ [
+ torch.tensor(
+ [
+ [self.char2id[c] for c in s + "#" * (len_max - len(s))]
+ for s in train_sequences
+ ]
+ )
+ ],
+ 0,
+ ).to(device)
+
+ len_max = max([len(x) for x in test_sequences])
+ self.test_input = torch.cat(
+ [
+ torch.tensor(
+ [
+ [self.char2id[c] for c in s + "#" * (len_max - len(s))]
+ for s in test_sequences
+ ]
+ )
+ ],
+ 0,
+ ).to(device)
+
+ self.nb_codes = max(self.train_input.max(), self.test_input.max()) + 1
+
+ def batches(self, split="train", nb_to_use=-1, desc=None):
+ assert split in {"train", "test"}
+ input = self.train_input if split == "train" else self.test_input
+ if nb_to_use > 0:
+ input = input[:nb_to_use]
+ if desc is None:
+ desc = f"epoch-{split}"
+ for batch in tqdm.tqdm(
+ input.split(self.batch_size), dynamic_ncols=True, desc=desc
+ ):
+ if split == "train":
+ last = (batch != self.filler).max(0).values.nonzero().max() + 1
+ batch = batch[:, :last]
+ yield batch
+
+ def vocabulary_size(self):
+ return self.nb_codes
+
+ def seq2str(self, s):
+ return "".join([self.id2char[k.item()] for k in s])
+
+ def produce_results(self, n_epoch, model):
+ with torch.autograd.no_grad():
+ t = model.training
+ model.eval()
+
+ def compute_nb_correct(input):
+ result = input.clone()
+ ar_mask = (result == self.space).long().cumsum(dim=1).clamp(max=1)
+ result = (1 - ar_mask) * result + ar_mask * self.filler