def get_data_history(self,
idx,
t_sep,
fillings,
model_args=None,
label=None):
res = {}
# max_len_commands = 0
# len_path = len(t_sep)
if model_args is None:
model_args = self.model_args
pad_len = 0
t_sep.extend([torch.empty(0, 14)] * pad_len)
# print("t_sep",len(t_sep))
t_grouped = [
SVGTensor.from_data(torch.cat(t_sep, dim=0),
PAD_VAL=self.PAD_VAL).add_eos().add_sos()
]
t_normal = []
for t in t_sep:
s = SVGTensor.from_data(t, PAD_VAL=self.PAD_VAL)
# print(1,len(s.commands))
j = s.add_eos().add_sos().pad(seq_len=20 + 2)
# print(2,len(s.commands))
t_normal.append(j)
for arg in set(model_args):
if "_grouped" in arg:
arg_ = arg.split("_grouped")[0]
t_list = t_grouped
else:
arg_ = arg
t_list = t_normal
if arg_ == "tensor":
res[arg] = t_list
if arg_ == "commands":
res[arg] = torch.stack([t.cmds() for t in t_list])
if arg_ == "args_rel":
res[arg] = torch.stack([t.get_relative_args() for t in t_list])
if arg_ == "args":
res[arg] = torch.stack([t.args() for t in t_list])
if "filling" in model_args:
res["filling"] = torch.stack(
[torch.tensor(t.filling) for t in t_sep]).unsqueeze(-1)
if "label" in model_args:
res["label"] = label
return res
def get_data(self, idx, t_sep, fillings, model_args=None, label=None):
res = {}
# max_len_commands = 0
# len_path = len(t_sep)
if model_args is None:
model_args = self.model_args
if len(t_sep) > self.MAX_NUM_GROUPS:
return None
pad_len = max(self.MAX_NUM_GROUPS - len(t_sep), 0)
t_sep.extend([torch.empty(0, 14)] * pad_len)
t_grouped = [
SVGTensor.from_data(torch.cat(t_sep, dim=0),
PAD_VAL=self.PAD_VAL).add_eos().add_sos().pad(
seq_len=self.MAX_TOTAL_LEN + 2)
]
t_normal = []
for t in t_sep:
s = SVGTensor.from_data(t, PAD_VAL=self.PAD_VAL)
if len(s.commands) > self.MAX_SEQ_LEN:
return None
t_normal.append(
s.add_eos().add_sos().pad(seq_len=self.MAX_SEQ_LEN + 2))
for arg in set(model_args):
if "_grouped" in arg:
arg_ = arg.split("_grouped")[0]
t_list = t_grouped
else:
arg_ = arg
t_list = t_normal
if arg_ == "tensor":
res[arg] = t_list
if arg_ == "commands":
res[arg] = torch.stack([t.cmds() for t in t_list])
if arg_ == "args_rel":
res[arg] = torch.stack([t.get_relative_args() for t in t_list])
if arg_ == "args":
res[arg] = torch.stack([t.args() for t in t_list])
if "filling" in model_args:
res["filling"] = torch.stack(
[torch.tensor(t.filling) for t in t_sep]).unsqueeze(-1)
if "label" in model_args:
res["label"] = label
return res
def visualize(self, model, output, train_vars, step, epoch, summary_writer,
visualization_dir):
device = next(model.parameters()).device
# Reconstruction
for i, data in enumerate(train_vars.x_inputs_train):
model_args = batchify((data[key] for key in self.model_args),
device)
commands_y, args_y = model.module.greedy_sample(*model_args)
tensor_pred = SVGTensor.from_cmd_args(commands_y[0].cpu(),
args_y[0].cpu())
try:
svg_path_sample = SVG.from_tensor(
tensor_pred.data, viewbox=Bbox(256),
allow_empty=True).normalize().split_paths().set_color(
"random")
except:
continue
tensor_target = data["tensor_grouped"][0].copy().drop_sos().unpad()
svg_path_gt = SVG.from_tensor(
tensor_target.data, viewbox=Bbox(
256)).normalize().split_paths().set_color("random")
img = make_grid([svg_path_sample,
svg_path_gt]).draw(do_display=False,
return_png=True,
fill=False,
with_points=False)
summary_writer.add_image(f"reconstructions_train/{i}",
TF.to_tensor(img), step)
def decode(z):
commands_y, args_y = model.greedy_sample(z=z)
tensor_pred = SVGTensor.from_cmd_args(commands_y[0].cpu(), args_y[0].cpu())
svg_path_sample = SVG.from_tensor(tensor_pred.data, viewbox=Bbox(256))
return svg_path_sample