async def send_command_help(self, command):
doc = Embed.from_dict({
'title': f'{self.clean_prefix}{command.name} {command.signature}',
'description': command.help,
'color': get_color().value
})
if command.aliases:
doc.add_field(name='Aliases',
inline=True,
value=', '.join(command.aliases))
if command.cog:
doc.add_field(name='Category',
inline=True,
value=command.cog.qualified_name)
if command._max_concurrency and getattr(
command._max_concurrency, 'per', None) != BucketType.default:
doc.add_field(name='Concurrency',
value='**%s** time(s) per **%s**' %
(command._max_concurrency.number,
command._max_concurrency.per.name),
inline=True)
await self.get_destination().send(embed=doc)
async def meme(self, ctx, site=None):
"""Fetch a meme from a defined or a random resource (always an image)"""
if not site:
site = random.choice([*self._generators.keys()])
try:
pic, info = self._generators[site]()
except KeyError:
raise BadArgument("Not a valid site from the list of sites")
await ctx.send(embed=Embed.from_dict({
'image': {
'url': pic
},
'title':
info.get('title', ''),
'author': {
'name': info.get('author', '')
},
'timestamp':
call(getattr(info.get('creation'), 'isoformat', None), ''),
'color':
get_color().value
}))
def projection_dim_exp_fig():
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
for ds in ['mini', 'tiered']:
ax = plt.subplot(111)
with open(f'exp_projection_dim_{ds}.pickle', 'rb') as handle:
exp_dict = pickle.load(handle)
x = exp_dict['args']
y_lim = np.inf
for part in ['val', 'test']:
name = f'{part}'
res, _ = exp_dict[name]
plt.plot(x, res, label=name, color=get_color(name), linewidth=2)
y_lim = min(y_lim, np.min(res))
ax.legend(loc='lower center', fancybox=True, shadow=True, ncol=3)
# plt.ylim(bottom=y_lim - 1)
plt.grid()
fig = plt.figure(1)
plt.xlabel("reduced dimension")
plt.ylabel("accuracy")
plt.savefig(f"projection-dim-exp-{ds}.png")
plt.close(fig)
def n_query_exp_fig():
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
for ds in ['mini', 'tiered']:
with open(f'exp_n_queries_{ds}.pickle', 'rb') as handle:
exp_dict = pickle.load(handle)
x = np.array(list(exp_dict.keys()))
values = np.array(list(exp_dict.values()))
y = values[:, 0, :]
ax = plt.subplot(111)
exp_name = ['simple', 'sub', 'ica', 'ica & bkm', 'ica & msp']
y_lim = None
for n in [*range(y.shape[1])]:
# plt.plot(x, y[:, n].reshape(-1, ), label=exp_name[n])
plt.plot(x[1:], y[:, n].reshape(-1, )[1:], label=exp_name[n], color=get_color(exp_name[n]), linewidth=2)
if exp_name[n] == 'simple':
y_lim = np.min(y[:, n])
ax.legend(loc='lower center', fancybox=True, shadow=True, ncol=3)
plt.ylim(bottom=y_lim - 3)
plt.grid()
fig = plt.figure(1)
plt.xlabel("queries")
plt.ylabel("accuracy")
plt.savefig(f"num-query-exp-{ds}.png")
plt.close(fig)
def noise_exp_fig():
import matplotlib
matplotlib.use('Agg')
import matplotlib.pyplot as plt
for ds in ['mini', 'tiered']:
with open(f'exp_noise_semi_{ds}.pickle', 'rb') as handle:
exp_dict = pickle.load(handle)
x = np.array(list(exp_dict.keys()))
values = np.array(list(exp_dict.values()))
y = values[:, 0, :]
x_plot = range(x.shape[0])
ax = plt.subplot(111)
exp_name = ['ica', 'ica & bkm', 'ica & msp']
y_lim = None
for n in [*range(y.shape[1])]:
plt.plot(x_plot, y[:, n].reshape(-1, ), label=exp_name[n], color=get_color(exp_name[n]), linewidth=2)
if exp_name[n] == 'ica & msp':
y_lim = np.max(y[:, n])
if ds == 'mini':
compare_works = {
'lst': [70.1, 68.0, 66.00, 64.0, 62.4, 60.8, 60.4, 60.0],
'tpn': [62.7, 62.4, 61.85, 61.3, 61.00, 60.7, 59.95, 59.2],
'skm': [62.1, 61.9, 61.40, 60.9, 60.45, 60.0, 59.30, 58.6],
}
else:
compare_works = {
'lst': [77.7, 76.25, 74.88, 73.5, 72.10, 70.7, 69.35, 68.0],
'tpn': [72.1, 72.00, 71.75, 71.5, 71.20, 70.9, 69.65, 68.4],
'skm': [68.6, 67.50, 67.25, 67.0, 66.10, 65.2, 64.88, 64.0],
}
for k, v in compare_works.items():
plt.plot(x_plot, v, label=k, color=get_color(k), linewidth=2)
ax.legend(loc='upper center', fancybox=True, shadow=True, ncol=3)
plt.ylim(top=y_lim + 3)
plt.xticks(x_plot, x)
plt.grid()
fig = plt.figure(1)
plt.xlabel("distracting classes")
plt.ylabel("accuracy")
plt.savefig(f"exp-noise-semi-{ds}.png")
plt.close(fig)
def __init__(self, cog: Cog, prefix: str, *, per_page=3):
if not isinstance(cog, Cog):
raise TypeError("a valid cog wasn't passed")
self._prefix = prefix
self._doc = Embed(title=cog.qualified_name,
description=cog.description,
color=get_color())
super().__init__(cog.get_commands(), per_page=per_page)
def visualize_label(label):
color = get_color()
mask = np.zeros(label.shape + (3, ))
obj_id = np.unique(label)
# embed()
for obj in obj_id:
if obj > 0:
mask[label == obj] = color[int(obj)]
return mask
def __init__(self, group: Group, prefix: str, *, per_page=3):
if not isinstance(group, Group):
raise TypeError("a valid group wasn't passed")
self._prefix = prefix
self._doc = Embed(title=f'{prefix}{group.name} {group.signature}',
description=format_description(group.help),
color=get_color())
super().__init__(self._flatten_commands(group.commands),
per_page=per_page)
def visualize(self, label, tracker_id):
# only for inference
color = get_color()
masks = torch.zeros(*label.size(), 3, dtype=torch.uint8)
labels = torch.zeros_like(label)
max_id = torch.max(label).item()
for oid in range(max_id + 1):
masks[label == oid] = torch.tensor(color[tracker_id[oid]])
labels[label == oid] = torch.tensor([tracker_id[oid] + 1])
masks = masks * (self.fgmask > 0.5).cpu().squeeze().unsqueeze(-1)
labels = labels.float() * (self.fgmask > 0.5).cpu().squeeze().float()
# embed()
return masks, labels.unsqueeze(-1)
async def send_bot_help(self, mapping):
doc = Embed(color=get_color())
for cog, commands in mapping.items():
command_list = '\n'.join(f'{self.clean_prefix}{cmd}'
for cmd in commands)
if not cog:
doc.description = command_list
else:
doc.add_field(name=cog.qualified_name,
value=command_list,
inline=False)
await self.get_destination().send(embed=doc)
async def send_error_message(self, error):
await self.context.send(
embed=Embed(description=error, color=get_color().value))
