def main(args):
config = get_config(args.model)
net = load_model_path(args.model, config)
t_len = 200
with torch.no_grad():
net.reset()
ins = torch.zeros((12))
for j in range(t_len):
def main(args):
config = get_config(args.model, to_bunch=True)
net = load_model_path(args.model, config)
data, loss = test_model(net, config, n_tests=150)
ys = [[] for i in range(net.args.T)]
ys_stim = [[] for i in range(net.args.T)]
for d in data:
context, idx, trial, x, y, out, loss = d
y_ready, y_set, y_go = trial.rsg
y_prod = np.argmax(out >= 1)
t_y = y_set - y_ready
t_p = y_prod - y_set
t_ym = y_go - y_set
ys[context].append((t_y, t_p))
ys_stim[context].append((t_y, t_ym))
ys = [list(zip(*np.array(y))) for y in ys]
ys_stim = [list(zip(*np.array(y))) for y in ys_stim]
for i in range(net.args.T):
plt.scatter(ys_stim[i][0],
ys_stim[i][1],
marker='o',
c='black',
s=20,
edgecolors=cols[i])
plt.scatter(ys[i][0], ys[i][1], color=cols[i], s=15)
plt.xlabel('desired t_p')
plt.ylabel('produced t_p')
plt.show()
def main(args):
config = get_config(args.model, to_bunch=True)
net = load_model_path(args.model, config)
if len(args.dataset) == 0:
args.dataset = config.dataset
n_reps = 1000
# don't show these contexts
context_filter = []
_, loader = create_loaders(args.dataset,
config,
split_test=False,
test_size=n_reps,
context_filter=context_filter)
x, y, trials = next(iter(loader))
A = get_states(net, x)
t_type = type(trials[0])
if t_type == RSG:
pca_rsg(args, A, trials, n_reps)
elif t_type in [DelayProAnti, MemoryProAnti]:
pca_dmpa(args, A, trials, n_reps)
def main(args):
files = os.scandir(args.dir)
for fn in os.scandir(args.dir):
if os.path.basename(fn).startswith('checkpoints'):
ckpt_folder = fn
elif os.path.basename(fn).startswith('config'):
config_path = fn
models = os.scandir(ckpt_folder)
with open(config_path, 'r') as f:
config = json.load(f)
norms = []
for ix, s in enumerate(models):
print(s.path)
model = load_model_path(s.path, config)
# J = model.reservoir.J.weight.data.numpy()
Wf = model.W_f.weight.data.numpy()
if ix == 0:
# last_J = J
last_Wf = Wf
continue
# dif = J - last_J
dif = Wf - last_Wf
norms.append(np.linalg.norm(dif))
# last_J = J
last_Wf = Wf
plt.plot(norms)
plt.savefig('figures/weight_changes.png')
config = get_config(args.model)
if args.noise != 0:
J = model['W_f.weight']
v = J.std()
shp = J.shape
model['W_f.weight'] += torch.normal(0, v * .5, shp)
J = model['W_ro.weight']
v = J.std()
shp = J.shape
model['W_ro.weight'] += torch.normal(0, v * .5, shp)
config = fill_undefined_args(args, config, overwrite_none=True)
net = load_model_path(args.model, config=config)
if args.test_all:
_, loss2 = test_model(net, config)
print('avg summed loss (all):', loss2)
if not args.no_plot:
data, loss = test_model(net, config, n_tests=6)
print('avg summed loss (plotted):', loss)
run_id = '/'.join(args.model.split('/')[-3:-1])
fig, ax = plt.subplots(2, 3, sharex=True, sharey=True, figsize=(12, 7))
if 'goals' in config.dataset:
p_fn = get_potential(config)
dt = dt[dt.mnoise == 6]
intervals = [{}, {}, {}]
for j, dset in enumerate(dsets):
subset = dt[dt.dset == dset]
for iterr in range(len(subset)):
job_id = subset.iloc[iterr].slurm_id
model_folder = os.path.join('..', 'logs', run_id, str(job_id))
model_path = os.path.join(model_folder, 'model_best.pth')
config = get_config(model_path, ctype='model', to_bunch=True)
config.m_noise = 0
config.dataset = dset_map[config.dataset]
net = load_model_path(model_path, config=config)
data, loss = test_model(net, config, n_tests=200, dset_base='../')
dset = load_rb(os.path.join('..', config.dataset))
distr = {}
for k in range(len(data)):
dset_idx, x, _, z, _ = data[k]
r, s, g = dset[dset_idx][2]
t_first = torch.nonzero(z >= 1)
if len(t_first) > 0:
t_first = t_first[0, 0]
else:
t_first = len(x)
with open(args.model, 'rb') as f:
model = torch.load(f)
if args.noise != 0:
J = model['W_f.weight']
v = J.std()
shp = J.shape
model['W_f.weight'] += torch.normal(0, v * .5, shp)
J = model['W_ro.weight']
v = J.std()
shp = J.shape
model['W_ro.weight'] += torch.normal(0, v * .5, shp)
net = load_model_path(args.model, params={'dset': args.dataset, 'out_act': args.out_act})
dset = load_rb(args.dataset)
data = test_model(net, dset, n_tests=0)
run_id = '/'.join(args.model.split('/')[-3:-1])
fig, ax = plt.subplots(3,4,sharex=True, sharey=True, figsize=(12,7))
for i, ax in enumerate(fig.axes):
ix, x, y, z, loss = data[i]
xr = np.arange(len(x))
ax.axvline(x=0, color='dimgray', alpha = 1)
ax.axhline(y=0, color='dimgray', alpha = 1)
ax.grid(True, which='major', lw=1, color='lightgray', alpha=0.4)
ax.spines['top'].set_visible(False)
