def test_stick_transition():
K = 3
D = 2
M = 0
model = HMM(K=K, D=D, M=M, transition='sticky', observation='gaussian', transition_kwargs=dict(alpha=1, kappa=100))
print("alpha = {}, kappa = {}".format(model.transition.alpha, model.transition.kappa))
data = torch.tensor([[2,3], [4,5], [6, 7]], dtype=torch.float64)
log_prob = model.log_likelihood(data)
print("log_prob", log_prob)
samples_z, samples_x = model.sample(10)
samples_log_prob = model.log_probability(samples_x)
print("sample log_prob", samples_log_prob)
print("model transition\n", model.transition.stationary_transition_matrix)
model1 = HMM(K=K, D=D, M=M, transition='sticky', observation='gaussian', transition_kwargs=dict(alpha=1, kappa=0.1))
print("Before fit")
print("model1 transition\n", model1.transition.stationary_transition_matrix)
losses, opt = model1.fit(samples_x)
print("After fit")
print("model1 transition\n", model1.transition.stationary_transition_matrix)
model2 = HMM(K=K, D=D, M=M, transition='sticky', observation='gaussian', transition_kwargs=dict(alpha=1, kappa=20))
print("Before fit")
print("model2 transition\n", model2.transition.stationary_transition_matrix)
losses, opt = model2.fit(samples_x)
print("After fit")
print("model2 transition\n", model2.transition.stationary_transition_matrix)
model3 = HMM(K=K, D=D, M=M, transition='sticky', observation='gaussian', transition_kwargs=dict(alpha=1, kappa=100))
print("Before fit")
print("model3 transition\n", model3.transition.stationary_transition_matrix)
losses, opt = model3.fit(samples_x)
print("After fit")
print("model3 transition\n", model3.transition.stationary_transition_matrix)
def main(job_name, downsample_n, filter_traj, load_model, load_model_dir, load_opt_dir,
transition, sticky_alpha, sticky_kappa, acc_factor, k, x_grids, y_grids, n_x, n_y,
train_model, pbar_update_interval, video_clips, torch_seed, np_seed,
list_of_num_iters, list_of_lr, sample_t, quiver_scale):
if job_name is None:
raise ValueError("Please provide the job name.")
K = k
sample_T = sample_t
video_clip_start, video_clip_end = [int(x) for x in video_clips.split(",")]
list_of_num_iters = [int(x) for x in list_of_num_iters.split(",")]
list_of_lr = [float(x) for x in list_of_lr.split(",")]
assert len(list_of_num_iters) == len(list_of_lr), "Length of list_of_num_iters must match length of list-of_lr."
for lr in list_of_lr:
if lr > 1:
raise ValueError("Learning rate should not be larger than 1!")
repo = git.Repo('.', search_parent_directories=True) # SocialBehaviorectories=True)
repo_dir = repo.working_tree_dir # SocialBehavior
torch.manual_seed(torch_seed)
np.random.seed(np_seed)
########################## data ########################
data_dir = repo_dir + '/SocialBehaviorptc/data/trajs_all'
trajs = joblib.load(data_dir)
traj = trajs[36000*video_clip_start:36000*video_clip_end]
traj = downsample(traj, downsample_n)
if filter_traj:
traj = filter_traj_by_speed(traj, q1=0.99, q2=0.99)
data = torch.tensor(traj, dtype=torch.float64)
######################### model ####################
# model
D = 4
M = 0
Df = 4
if load_model:
print("Loading the model from ", load_model_dir)
model = joblib.load(load_model_dir)
tran = model.observation.transformation
K = model.K
n_x = len(tran.x_grids) - 1
n_y = len(tran.y_grids) - 1
acc_factor = tran.transformations_a[0].acc_factor
else:
print("Creating the model...")
bounds = np.array([[ARENA_XMIN, ARENA_XMAX], [ARENA_YMIN, ARENA_YMAX],
[ARENA_XMIN, ARENA_XMAX], [ARENA_YMIN, ARENA_YMAX]])
# grids
if x_grids is None:
x_grid_gap = (ARENA_XMAX - ARENA_XMIN) / n_x
x_grids = [ARENA_XMIN + i * x_grid_gap for i in range(n_x + 1)]
else:
x_grids = [float(x) for x in x_grids.split(",")]
n_x = len(x_grids) - 1
if y_grids is None:
y_grid_gap = (ARENA_YMAX - ARENA_YMIN) / n_y
y_grids = [ARENA_YMIN + i * y_grid_gap for i in range(n_y + 1)]
else:
y_grids = [float(x) for x in y_grids.split(",")]
n_y = len(y_grids) - 1
if acc_factor is None:
acc_factor = downsample_n * 10
tran = GridTransformation(K=K, D=D, x_grids=x_grids, y_grids=y_grids, unit_transformation="direction",
Df=Df, feature_vec_func=f_corner_vec_func, acc_factor=acc_factor)
obs = ARTruncatedNormalObservation(K=K, D=D, M=M, lags=1, bounds=bounds, transformation=tran)
if transition == 'sticky':
transition_kwargs = dict(alpha=sticky_alpha, kappa=sticky_kappa)
else:
transition_kwargs = None
model = HMM(K=K, D=D, M=M, transition=transition, observation=obs, transition_kwargs=transition_kwargs)
model.observation.mus_init = data[0] * torch.ones(K, D, dtype=torch.float64)
# save experiment params
exp_params = {"job_name": job_name,
'downsample_n': downsample_n,
"load_model": load_model,
"load_model_dir": load_model_dir,
"load_opt_dir": load_opt_dir,
"transition": transition,
"sticky_alpha": sticky_alpha,
"sticky_kappa": sticky_kappa,
"acc_factor": acc_factor,
"K": K,
"n_x": n_x,
"n_y": n_y,
"x_grids": x_grids,
"y_grids": y_grids,
"train_model": train_model,
"pbar_update_interval": pbar_update_interval,
"list_of_num_iters": list_of_num_iters,
"list_of_lr": list_of_lr,
"video_clip_start": video_clip_start,
"video_clip_end": video_clip_end,
"sample_T": sample_T}
print("Experiment params:")
print(exp_params)
rslt_dir = addDateTime("rslts/" + job_name)
rslt_dir = os.path.join(repo_dir, rslt_dir)
if not os.path.exists(rslt_dir):
os.makedirs(rslt_dir)
print("Making result directory...")
print("Saving to rlst_dir: ", rslt_dir)
with open(rslt_dir+"/exp_params.json", "w") as f:
json.dump(exp_params, f, indent=4, cls=NumpyEncoder)
# compute memories
masks_a, masks_b = tran.get_masks(data[:-1])
feature_vecs_a = f_corner_vec_func(data[:-1, 0:2])
feature_vecs_b = f_corner_vec_func(data[:-1, 2:4])
m_kwargs_a = dict(feature_vecs=feature_vecs_a)
m_kwargs_b = dict(feature_vecs=feature_vecs_b)
##################### training ############################
if train_model:
print("start training")
list_of_losses = []
if load_opt_dir != "":
opt = joblib.load(load_opt_dir)
else:
opt = None
for i, (num_iters, lr) in enumerate(zip(list_of_num_iters, list_of_lr)):
losses, opt = model.fit(data, optimizer=opt, method='adam', num_iters=num_iters, lr=lr,
masks=(masks_a, masks_b), pbar_update_interval=pbar_update_interval,
memory_kwargs_a=m_kwargs_a, memory_kwargs_b=m_kwargs_b)
list_of_losses.append(losses)
checkpoint_dir = rslt_dir + "/checkpoint_{}".format(i)
if not os.path.exists(checkpoint_dir):
os.makedirs(checkpoint_dir)
print("Creating checkpoint_{} directory...".format(i))
# save model and opt
joblib.dump(model, checkpoint_dir+"/model")
joblib.dump(opt, checkpoint_dir+"/optimizer")
# save rest
rslt_saving(checkpoint_dir, model, Df, data, masks_a, masks_b, m_kwargs_a, m_kwargs_b, sample_T,
train_model, losses, quiver_scale)
else:
# only save the results
rslt_saving(rslt_dir, model, Df, data, masks_a, masks_b, m_kwargs_a, m_kwargs_b, sample_T,
False, [], quiver_scale)
print("Finish running!")
momentum_vecs = MomentumInteractionTransformation._compute_momentum_vecs(
data[:-1], lags=momentum_lags)
interaction_vecs = MomentumInteractionTransformation._compute_interaction_vecs(
data[:-1])
out = model.log_likelihood(data,
momentum_vecs=momentum_vecs,
interaction_vecs=interaction_vecs)
print(out)
##################### training ############################
num_iters = 10
losses, opt = model.fit(data,
num_iters=num_iters,
lr=0.001,
momentum_vecs=momentum_vecs,
interaction_vecs=interaction_vecs)
##################### sampling ############################
print("start sampling")
sample_z, sample_x = model.sample(30)
#################### inference ###########################
print("inferiring most likely states...")
z = model.most_likely_states(data,
momentum_vecs=momentum_vecs,
interaction_vecs=interaction_vecs)
#print("k step prediction")
#x_predict = k_step_prediction_for_coupled_momentum_model(model, z, data, momentum_vecs=momentum_vecs, features=features)
def main(job_name, cuda_num, downsample_n, filter_traj, gp_version, load_model,
load_model_dir, load_opt_dir, transition, sticky_alpha, sticky_kappa,
acc_factor, k, x_grids, y_grids, n_x, n_y, rs_factor, rs, train_rs,
train_model, pbar_update_interval, video_clips, held_out_proportion,
torch_seed, np_seed, list_of_num_iters, ckpts_not_to_save, list_of_lr,
list_of_k_steps, sample_t, quiver_scale):
if job_name is None:
raise ValueError("Please provide the job name.")
cuda_num = int(cuda_num)
device = torch.device(
"cuda:{}".format(cuda_num) if torch.cuda.is_available() else "cpu")
print("Using device {} \n\n".format(device))
K = k
sample_T = sample_t
rs_factor = np.array([float(x) for x in rs_factor.split(",")])
if rs_factor[0] == 0 and rs_factor[1] == 0:
rs_factor = None
rs = float(rs)
video_clip_start, video_clip_end = [
float(x) for x in video_clips.split(",")
]
list_of_num_iters = [int(x) for x in list_of_num_iters.split(",")]
list_of_lr = [float(x) for x in list_of_lr.split(",")]
list_of_k_steps = [int(x) for x in list_of_k_steps.split(",")]
assert len(list_of_num_iters) == len(
list_of_lr
), "Length of list_of_num_iters must match length of list_of_lr."
for lr in list_of_lr:
if lr > 1:
raise ValueError("Learning rate should not be larger than 1!")
ckpts_not_to_save = [int(x) for x in ckpts_not_to_save.split(',')
] if ckpts_not_to_save else []
repo = git.Repo(
'.', search_parent_directories=True) # SocialBehaviorectories=True)
repo_dir = repo.working_tree_dir # SocialBehavior
torch.manual_seed(torch_seed)
np.random.seed(np_seed)
########################## data ########################
data_dir = repo_dir + '/SocialBehaviorptc/data/trajs_all'
trajs = joblib.load(data_dir)
traj = trajs[int(36000 * video_clip_start):int(36000 * video_clip_end)]
traj = downsample(traj, downsample_n)
if filter_traj:
traj = filter_traj_by_speed(traj, q1=0.99, q2=0.99)
data = torch.tensor(traj, dtype=torch.float64, device=device)
assert 0 <= held_out_proportion <= 0.4, \
"held_out-portion should be between 0 and 0.4 (inclusive), but is {}".format(held_out_proportion)
T = data.shape[0]
breakpoint = int(T * (1 - held_out_proportion))
training_data = data[:breakpoint]
valid_data = data[breakpoint:]
######################### model ####################
# model
D = 4
M = 0
Df = 4
if load_model:
print("Loading the model from ", load_model_dir)
model = joblib.load(load_model_dir)
tran = model.observation.transformation
K = model.K
n_x = len(tran.x_grids) - 1
n_y = len(tran.y_grids) - 1
acc_factor = tran.acc_factor
else:
print("Creating the model...")
bounds = np.array([[ARENA_XMIN, ARENA_XMAX], [ARENA_YMIN, ARENA_YMAX],
[ARENA_XMIN, ARENA_XMAX], [ARENA_YMIN, ARENA_YMAX]])
# grids
if x_grids is None:
x_grid_gap = (ARENA_XMAX - ARENA_XMIN) / n_x
x_grids = np.array(
[ARENA_XMIN + i * x_grid_gap for i in range(n_x + 1)])
else:
x_grids = np.array([float(x) for x in x_grids.split(",")])
n_x = len(x_grids) - 1
if y_grids is None:
y_grid_gap = (ARENA_YMAX - ARENA_YMIN) / n_y
y_grids = np.array(
[ARENA_YMIN + i * y_grid_gap for i in range(n_y + 1)])
else:
y_grids = np.array([float(x) for x in y_grids.split(",")])
n_y = len(y_grids) - 1
if acc_factor is None:
acc_factor = downsample_n * 10
tran = GPGridTransformation(K=K,
D=D,
x_grids=x_grids,
y_grids=y_grids,
Df=Df,
feature_vec_func=f_corner_vec_func,
acc_factor=acc_factor,
rs_factor=rs_factor,
rs=None,
train_rs=train_rs,
device=device,
version=gp_version)
obs = ARTruncatedNormalObservation(K=K,
D=D,
M=M,
lags=1,
bounds=bounds,
transformation=tran,
device=device)
if transition == 'sticky':
transition_kwargs = dict(alpha=sticky_alpha, kappa=sticky_kappa)
else:
transition_kwargs = None
model = HMM(K=K,
D=D,
M=M,
transition=transition,
observation=obs,
transition_kwargs=transition_kwargs,
device=device)
model.observation.mus_init = training_data[0] * torch.ones(
K, D, dtype=torch.float64, device=device)
# save experiment params
exp_params = {
"job_name": job_name,
'downsample_n': downsample_n,
"filter_traj": filter_traj,
"load_model": load_model,
"gp_version": gp_version,
"load_model_dir": load_model_dir,
"load_opt_dir": load_opt_dir,
"transition": transition,
"sticky_alpha": sticky_alpha,
"sticky_kappa": sticky_kappa,
"acc_factor": acc_factor,
"K": K,
"x_grids": x_grids,
"y_grids": y_grids,
"n_x": n_x,
"n_y": n_y,
"rs_factor": get_np(tran.rs_factor),
"rs": rs,
"train_rs": train_rs,
"train_model": train_model,
"pbar_update_interval": pbar_update_interval,
"video_clip_start": video_clip_start,
"video_clip_end": video_clip_end,
"held_out_proportion": held_out_proportion,
"torch_seed": torch_seed,
"np_seed": np_seed,
"list_of_num_iters": list_of_num_iters,
"list_of_lr": list_of_lr,
"list_of_k_steps": list_of_k_steps,
"sample_T": sample_T,
"quiver_scale": quiver_scale
}
print("Experiment params:")
print(exp_params)
rslt_dir = addDateTime("rslts/gpgrid/" + job_name)
rslt_dir = os.path.join(repo_dir, rslt_dir)
if not os.path.exists(rslt_dir):
os.makedirs(rslt_dir)
print("Making result directory...")
print("Saving to rlst_dir: ", rslt_dir)
with open(rslt_dir + "/exp_params.json", "w") as f:
json.dump(exp_params, f, indent=4, cls=NumpyEncoder)
# compute memory
print("Computing memory...")
def get_memory_kwargs(data, train_rs):
feature_vecs_a = f_corner_vec_func(data[:-1, 0:2])
feature_vecs_b = f_corner_vec_func(data[:-1, 2:4])
gpt_idx_a, grid_idx_a = tran.get_gpt_idx_and_grid_idx_for_batch(
data[:-1, 0:2])
gpt_idx_b, grid_idx_b = tran.get_gpt_idx_and_grid_idx_for_batch(
data[:-1, 2:4])
if train_rs:
nearby_gpts_a = tran.gridpoints[gpt_idx_a]
dist_sq_a = (data[:-1, None, 0:2] - nearby_gpts_a)**2
nearby_gpts_b = tran.gridpoints[gpt_idx_b]
dist_sq_b = (data[:-1, None, 2:4] - nearby_gpts_b)**2
return dict(feature_vecs_a=feature_vecs_a,
feature_vecs_b=feature_vecs_b,
gpt_idx_a=gpt_idx_a,
gpt_idx_b=gpt_idx_b,
grid_idx_a=grid_idx_a,
grid_idx_b=grid_idx_b,
dist_sq_a=dist_sq_a,
dist_sq_b=dist_sq_b)
else:
coeff_a = tran.get_gp_coefficients(data[:-1, 0:2], 0, gpt_idx_a,
grid_idx_a)
coeff_b = tran.get_gp_coefficients(data[:-1, 2:4], 0, gpt_idx_b,
grid_idx_b)
return dict(feature_vecs_a=feature_vecs_a,
feature_vecs_b=feature_vecs_b,
gpt_idx_a=gpt_idx_a,
gpt_idx_b=gpt_idx_b,
grid_idx_a=grid_idx_a,
grid_idx_b=grid_idx_b,
coeff_a=coeff_a,
coeff_b=coeff_b)
memory_kwargs = get_memory_kwargs(training_data, train_rs)
valid_data_memory_kwargs = get_memory_kwargs(valid_data, train_rs)
log_prob = model.log_likelihood(training_data, **memory_kwargs)
##################### training ############################
if train_model:
print("start training")
list_of_losses = []
if load_opt_dir != "":
opt = joblib.load(load_opt_dir)
else:
opt = None
for i, (num_iters, lr) in enumerate(zip(list_of_num_iters,
list_of_lr)):
training_losses, opt, valid_losses = model.fit(
training_data,
optimizer=opt,
method='adam',
num_iters=num_iters,
lr=lr,
pbar_update_interval=pbar_update_interval,
valid_data=valid_data,
valid_data_memory_kwargs=valid_data_memory_kwargs,
**memory_kwargs)
list_of_losses.append(training_losses)
checkpoint_dir = rslt_dir + "/checkpoint_{}".format(i)
if not os.path.exists(checkpoint_dir):
os.makedirs(checkpoint_dir)
print("Creating checkpoint_{} directory...".format(i))
# save model and opt
joblib.dump(model, checkpoint_dir + "/model")
joblib.dump(opt, checkpoint_dir + "/optimizer")
# save losses
losses = dict(training_loss=training_losses,
valid_loss=valid_losses)
joblib.dump(losses, checkpoint_dir + "/losses")
plt.figure()
plt.plot(training_losses)
plt.title("training loss")
plt.savefig(checkpoint_dir + "/training_losses.jpg")
plt.close()
plt.figure()
plt.plot(valid_losses)
plt.title("validation loss")
plt.savefig(checkpoint_dir + "/valid_losses.jpg")
plt.close()
# save rest
if i in ckpts_not_to_save:
print("ckpt {}: skip!\n".format(i))
continue
with torch.no_grad():
rslt_saving(rslt_dir=checkpoint_dir,
model=model,
data=training_data,
memory_kwargs=memory_kwargs,
list_of_k_steps=list_of_k_steps,
sample_T=sample_T,
quiver_scale=quiver_scale,
valid_data=valid_data,
valid_data_memory_kwargs=valid_data_memory_kwargs,
device=device)
else:
# only save the results
rslt_saving(rslt_dir=rslt_dir,
model=model,
data=training_data,
memory_kwargs=memory_kwargs,
list_of_k_steps=list_of_k_steps,
sample_T=sample_T,
quiver_scale=quiver_scale,
valid_data=valid_data,
valid_data_memory_kwargs=valid_data_memory_kwargs,
device=device)
print("Finish running!")
import matplotlib.pyplot as plt torch.manual_seed(0) np.random.seed(0) # test fitting K = 3 D = 2 lags= 5 trans1 = LinearTransformation(K=K, D=D, lags=lags) obs1 = ARGaussianObservation(K=K, D=D, transformation=trans1) model1 = HMM(K=K,D=D, observation=obs1) T = 100 sample_z, sample_x = model1.sample(T) model2 = HMM(K=K, D=D, observation='gaussian', observation_kwargs=dict(lags=lags)) lls, opt = model2.fit(sample_x, num_iters=2000, lr=0.001) z_infer = model2.most_likely_states(sample_x) x_predict = k_step_prediction(model2, z_infer, sample_x) plt.figure() plt.plot(x_predict[:,0], label='prediction') plt.plot(sample_x[:,0], label='truth') plt.show()
def test_model():
torch.manual_seed(0)
np.random.seed(0)
T = 100
D = 4
# data = np.array([[1.0, 1.0, 1.0, 6.0], [3.0, 6.0, 8.0, 6.0],
# [4.0, 7.0, 8.0, 5.0], [6.0, 7.0, 5.0, 6.0], [8.0, 2.0, 6.0, 1.0]])
data = np.random.randn(T, D)
data = torch.tensor(data, dtype=torch.float64)
xmax = max(np.max(data[:, 0].numpy()), np.max(data[:, 2].numpy()))
xmin = min(np.min(data[:, 0].numpy()), np.min(data[:, 2].numpy()))
ymax = max(np.max(data[:, 1].numpy()), np.max(data[:, 3].numpy()))
ymin = min(np.min(data[:, 1].numpy()), np.min(data[:, 3].numpy()))
bounds = np.array([[xmin - 1, xmax + 1], [ymin - 1, ymax + 1],
[xmin - 1, xmax + 1], [ymin - 1, ymax + 1]])
def toy_feature_vec_func(s):
"""
:param s: self, (T, 2)
:param o: other, (T, 2)
:return: features, (T, Df, 2)
"""
corners = torch.tensor([[0, 0], [0, 8], [10, 0], [10, 8]],
dtype=torch.float64)
return feature_direction_vec(s, corners)
K = 3
Df = 4
lags = 1
tran = UniLSTMTransformation(K=K,
D=D,
Df=Df,
feature_vec_func=toy_feature_vec_func,
lags=lags,
dh=10)
# observation
obs = ARTruncatedNormalObservation(K=K,
D=D,
lags=lags,
bounds=bounds,
transformation=tran)
# model
model = HMM(K=K, D=D, observation=obs)
print("calculating log likelihood")
feature_vecs_a = toy_feature_vec_func(data[:-1, 0:2])
feature_vecs_b = toy_feature_vec_func(data[:-1, 2:4])
feature_vecs = (feature_vecs_a, feature_vecs_b)
packed_data = get_packed_data((data[:-1]), lags=lags)
model.log_likelihood(data,
feature_vecs=feature_vecs,
packed_data=packed_data)
# fit
losses, _ = model.fit(data,
optimizer=None,
method="adam",
num_iters=50,
feature_vecs=feature_vecs,
packed_data=packed_data)
plt.figure()
plt.plot(losses)
plt.show()
# most-likely-z
print("Most likely z...")
z = model.most_likely_states(data,
feature_vecs=feature_vecs,
packed_data=packed_data)
# prediction
if data.shape[0] <= 1000:
data_to_predict = data
else:
data_to_predict = data[-1000:]
print("0 step prediction")
if data.shape[0] <= 1000:
data_to_predict = data
else:
data_to_predict = data[-1000:]
x_predict = k_step_prediction_for_lstm_model(model,
z,
data_to_predict,
feature_vecs=feature_vecs)
x_predict_err = np.mean(np.abs(x_predict - data_to_predict.numpy()),
axis=0)
print("10 step prediction")
x_predict_2 = k_step_prediction(model, z, data_to_predict, k=10)
x_predict_2_err = np.mean(np.abs(x_predict_2 -
data_to_predict[10:].numpy()),
axis=0)
# samples
print("sampling...")
sample_T = 5
sample_z, sample_x = model.sample(sample_T)
def test_model():
torch.manual_seed(0)
np.random.seed(0)
T = 5
x_grids = np.array([0.0, 5.0, 10.0])
y_grids = np.array([0.0, 4.0, 8.0])
data = np.array([[1.0, 1.0, 1.0, 6.0], [3.0, 6.0, 8.0, 6.0],
[4.0, 7.0, 8.0, 5.0], [6.0, 7.0, 5.0, 6.0],
[8.0, 2.0, 6.0, 1.0]])
data = torch.tensor(data, dtype=torch.float64)
def toy_feature_vec_func(s):
"""
:param s: self, (T, 2)
:param o: other, (T, 2)
:return: features, (T, Df, 2)
"""
corners = torch.tensor([[0, 0], [0, 8], [10, 0], [10, 8]],
dtype=torch.float64)
return feature_direction_vec(s, corners)
K = 3
D = 4
M = 0
Df = 4
bounds = np.array([[0.0, 10.0], [0.0, 8.0], [0.0, 10.0], [0.0, 8.0]])
tran = LinearGridTransformation(K=K,
D=D,
x_grids=x_grids,
y_grids=y_grids,
Df=Df,
feature_vec_func=toy_feature_vec_func)
obs = ARTruncatedNormalObservation(K=K,
D=D,
M=0,
lags=1,
bounds=bounds,
transformation=tran)
model = HMM(K=K, D=D, M=M, transition="stationary", observation=obs)
model.observation.mus_init = data[0] * torch.ones(
K, D, dtype=torch.float64)
# calculate memory
gridpoints_idx_a = tran.get_gridpoints_idx_for_batch(data[:-1, 0:2])
gridpoints_idx_b = tran.get_gridpoints_idx_for_batch(data[:-1, 2:4])
gridpoints_a = tran.get_gridpoints_for_batch(gridpoints_idx_a)
gridpoints_b = tran.get_gridpoints_for_batch(gridpoints_idx_b)
feature_vecs_a = toy_feature_vec_func(data[:-1, 0:2])
feature_vecs_b = toy_feature_vec_func(data[:-1, 2:4])
gridpoints_idx = (gridpoints_idx_a, gridpoints_idx_b)
gridpoints = (gridpoints_a, gridpoints_b)
feature_vecs = (feature_vecs_a, feature_vecs_b)
# fit
losses, opt = model.fit(data,
optimizer=None,
method='adam',
num_iters=100,
lr=0.01,
pbar_update_interval=10,
gridpoints=gridpoints,
gridpoints_idx=gridpoints_idx,
feature_vecs=feature_vecs)
plt.figure()
plt.plot(losses)
plt.show()
# most-likely-z
print("Most likely z...")
z = model.most_likely_states(data,
gridpoints_idx=gridpoints_idx,
feature_vecs=feature_vecs)
# prediction
print("0 step prediction")
if data.shape[0] <= 1000:
data_to_predict = data
else:
data_to_predict = data[-1000:]
x_predict = k_step_prediction_for_lineargrid_model(
model,
z,
data_to_predict,
gridpoints_idx=gridpoints_idx,
feature_vecs=feature_vecs)
x_predict_err = np.mean(np.abs(x_predict - data_to_predict.numpy()),
axis=0)
print("2 step prediction")
x_predict_2 = k_step_prediction(model, z, data_to_predict, k=2)
x_predict_2_err = np.mean(np.abs(x_predict_2 -
data_to_predict[2:].numpy()),
axis=0)
# samples
sample_T = 5
sample_z, sample_x = model.sample(sample_T)
def test_model():
torch.manual_seed(0)
np.random.seed(0)
T = 5
x_grids = np.array([0.0, 10.0])
y_grids = np.array([0.0, 8.0])
bounds = np.array([[0.0, 10.0], [0.0, 8.0], [0.0, 10.0], [0.0, 8.0]])
data = np.array([[1.0, 1.0, 1.0, 6.0], [3.0, 6.0, 8.0, 6.0],
[4.0, 7.0, 8.0, 5.0], [6.0, 7.0, 5.0, 6.0],
[8.0, 2.0, 6.0, 1.0]])
data = torch.tensor(data, dtype=torch.float64)
K = 3
D = 4
M = 0
obs = GPObservation(K=K,
D=D,
x_grids=x_grids,
y_grids=y_grids,
bounds=bounds,
train_rs=True)
correct_kerneldist_gg = torch.tensor(
[[0., 0., 64., 64., 100., 100., 164., 164.],
[0., 0., 64., 64., 100., 100., 164., 164.],
[64., 64., 0., 0., 164., 164., 100., 100.],
[64., 64., 0., 0., 164., 164., 100., 100.],
[100., 100., 164., 164., 0., 0., 64., 64.],
[100., 100., 164., 164., 0., 0., 64., 64.],
[164., 164., 100., 100., 64., 64., 0., 0.],
[164., 164., 100., 100., 64., 64., 0., 0.]],
dtype=torch.float64)
assert torch.all(torch.eq(correct_kerneldist_gg,
obs.kernel_distsq_gg)), obs.kernel_distsq_gg
log_prob_nocache = obs.log_prob(data)
print("log_prob_nocache = {}".format(log_prob_nocache))
kernel_distsq_xg_a = kernel_distsq_doubled(data[:-1, 0:2],
obs.inducing_points)
kernel_distsq_xg_b = kernel_distsq_doubled(data[:-1, 2:4],
obs.inducing_points)
correct_kernel_distsq_xg_a = torch.tensor(
[[2., 2., 50., 50., 82., 82., 130., 130.],
[2., 2., 50., 50., 82., 82., 130., 130.],
[45., 45., 13., 13., 85., 85., 53., 53.],
[45., 45., 13., 13., 85., 85., 53., 53.],
[65., 65., 17., 17., 85., 85., 37., 37.],
[65., 65., 17., 17., 85., 85., 37., 37.],
[85., 85., 37., 37., 65., 65., 17., 17.],
[85., 85., 37., 37., 65., 65., 17., 17.]],
dtype=torch.float64)
assert torch.all(torch.eq(correct_kernel_distsq_xg_a,
kernel_distsq_xg_a)), kernel_distsq_xg_a
memory_kwargs = dict(kernel_distsq_xg_a=kernel_distsq_xg_a,
kernel_distsq_xg_b=kernel_distsq_xg_b)
log_prob = obs.log_prob(data, **memory_kwargs)
print("log_prob = {}".format(log_prob))
assert torch.all(torch.eq(log_prob_nocache, log_prob))
Sigma_a, A_a = obs.get_gp_cache(data[:-1, 0:2], 0, **memory_kwargs)
Sigma_b, A_b = obs.get_gp_cache(data[:-1, 2:4], 1, **memory_kwargs)
memory_kwargs_2 = dict(Sigma_a=Sigma_a, A_a=A_a, Sigma_b=Sigma_b, A_b=A_b)
print("calculating log prob 2...")
log_prob2 = obs.log_prob(data, **memory_kwargs_2)
assert torch.all(torch.eq(log_prob, log_prob2))
model = HMM(K=K, D=D, M=M, transition="stationary", observation=obs)
model.observation.mus_init = data[0] * torch.ones(
K, D, dtype=torch.float64)
# fit
losses, opt = model.fit(data,
optimizer=None,
method='adam',
num_iters=100,
lr=0.01,
pbar_update_interval=10,
**memory_kwargs)
plt.figure()
plt.plot(losses)
plt.show()
# most-likely-z
print("Most likely z...")
z = model.most_likely_states(data, **memory_kwargs)
# prediction
print("0 step prediction")
if data.shape[0] <= 1000:
data_to_predict = data
else:
data_to_predict = data[-1000:]
x_predict = k_step_prediction_for_gpmodel(model, z, data_to_predict,
**memory_kwargs)
x_predict_err = np.mean(np.abs(x_predict - data_to_predict.numpy()),
axis=0)
print("2 step prediction")
x_predict_2 = k_step_prediction(model, z, data_to_predict, k=2)
x_predict_2_err = np.mean(np.abs(x_predict_2 -
data_to_predict[2:].numpy()),
axis=0)
# samples
sample_T = 5
sample_z, sample_x = model.sample(sample_T)
# observation
obs = ARTruncatedNormalObservation(K=K, D=D, M=M, lags=momentum_lags, bounds=bounds, transformation=tran)
# model
model = HMM(K=K, D=D, M=M, observation=obs)
log_prob = model.log_likelihood(data, masks=(masks_a, masks_b),
memory_kwargs_a=m_kwargs_a, memory_kwargs_b=m_kwargs_b)
log_prob_2 = model.log_likelihood(data)
assert torch.eq(log_prob, log_prob_2)
print(log_prob)
# training
print("start training...")
num_iters = 10
losses, opt = model.fit(data, num_iters=num_iters, lr=0.001, masks=(masks_a, masks_b),
memory_kwargs_a=m_kwargs_a, memory_kwargs_b=m_kwargs_b)
# sampling
print("start sampling")
sample_z, sample_x = model.sample(T)
# inference
print("inferiring most likely states...")
z = model.most_likely_states(data, masks=(masks_a, masks_b),
memory_kwargs_a=m_kwargs_a, memory_kwargs_b=m_kwargs_b)
print("0 step prediction")
x_predict = k_step_prediction_for_grid_model(model, z, data, memory_kwargs_a=m_kwargs_a, memory_kwargs_b=m_kwargs_b)
print("k step prediction")
x_predict_10 = k_step_prediction(model, z, data, 10)
momentum_vecs = MomentumFeatureTransformation._compute_momentum_vecs(
data[:-1], lags=momentum_lags)
features = MomentumFeatureTransformation._compute_features(
feature_funcs=feature_func_single, inputs=data[:-1])
out = model.log_likelihood(data,
momentum_vecs=momentum_vecs,
features=features)
print(out)
##################### training ############################
num_iters = 10
losses, opt = model.fit(data,
num_iters=num_iters,
lr=0.001,
momentum_vecs=momentum_vecs,
features=features)
##################### sampling ############################
print("start sampling")
sample_z, sample_x = model.sample(30)
#################### inference ###########################
print("inferiring most likely states...")
z = model.most_likely_states(data,
momentum_vecs=momentum_vecs,
features=features)
print("k step prediction")
x_predict = k_step_prediction_for_momentum_feature_model(
def main(job_name, cuda_num, data_type, downsample_n, filter_traj, lg_version, use_log_prior, no_boundary_prior,
add_log_diagonal_prior, log_prior_sigma_sq,
load_model, load_model_dir, load_opt_dir, animal, reset_prior_info,
transition, sticky_alpha, sticky_kappa, acc_factor, k, x_grids, y_grids, n_x, n_y,
train_model, pbar_update_interval, prop_start_end, video_clips, held_out_proportion, torch_seed, np_seed,
list_of_num_iters, ckpts_not_to_save, list_of_lr, list_of_k_steps, sample_t, quiver_scale):
if job_name is None:
raise ValueError("Please provide the job name.")
assert animal in ['both', 'virgin', 'mother'], animal
cuda_num = int(cuda_num)
device = torch.device("cuda:{}".format(cuda_num) if torch.cuda.is_available() else "cpu")
print("Using device {} \n\n".format(device))
K = k
sample_T = sample_t
video_clip_start, video_clip_end = [float(x) for x in video_clips.split(",")]
start, end = [float(x) for x in prop_start_end.split(",")]
log_prior_sigma_sq = float(log_prior_sigma_sq)
list_of_num_iters = [int(x) for x in list_of_num_iters.split(",")]
list_of_lr = [float(x) for x in list_of_lr.split(",")]
# TODO: test if that works
list_of_k_steps = [int(x) for x in list_of_k_steps.split(",")] if list_of_k_steps else []
assert len(list_of_num_iters) == len(list_of_lr), "Length of list_of_num_iters must match length of list_of_lr."
for lr in list_of_lr:
if lr > 1:
raise ValueError("Learning rate should not be larger than 1!")
ckpts_not_to_save = [int(x) for x in ckpts_not_to_save.split(',')] if ckpts_not_to_save else []
repo = git.Repo('.', search_parent_directories=True) # SocialBehaviorectories=True)
repo_dir = repo.working_tree_dir # SocialBehavior
torch.manual_seed(torch_seed)
np.random.seed(np_seed)
########################## data ########################
if data_type == 'full':
data_dir = repo_dir + '/SocialBehaviorptc/data/trajs_all'
traj = joblib.load(data_dir)
traj = traj[int(36000 * video_clip_start):int(36000 * video_clip_end)]
elif data_type == 'selected_010_virgin':
assert animal == 'virgin', "animal much be 'virgin', but got {}.".format(animal)
data_dir = repo_dir + '/SocialBehaviorptc/data/traj_010_virgin_selected'
traj = joblib.load(data_dir)
T = len(traj)
traj = traj[int(T*start): int(T*end)]
else:
raise ValueError("unsupported data type: {}".format(data_type))
if animal == 'virgin':
traj = traj[:,0:2]
elif animal == 'mother':
traj = traj[:,2:4]
traj = downsample(traj, downsample_n)
if filter_traj:
traj = filter_traj_by_speed(traj, q1=0.99, q2=0.99)
data = torch.tensor(traj, dtype=torch.float64, device=device)
assert 0 <= held_out_proportion <= 0.4, \
"held_out-portion should be between 0 and 0.4 (inclusive), but is {}".format(held_out_proportion)
T = data.shape[0]
breakpoint = int(T*(1-held_out_proportion))
training_data = data[:breakpoint]
valid_data = data[breakpoint:]
sample_T = training_data.shape[0]
######################### model ####################
# model
D = data.shape[1]
assert D == 2 or 4, D
M = 0
Df = 4
if load_model:
print("Loading the model from ", load_model_dir)
if reset_prior_info:
pretrained_model = joblib.load(load_model_dir)
pretrained_transition = pretrained_model.transition
pretrained_observation = pretrained_model.observation
pretrained_tran = pretrained_model.observation.transformation
# set prior info
pretrained_tran.use_log_prior = use_log_prior
pretrained_tran.no_boundary_prior = no_boundary_prior
pretrained_tran.add_log_diagonal_prior = add_log_diagonal_prior
pretrained_tran.log_prior_sigma_sq = torch.tensor(log_prior_sigma_sq, dtype=torch.float64, device=device)
acc_factor = pretrained_tran.acc_factor
K = pretrained_model.K
obs = ARTruncatedNormalObservation(K=K, D=D, M=0, obs=pretrained_observation, device=device)
tran = obs.transformation
if transition == 'sticky':
transition_kwargs = dict(alpha=sticky_alpha, kappa=sticky_kappa)
else:
transition_kwargs = None
model = HMM(K=K, D=D, M=M, pi0=get_np(pretrained_model.pi0), Pi=get_np(pretrained_transition.Pi),
transition=transition, observation=obs, transition_kwargs=transition_kwargs,
device=device)
model.observation.mus_init = training_data[0] * torch.ones(K, D, dtype=torch.float64, device=device)
else:
model = joblib.load(load_model_dir)
tran = model.observation.transformation
K = model.K
n_x = len(tran.x_grids) - 1
n_y = len(tran.y_grids) - 1
if isinstance(model, LinearGridTransformation):
acc_factor = tran.acc_factor
use_log_prior = tran.use_log_prior
no_boundary_prior = tran.no_boundary_prior
add_log_diagonal_prior = tran.add_log_diagonal_prior
log_prior_sigma_sq = get_np(tran.log_prior_sigma_sq)
else:
if D == 4:
bounds = np.array([[ARENA_XMIN, ARENA_XMAX], [ARENA_YMIN, ARENA_YMAX],
[ARENA_XMIN, ARENA_XMAX], [ARENA_YMIN, ARENA_YMAX]])
else:
bounds = np.array([[ARENA_XMIN, ARENA_XMAX], [ARENA_YMIN, ARENA_YMAX]])
# grids
if x_grids is None:
x_grid_gap = (ARENA_XMAX - ARENA_XMIN) / n_x
x_grids = np.array([ARENA_XMIN + i * x_grid_gap for i in range(n_x + 1)])
else:
x_grids = np.array([float(x) for x in x_grids.split(",")])
n_x = len(x_grids) - 1
if y_grids is None:
y_grid_gap = (ARENA_YMAX - ARENA_YMIN) / n_y
y_grids = np.array([ARENA_YMIN + i * y_grid_gap for i in range(n_y + 1)])
else:
y_grids = np.array([float(x) for x in y_grids.split(",")])
n_y = len(y_grids) - 1
if acc_factor is None:
acc_factor = downsample_n * 10
print("Creating the model...")
if animal == 'both':
tran = LinearGridTransformation(K=K, D=D, x_grids=x_grids, y_grids=y_grids,
Df=Df, feature_vec_func=f_corner_vec_func, acc_factor=acc_factor,
use_log_prior=use_log_prior, no_boundary_prior=no_boundary_prior,
add_log_diagonal_prior=add_log_diagonal_prior,
log_prior_sigma_sq=log_prior_sigma_sq, device=device, version=lg_version)
else:
tran = SingleLinearGridTransformation(K=K, D=D, x_grids=x_grids, y_grids=y_grids, device=device)
obs = ARTruncatedNormalObservation(K=K, D=D, M=M, lags=1, bounds=bounds, transformation=tran, device=device)
if transition == 'sticky':
transition_kwargs = dict(alpha=sticky_alpha, kappa=sticky_kappa)
elif transition == 'grid':
transition_kwargs = dict(x_grids=x_grids, y_grids=y_grids)
else:
transition_kwargs = None
model = HMM(K=K, D=D, M=M, transition=transition, observation=obs, transition_kwargs=transition_kwargs,
device=device)
model.observation.mus_init = training_data[0] * torch.ones(K, D, dtype=torch.float64, device=device)
# save experiment params
exp_params = {"job_name": job_name,
'downsample_n': downsample_n,
'data_type': data_type,
"filter_traj": filter_traj,
"lg_version": lg_version,
"use_log_prior": use_log_prior,
"add_log_diagonal_prior": add_log_diagonal_prior,
"no_boundary_prior": no_boundary_prior,
"log_prior_sigma_sq": log_prior_sigma_sq,
"load_model": load_model,
"load_model_dir": load_model_dir,
"animal": animal,
"reset_prior_info": reset_prior_info,
"load_opt_dir": load_opt_dir,
"transition": transition,
"sticky_alpha": sticky_alpha,
"sticky_kappa": sticky_kappa,
"acc_factor": acc_factor,
"K": K,
"x_grids": x_grids,
"y_grids": y_grids,
"n_x": n_x,
"n_y": n_y,
"train_model": train_model,
"pbar_update_interval": pbar_update_interval,
"video_clip_start": video_clip_start,
"video_clip_end": video_clip_end,
"start_percentage": start,
"end_percentage": end,
"held_out_proportion": held_out_proportion,
"torch_seed": torch_seed,
"np_seed": np_seed,
"list_of_num_iters": list_of_num_iters,
"ckpts_not_to_save": ckpts_not_to_save,
"list_of_lr": list_of_lr,
"list_of_k_steps": list_of_k_steps,
"sample_T": sample_T,
"quiver_scale": quiver_scale}
print("Experiment params:")
print(exp_params)
rslt_dir = addDateTime("rslts/lineargrid/{}/{}".format(animal, job_name))
rslt_dir = os.path.join(repo_dir, rslt_dir)
if not os.path.exists(rslt_dir):
os.makedirs(rslt_dir)
print("Making result directory...")
print("Saving to rlst_dir: ", rslt_dir)
with open(rslt_dir+"/exp_params.json", "w") as f:
json.dump(exp_params, f, indent=4, cls=NumpyEncoder)
# compute memory
if transition == "grid":
print("Computing transition memory...")
joint_grid_idx = model.transition.get_grid_idx(training_data[:-1])
transition_memory_kwargs = dict(joint_grid_idx=joint_grid_idx)
valid_joint_grid_idx = model.transition.get_grid_idx(valid_data[:-1])
valid_data_transition_memory_kwargs = dict(joint_grid_idx=valid_joint_grid_idx)
else:
transition_memory_kwargs = None
valid_data_transition_memory_kwargs = None
print("Computing observation memory...")
if animal == 'both':
gridpoints_idx_a = tran.get_gridpoints_idx_for_batch(training_data[:-1, 0:2]) # (T-1, n_gps, 4)
gridpoints_idx_b = tran.get_gridpoints_idx_for_batch(training_data[:-1, 2:4]) # (T-1, n_gps, 4)
gridpoints_a = tran.get_gridpoints_for_batch(gridpoints_idx_a) # (T-1, d, 2)
gridpoints_b = tran.get_gridpoints_for_batch(gridpoints_idx_b) # (T-1, d, 2)
feature_vecs_a = f_corner_vec_func(training_data[:-1, 0:2]) # (T, Df, 2)
feature_vecs_b = f_corner_vec_func(training_data[:-1, 2:4]) # (T, Df, 2)
gridpoints_idx = (gridpoints_idx_a, gridpoints_idx_b)
gridpoints = (gridpoints_a, gridpoints_b)
feature_vecs = (feature_vecs_a, feature_vecs_b)
memory_kwargs = dict(gridpoints_idx=gridpoints_idx, gridpoints=gridpoints, feature_vecs=feature_vecs)
if len(valid_data) > 0:
gridpoints_idx_a_v = tran.get_gridpoints_idx_for_batch(valid_data[:-1, 0:2]) # (T-1, n_gps, 4)
gridpoints_idx_b_v = tran.get_gridpoints_idx_for_batch(valid_data[:-1, 2:4]) # (T-1, n_gps, 4)
gridpoints_a_v = tran.get_gridpoints_for_batch(gridpoints_idx_a_v) # (T-1, d, 2)
gridpoints_b_v = tran.get_gridpoints_for_batch(gridpoints_idx_b_v) # (T-1, d, 2)
feature_vecs_a_v = f_corner_vec_func(valid_data[:-1, 0:2]) # (T, Df, 2)
feature_vecs_b_v = f_corner_vec_func(valid_data[:-1, 2:4]) # (T, Df, 2)
gridpoints_idx_v = (gridpoints_idx_a_v, gridpoints_idx_b_v)
gridpoints_v = (gridpoints_a_v, gridpoints_b_v)
feature_vecs_v = (feature_vecs_a_v, feature_vecs_b_v)
valid_data_memory_kwargs = dict(gridpoints_idx=gridpoints_idx_v, gridpoints=gridpoints_v,
feature_vecs=feature_vecs_v)
else:
valid_data_memory_kwargs = {}
else:
def get_memory_kwargs(data):
if data is None or data.shape[0] == 0:
return {}
gridpoints_idx = tran.get_gridpoints_idx_for_batch(data)
gridpoints = tran.gridpoints[gridpoints_idx]
coeffs = tran.get_lp_coefficients(data, gridpoints[:, 0], gridpoints[:, 3], device=device)
return dict(gridpoints_idx=gridpoints_idx, coeffs=coeffs)
memory_kwargs = get_memory_kwargs(training_data[:-1])
valid_data_memory_kwargs = get_memory_kwargs(valid_data[:-1])
log_prob = model.log_likelihood(training_data, transition_memory_kwargs=transition_memory_kwargs, **memory_kwargs)
print("log_prob = {}".format(log_prob))
##################### training ############################
if train_model:
print("start training")
list_of_losses = []
if load_opt_dir != "":
opt = joblib.load(load_opt_dir)
else:
opt = None
for i, (num_iters, lr) in enumerate(zip(list_of_num_iters, list_of_lr)):
training_losses, opt, valid_losses, _ = \
model.fit(training_data, optimizer=opt, method='adam', num_iters=num_iters, lr=lr,
pbar_update_interval=pbar_update_interval, valid_data=valid_data,
transition_memory_kwargs=transition_memory_kwargs,
valid_data_transition_memory_kwargs=valid_data_transition_memory_kwargs,
valid_data_memory_kwargs=valid_data_memory_kwargs, **memory_kwargs)
list_of_losses.append(training_losses)
checkpoint_dir = rslt_dir + "/checkpoint_{}".format(i)
if not os.path.exists(checkpoint_dir):
os.makedirs(checkpoint_dir)
print("Creating checkpoint_{} directory...".format(i))
# save model and opt
joblib.dump(model, checkpoint_dir+"/model")
joblib.dump(opt, checkpoint_dir+"/optimizer")
# save losses
losses = dict(training_loss=training_losses, valid_loss=valid_losses)
joblib.dump(losses, checkpoint_dir+"/losses")
plt.figure()
plt.plot(training_losses)
plt.title("training loss")
plt.savefig(checkpoint_dir + "/training_losses.jpg")
plt.close()
plt.figure()
plt.plot(valid_losses)
plt.title("validation loss")
plt.savefig(checkpoint_dir + "/valid_losses.jpg")
plt.close()
# save rest
if i in ckpts_not_to_save:
print("ckpt {}: skip!\n".format(i))
continue
with torch.no_grad():
rslt_saving(rslt_dir=checkpoint_dir, model=model, data=training_data, animal=animal,
memory_kwargs=memory_kwargs,
list_of_k_steps=list_of_k_steps, sample_T=sample_T,
quiver_scale=quiver_scale, valid_data=valid_data,
valid_data_memory_kwargs=valid_data_memory_kwargs, device=device)
else:
# only save the results
rslt_saving(rslt_dir=rslt_dir, model=model, data=training_data, animal=animal,
memory_kwargs=memory_kwargs,
list_of_k_steps=list_of_k_steps, sample_T=sample_T,
quiver_scale=quiver_scale, valid_data=valid_data,
valid_data_memory_kwargs=valid_data_memory_kwargs, device=device)
print("Finish running!")
D = 4
M = 0
obs = TruncatedNormalObservation(K=K, D=D, M=M, bounds=bounds)
# model
model = HMM(K=K, D=D, M=M, observation=obs)
# log like
log_prob = model.log_probability(data)
print("log probability = ", log_prob)
# training
print("start training...")
num_iters = 10
losses, opt = model.fit(data, num_iters=num_iters, lr=0.001)
# sampling
samplt_T = T
print("start sampling")
sample_z, sample_x = model.sample(samplt_T)
print("start sampling based on with_noise")
sample_z2, sample_x2 = model.sample(T, with_noise=True)
# inference
print("inferiring most likely states...")
z = model.most_likely_states(data)
print("0 step prediction")
x_predict = k_step_prediction(model, z, data)
def main(job_name, downsample_n, mouse, filter_traj, load_model,
load_model_dir, load_opt_dir, transition, sticky_alpha, sticky_kappa,
k, x_grids, y_grids, n_x, n_y, not_train_mu, initialize_mu,
train_model, pbar_update_interval, video_clips, torch_seed, np_seed,
list_of_num_iters, list_of_lr, sample_t, quiver_scale):
if job_name is None:
raise ValueError("Please provide the job name.")
K = k
sample_T = sample_t
video_clip_start, video_clip_end = [int(x) for x in video_clips.split(",")]
list_of_num_iters = [int(x) for x in list_of_num_iters.split(",")]
list_of_lr = [float(x) for x in list_of_lr.split(",")]
assert len(list_of_num_iters) == len(
list_of_lr
), "Length of list_of_num_iters must match length of list-of_lr."
for lr in list_of_lr:
if lr > 1:
raise ValueError("Learning rate should not be larger than 1!")
repo = git.Repo(
'.', search_parent_directories=True) # SocialBehaviorectories=True)
repo_dir = repo.working_tree_dir # SocialBehavior
torch.manual_seed(torch_seed)
np.random.seed(np_seed)
# specify grids for sanity checks and plotting
if x_grids is None:
x_grid_gap = (ARENA_XMAX - ARENA_XMIN) / n_x
x_grids = np.array(
[ARENA_XMIN + i * x_grid_gap for i in range(n_x + 1)])
else:
x_grids = np.array([float(x) for x in x_grids.split(",")])
n_x = len(x_grids) - 1
if y_grids is None:
y_grid_gap = (ARENA_YMAX - ARENA_YMIN) / n_y
y_grids = np.array(
[ARENA_YMIN + i * y_grid_gap for i in range(n_y + 1)])
else:
y_grids = np.array([float(x) for x in y_grids.split(",")])
n_y = len(y_grids) - 1
########################## data ########################
data_dir = repo_dir + '/SocialBehaviorptc/data/trajs_all'
trajs = joblib.load(data_dir)
traj = trajs[36000 * video_clip_start:36000 * video_clip_end]
traj = downsample(traj, downsample_n)
if filter_traj:
traj = filter_traj_by_speed(traj, q1=0.99, q2=0.99)
if mouse == "both":
data = torch.tensor(traj, dtype=torch.float64)
D = 4
elif mouse == "virgin":
data = torch.tensor(traj[:, 0:2], dtype=torch.float64)
D = 2
elif mouse == "mother":
data = torch.tensor(traj[:, 2:4], dtype=torch.float64)
D = 2
else:
raise ValueError(
"mouse must be chosen from 'both', 'virgin', 'mother'.")
######################### model ####################
# model
M = 0
if load_model:
print("Loading the model from ", load_model_dir)
model = joblib.load(load_model_dir)
K = model.K
cluster_locations = get_np(model.observation.mus)
else:
print("Creating the model...")
if D == 4:
bounds = np.array([[ARENA_XMIN, ARENA_XMAX],
[ARENA_YMIN, ARENA_YMAX],
[ARENA_XMIN, ARENA_XMAX],
[ARENA_YMIN, ARENA_YMAX]])
else:
bounds = np.array([[ARENA_XMIN, ARENA_XMAX],
[ARENA_YMIN, ARENA_YMAX]])
if transition == 'sticky':
transition_kwargs = dict(alpha=sticky_alpha, kappa=sticky_kappa)
else:
transition_kwargs = None
obs = DynamicLocationObservation(K=K, D=D, M=M, bounds=bounds)
model = HMM(K=K,
D=D,
M=M,
transition=transition,
observation=obs,
transition_kwargs=transition_kwargs)
if initialize_mu:
if mouse == 'both':
assert K == (n_x * n_y)**2, "K should be equal to (n_x*n_y)**2"
else:
assert K == n_x * n_y, "K should be equal to n_x * n_y"
cluster_locations = np.array([((x_grids[i] + x_grids[i + 1]) / 2,
(y_grids[j] + y_grids[j + 1]) / 2)
for i in range(n_x)
for j in range(n_y)])
if mouse == "both":
cluster_locations = np.array([
np.concatenate((loc_a, loc_b))
for loc_a in cluster_locations
for loc_b in cluster_locations
])
assert cluster_locations.shape == (K, D)
train_mu = not not_train_mu
model.observation.transformation.mus_loc = \
torch.tensor(cluster_locations, dtype=torch.float64, requires_grad=train_mu)
else:
cluster_locations = get_np(
model.observation.transformation.mus_loc)
# save experiment params
exp_params = {
"job_name": job_name,
'downsample_n': downsample_n,
"mouse": mouse,
"load_model": load_model,
"load_model_dir": load_model_dir,
"load_opt_dir": load_opt_dir,
"transition": transition,
"sticky_alpha": sticky_alpha,
"sticky_kappa": sticky_kappa,
"K": K,
"n_x": n_x,
"n_y": n_y,
"x_grids": x_grids,
"y_grids": y_grids,
"cluster_locations": cluster_locations,
"not_train_mu": not_train_mu,
"initialize_mu": initialize_mu,
"train_model": train_model,
"pbar_update_interval": pbar_update_interval,
"list_of_num_iters": list_of_num_iters,
"list_of_lr": list_of_lr,
"video_clip_start": video_clip_start,
"video_clip_end": video_clip_end,
"sample_T": sample_T
}
print("Experiment params:")
print(exp_params)
rslt_dir = addDateTime("rslts/dynamic_loc/" + job_name)
rslt_dir = os.path.join(repo_dir, rslt_dir)
if not os.path.exists(rslt_dir):
os.makedirs(rslt_dir)
print("Making result directory...")
print("Saving to rlst_dir: ", rslt_dir)
with open(rslt_dir + "/exp_params.json", "w") as f:
json.dump(exp_params, f, indent=4, cls=NumpyEncoder)
##################### training ############################
if train_model:
initial_dir = rslt_dir + "/initial"
if not os.path.exists(initial_dir):
os.makedirs(initial_dir)
print("\nCreating initial directory...")
rslt_saving(initial_dir, model, data, mouse, sample_T, False, [],
quiver_scale, x_grids, y_grids)
print("start training")
list_of_losses = []
if load_opt_dir != "":
opt = joblib.load(load_opt_dir)
else:
opt = None
for i, (num_iters, lr) in enumerate(zip(list_of_num_iters,
list_of_lr)):
losses, opt = model.fit(data,
optimizer=opt,
method='adam',
num_iters=num_iters,
lr=lr,
pbar_update_interval=pbar_update_interval)
list_of_losses.append(losses)
checkpoint_dir = rslt_dir + "/checkpoint_{}".format(i)
if not os.path.exists(checkpoint_dir):
os.makedirs(checkpoint_dir)
print("Creating checkpoint_{} directory...".format(i))
# save model and opt
joblib.dump(model, checkpoint_dir + "/model")
joblib.dump(opt, checkpoint_dir + "/optimizer")
# save rest
rslt_saving(checkpoint_dir, model, data, mouse, sample_T,
train_model, losses, quiver_scale, x_grids, y_grids)
else:
# only save the results
rslt_saving(rslt_dir, model, data, mouse, sample_T, False, [],
quiver_scale, x_grids, y_grids)
print("Finish running!")
def main(job_name, cuda_num, data_type, downsample_n, filter_traj, load_model,
load_model_dir, load_opt_dir, animal, transition, sticky_alpha,
sticky_kappa, k, x_grids, y_grids, n_x, n_y, rs, train_rs, train_vs,
train_model, pbar_update_interval, video_clips, prop_start_end,
held_out_proportion, torch_seed, np_seed, list_of_num_iters,
ckpts_not_to_save, list_of_lr, list_of_k_steps, sample_t,
quiver_scale):
if job_name is None:
raise ValueError("Please provide the job name.")
cuda_num = int(cuda_num)
device = torch.device(
"cuda:{}".format(cuda_num) if torch.cuda.is_available() else "cpu")
print("Using device {} \n\n".format(device))
assert animal in ['both', 'virgin', 'mother'], animal
K = k
sample_T = sample_t
rs = float(rs) if rs else None
video_clip_start, video_clip_end = [
float(x) for x in video_clips.split(",")
]
start, end = [float(x) for x in prop_start_end.split(",")]
list_of_num_iters = [int(x) for x in list_of_num_iters.split(",")]
list_of_lr = [float(x) for x in list_of_lr.split(",")]
# TODO: fix for no k_steps, k > 0
list_of_k_steps = [int(x) for x in list_of_k_steps.split(",")
] if list_of_k_steps else []
assert len(list_of_num_iters) == len(
list_of_lr
), "Length of list_of_num_iters must match length of list_of_lr."
for lr in list_of_lr:
if lr > 1:
raise ValueError("Learning rate should not be larger than 1!")
ckpts_not_to_save = [int(x) for x in ckpts_not_to_save.split(',')
] if ckpts_not_to_save else []
repo = git.Repo(
'.', search_parent_directories=True) # SocialBehaviorectories=True)
repo_dir = repo.working_tree_dir # SocialBehavior
torch.manual_seed(torch_seed)
np.random.seed(np_seed)
########################## data ########################
if data_type == 'full':
data_dir = repo_dir + '/SocialBehaviorptc/data/trajs_all'
traj = joblib.load(data_dir)
traj = traj[int(36000 * video_clip_start):int(36000 * video_clip_end)]
elif data_type == 'selected_010_virgin':
assert animal == 'virgin', "animal much be 'virgin', but got {}.".format(
animal)
data_dir = repo_dir + '/SocialBehaviorptc/data/traj_010_virgin_selected'
traj = joblib.load(data_dir)
T = len(traj)
traj = traj[int(T * start):int(T * end)]
else:
raise ValueError("unsupported data type: {}".format(data_type))
if animal == 'virgin':
traj = traj[:, 0:2]
elif animal == 'mother':
traj = traj[:, 2:4]
traj = downsample(traj, downsample_n)
if filter_traj:
traj = filter_traj_by_speed(traj, q1=0.99, q2=0.99)
data = torch.tensor(traj, dtype=torch.float64, device=device)
assert 0 <= held_out_proportion <= 0.4, \
"held_out-portion should be between 0 and 0.4 (inclusive), but is {}".format(held_out_proportion)
T = data.shape[0]
breakpoint = int(T * (1 - held_out_proportion))
training_data = data[:breakpoint]
valid_data = data[breakpoint:]
if sample_T is None:
sample_T = training_data.shape[0]
######################### model ####################
# model
D = data.shape[1]
assert D == 4 or D == 2, D
M = 0
if load_model:
print("Loading the model from ", load_model_dir)
model = joblib.load(load_model_dir)
obs = model.observation
K = model.K
assert D == model.D, "D = {}, model.D = {}".format(D, model.D)
n_x = len(obs.x_grids) - 1
n_y = len(obs.y_grids) - 1
bounds = obs.bounds
else:
print("Creating the model...")
if D == 4:
bounds = np.array([[ARENA_XMIN, ARENA_XMAX],
[ARENA_YMIN, ARENA_YMAX],
[ARENA_XMIN, ARENA_XMAX],
[ARENA_YMIN, ARENA_YMAX]])
else:
bounds = np.array([[ARENA_XMIN, ARENA_XMAX],
[ARENA_YMIN, ARENA_YMAX]])
# grids
if x_grids is None:
x_grid_gap = (ARENA_XMAX - ARENA_XMIN) / n_x
x_grids = np.array(
[ARENA_XMIN + i * x_grid_gap for i in range(n_x + 1)])
else:
x_grids = np.array([float(x) for x in x_grids.split(",")])
n_x = len(x_grids) - 1
if y_grids is None:
y_grid_gap = (ARENA_YMAX - ARENA_YMIN) / n_y
y_grids = np.array(
[ARENA_YMIN + i * y_grid_gap for i in range(n_y + 1)])
else:
y_grids = np.array([float(x) for x in y_grids.split(",")])
n_y = len(y_grids) - 1
mus_init = training_data[0] * torch.ones(
K, D, dtype=torch.float64, device=device)
if animal == 'both':
obs = GPObservation(K=K,
D=D,
mus_init=mus_init,
x_grids=x_grids,
y_grids=y_grids,
bounds=bounds,
rs=rs,
train_rs=train_rs,
train_vs=train_vs,
device=device)
else:
obs = GPObservationSingle(K=K,
D=D,
mus_init=mus_init,
x_grids=x_grids,
y_grids=y_grids,
bounds=bounds,
rs=rs,
train_rs=train_rs,
device=device)
if transition == 'sticky':
transition_kwargs = dict(alpha=sticky_alpha, kappa=sticky_kappa)
elif transition == 'grid':
transition_kwargs = dict(x_grids=x_grids, y_grids=y_grids)
else:
transition_kwargs = None
model = HMM(K=K,
D=D,
M=M,
transition=transition,
observation=obs,
transition_kwargs=transition_kwargs,
device=device)
# save experiment params
exp_params = {
"job_name": job_name,
"data_type": data_type,
'downsample_n': downsample_n,
"filter_traj": filter_traj,
"load_model": load_model,
"load_model_dir": load_model_dir,
"load_opt_dir": load_opt_dir,
"animal": animal,
"transition": transition,
"sticky_alpha": sticky_alpha,
"sticky_kappa": sticky_kappa,
"K": K,
"x_grids": x_grids,
"y_grids": y_grids,
"n_x": n_x,
"n_y": n_y,
"rs": rs,
"train_rs": train_rs,
"train_vs": train_vs,
"train_model": train_model,
"pbar_update_interval": pbar_update_interval,
"video_clip_start": video_clip_start,
"video_clip_end": video_clip_end,
"start_percentage": start,
"end_percentage": end,
"held_out_proportion": held_out_proportion,
"torch_seed": torch_seed,
"np_seed": np_seed,
"list_of_num_iters": list_of_num_iters,
"list_of_lr": list_of_lr,
"list_of_k_steps": list_of_k_steps,
"ckpts_not_to_save": ckpts_not_to_save,
"sample_T": sample_T,
"quiver_scale": quiver_scale
}
print("Experiment params:")
print(exp_params)
rslt_dir = addDateTime("rslts/gp/{}/{}".format(animal, job_name))
rslt_dir = os.path.join(repo_dir, rslt_dir)
if not os.path.exists(rslt_dir):
os.makedirs(rslt_dir)
print("Making result directory...")
print("Saving exp_params to rlst_dir: ", rslt_dir)
with open(rslt_dir + "/exp_params.json", "w") as f:
json.dump(exp_params, f, indent=4, cls=NumpyEncoder)
# compute memory
if transition == "grid":
print("Computing transition memory...")
joint_grid_idx = model.transition.get_grid_idx(training_data[:-1])
transition_memory_kwargs = dict(joint_grid_idx=joint_grid_idx)
valid_joint_grid_idx = model.transition.get_grid_idx(valid_data[:-1])
valid_data_transition_memory_kwargs = dict(
joint_grid_idx=valid_joint_grid_idx)
else:
transition_memory_kwargs = None
valid_data_transition_memory_kwargs = None
print("Computing observation memory...")
def get_memory_kwargs(data, train_rs):
if data is None or data.shape[0] == 0:
return {}
if animal == 'both':
kernel_distsq_xg_a = kernel_distsq(data[:-1, 0:2],
obs.inducing_points)
kernel_distsq_xg_b = kernel_distsq(data[:-1, 2:4],
obs.inducing_points)
kernel_distsq_dict = dict(kernel_distsq_xg_a=kernel_distsq_xg_a,
kernel_distsq_xg_b=kernel_distsq_xg_b)
else:
kernel_distsq_xg = kernel_distsq(data[:-1], obs.inducing_points)
kernel_distsq_dict = dict(kernel_distsq_xg=kernel_distsq_xg)
if train_rs:
return kernel_distsq_dict
else:
if animal == 'both':
Sigma_a, A_a = obs.get_gp_cache(data[:-1, 0:2], 0,
**kernel_distsq_dict)
Sigma_b, A_b = obs.get_gp_cache(data[:-1, 2:4], 1,
**kernel_distsq_dict)
return dict(Sigma_a=Sigma_a, A_a=A_a, Sigma_b=Sigma_b, A_b=A_b)
else:
Sigma, A = obs.get_gp_cache(data[:-1], **kernel_distsq_dict)
return dict(Sigma=Sigma, A=A)
memory_kwargs = get_memory_kwargs(training_data, train_rs)
valid_data_memory_kwargs = get_memory_kwargs(valid_data, train_rs)
log_prob = model.log_likelihood(
training_data,
transition_memory_kwargs=transition_memory_kwargs,
**memory_kwargs)
print("log_prob = {}".format(log_prob))
##################### training ############################
if train_model:
print("start training")
list_of_losses = []
if load_opt_dir != "":
opt = joblib.load(load_opt_dir)
else:
opt = None
for i, (num_iters, lr) in enumerate(zip(list_of_num_iters,
list_of_lr)):
training_losses, opt, valid_losses, rs_s = \
model.fit(training_data, optimizer=opt, method='adam', num_iters=num_iters, lr=lr,
pbar_update_interval=pbar_update_interval, valid_data=valid_data,
transition_memory_kwargs=transition_memory_kwargs,
valid_data_transition_memory_kwargs=valid_data_transition_memory_kwargs,
valid_data_memory_kwargs=valid_data_memory_kwargs, **memory_kwargs)
list_of_losses.append(training_losses)
checkpoint_dir = rslt_dir + "/checkpoint_{}".format(i)
if not os.path.exists(checkpoint_dir):
os.makedirs(checkpoint_dir)
print("Creating checkpoint_{} directory...".format(i))
# save model and opt
joblib.dump(model, checkpoint_dir + "/model")
joblib.dump(opt, checkpoint_dir + "/optimizer")
# save losses
losses = dict(training_loss=training_losses,
valid_loss=valid_losses)
joblib.dump(losses, checkpoint_dir + "/losses")
plt.figure()
plt.plot(training_losses)
plt.title("training loss")
plt.savefig(checkpoint_dir + "/training_losses.jpg")
plt.close()
plt.figure()
plt.plot(valid_losses)
plt.title("validation loss")
plt.savefig(checkpoint_dir + "/valid_losses.jpg")
plt.close()
rs_s = np.array(rs_s)
plt.figure(figsize=(10, 5))
plt.subplot(1, 2, 1)
plt.title("lengthscale(x)")
for k in range(K):
plt.plot(rs_s[:, k, 0], label='k={}'.format(k))
plt.legend()
plt.subplot(1, 2, 2)
plt.title("lengthscale(y)")
for k in range(K):
plt.plot(rs_s[:, k, 1], label='k={}'.format(k))
plt.legend()
plt.savefig(checkpoint_dir + "/observation_rs.jpg")
plt.close()
# save rest
if i in ckpts_not_to_save:
print("ckpt {}: skip!\n".format(i))
continue
with torch.no_grad():
rslt_saving(rslt_dir=checkpoint_dir,
model=model,
data=training_data,
animal=animal,
memory_kwargs=memory_kwargs,
list_of_k_steps=list_of_k_steps,
sample_T=sample_T,
quiver_scale=quiver_scale,
valid_data=valid_data,
transition_memory_kwargs=transition_memory_kwargs,
valid_data_transition_memory_kwargs=
valid_data_transition_memory_kwargs,
valid_data_memory_kwargs=valid_data_memory_kwargs,
device=device)
else:
# only save the results
rslt_saving(rslt_dir=rslt_dir,
model=model,
data=training_data,
animal=animal,
memory_kwargs=memory_kwargs,
list_of_k_steps=list_of_k_steps,
sample_T=sample_T,
quiver_scale=quiver_scale,
valid_data=valid_data,
transition_memory_kwargs=transition_memory_kwargs,
valid_data_transition_memory_kwargs=
valid_data_transition_memory_kwargs,
valid_data_memory_kwargs=valid_data_memory_kwargs,
device=device)
print("Finish running!")
def main(job_name, lstm_model_type, downsample_n, filter_traj, load_model,
load_model_dir, load_opt_dir, transition, sticky_alpha, sticky_kappa,
acc_factor, lags, dh, dhs, k, x_grids, y_grids, n_x, n_y, train_model,
pbar_update_interval, video_clips, torch_seed, np_seed,
list_of_num_iters, list_of_lr, list_of_k_steps, sample_t,
quiver_scale):
if job_name is None:
raise ValueError("Please provide the job name.")
assert lstm_model_type in [
"uni", "multi"
], "lstm_model_dypte must be choosen from 'uni' and 'multi'."
K = k
sample_T = sample_t
video_clip_start, video_clip_end = [
float(x) for x in video_clips.split(",")
]
if dhs == 'none':
dhs = None
else:
dhs = [int(d_hidden) for d_hidden in dhs.split(",")]
list_of_num_iters = [int(x) for x in list_of_num_iters.split(",")]
list_of_lr = [float(x) for x in list_of_lr.split(",")]
list_of_k_steps = [int(x) for x in list_of_k_steps.split(",")]
assert len(list_of_num_iters) == len(list_of_lr), \
"Length of list_of_num_iters must match length of list-of_lr," \
" but we have list_of_num_iters = {}, and list_of_lr = {}".format(list_of_num_iters, list_of_lr)
for lr in list_of_lr:
if lr > 1:
raise ValueError("Learning rate should not be larger than 1!")
repo = git.Repo(
'.', search_parent_directories=True) # SocialBehaviorectories=True)
repo_dir = repo.working_tree_dir # SocialBehavior
torch.manual_seed(torch_seed)
np.random.seed(np_seed)
########################## data ########################
data_dir = repo_dir + '/SocialBehaviorptc/data/trajs_all'
trajs = joblib.load(data_dir)
start = int(36000 * video_clip_start)
end = int(36000 * video_clip_end)
traj = trajs[start:end]
traj = downsample(traj, downsample_n)
if filter_traj:
traj = filter_traj_by_speed(traj, q1=0.99, q2=0.99)
data = torch.tensor(traj, dtype=torch.float64)
######################### model ####################
# model
D = 4
M = 0
Df = 4
if load_model:
print("Loading the model from ", load_model_dir)
model = joblib.load(load_model_dir)
tran = model.observation.transformation
assert isinstance(tran, (LSTMTransformation, UniLSTMTransformation)),\
"tran should be {}, but is {}".format(LSTMTransformation, type(tran))
K = model.K
acc_factor = tran.acc_factor
lstm_model_type = "uni" if isinstance(
tran, UniLSTMTransformation) else "multi"
else:
print("Creating the model...")
bounds = np.array([[ARENA_XMIN, ARENA_XMAX], [ARENA_YMIN, ARENA_YMAX],
[ARENA_XMIN, ARENA_XMAX], [ARENA_YMIN, ARENA_YMAX]])
if acc_factor is None:
acc_factor = downsample_n * 10
tran = LSTMTransformation(K=K,
D=D,
Df=Df,
feature_vec_func=f_corner_vec_func,
lags=lags,
dh=dh,
dhs=dhs,
acc_factor=acc_factor)
obs = ARTruncatedNormalObservation(K=K,
D=D,
M=M,
lags=lags,
bounds=bounds,
transformation=tran)
if transition == 'sticky':
transition_kwargs = dict(alpha=sticky_alpha, kappa=sticky_kappa)
else:
transition_kwargs = None
model = HMM(K=K,
D=D,
M=M,
transition=transition,
observation=obs,
transition_kwargs=transition_kwargs)
model.observation.mus_init = data[0] * torch.ones(
K, D, dtype=torch.float64)
# grids
if x_grids is None:
x_grid_gap = (ARENA_XMAX - ARENA_XMIN) / n_x
x_grids = np.array(
[ARENA_XMIN + i * x_grid_gap for i in range(n_x + 1)])
else:
x_grids = np.array([float(x) for x in x_grids.split(",")])
n_x = len(x_grids) - 1
if y_grids is None:
y_grid_gap = (ARENA_YMAX - ARENA_YMIN) / n_y
y_grids = np.array(
[ARENA_YMIN + i * y_grid_gap for i in range(n_y + 1)])
else:
y_grids = np.array([float(x) for x in y_grids.split(",")])
n_y = len(y_grids) - 1
# save experiment params
exp_params = {
"job_name": job_name,
"lstm_model_tyoe": lstm_model_type,
'downsample_n': downsample_n,
"filter_traj": filter_traj,
"load_model": load_model,
"load_model_dir": load_model_dir,
"load_opt_dir": load_opt_dir,
"transition": transition,
"sticky_alpha": sticky_alpha,
"sticky_kappa": sticky_kappa,
"acc_factor": acc_factor,
"K": K,
"lags": lags,
"dh": dh,
"dhs": dhs,
"n_x": n_x,
"n_y": n_y,
"x_grids": x_grids,
"y_grids": y_grids,
"train_model": train_model,
"pbar_update_interval": pbar_update_interval,
"list_of_num_iters": list_of_num_iters,
"list_of_lr": list_of_lr,
"video_clip_start": video_clip_start,
"video_clip_end": video_clip_end,
"list_of_k_steps": list_of_k_steps,
"sample_T": sample_T
}
print("Experiment params:")
print(exp_params)
rslt_dir = addDateTime("rslts/lstm/{}/{}".format(lstm_model_type,
job_name))
rslt_dir = os.path.join(repo_dir, rslt_dir)
if not os.path.exists(rslt_dir):
os.makedirs(rslt_dir)
print("Making result directory...")
print("Saving to rlst_dir: ", rslt_dir)
with open(rslt_dir + "/exp_params.json", "w") as f:
json.dump(exp_params, f, indent=4, cls=NumpyEncoder)
# compute memory
print("Computing memory...")
feature_vecs_a = f_corner_vec_func(data[:-1, 0:2]) # (T, Df, 2)
feature_vecs_b = f_corner_vec_func(data[:-1, 2:4]) # (T, Df, 2)
feature_vecs = (feature_vecs_a, feature_vecs_b)
packed_data = get_packed_data(data[:-1], lags=lags)
memory_kwargs = dict(packed_data=packed_data, feature_vecs=feature_vecs)
##################### training ############################
if train_model:
print("start training")
list_of_losses = []
if load_opt_dir != "":
opt = joblib.load(load_opt_dir)
else:
opt = None
for i, (num_iters, lr) in enumerate(zip(list_of_num_iters,
list_of_lr)):
losses, opt = model.fit(data,
optimizer=opt,
method='adam',
num_iters=num_iters,
lr=lr,
pbar_update_interval=pbar_update_interval,
**memory_kwargs)
list_of_losses.append(losses)
checkpoint_dir = rslt_dir + "/checkpoint_{}".format(i)
if not os.path.exists(checkpoint_dir):
os.makedirs(checkpoint_dir)
print("Creating checkpoint_{} directory...".format(i))
# save model and opt
joblib.dump(model, checkpoint_dir + "/model")
joblib.dump(opt, checkpoint_dir + "/optimizer")
# save rest
rslt_saving(checkpoint_dir,
model,
data,
memory_kwargs,
list_of_k_steps,
sample_T,
train_model,
losses,
quiver_scale,
x_grids=x_grids,
y_grids=y_grids)
else:
# only save the results
rslt_saving(rslt_dir,
model,
data,
memory_kwargs,
list_of_k_steps,
sample_T,
False, [],
quiver_scale,
x_grids=x_grids,
y_grids=y_grids)
print("Finish running!")