-
Notifications
You must be signed in to change notification settings - Fork 2
/
launch_experiment.py
385 lines (359 loc) · 16.6 KB
/
launch_experiment.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
import os
import pathlib
import numpy as np
import click
import json
import torch
from rlkit.envs import ENVS
from rlkit.envs.wrappers import NormalizedBoxEnv
from rlkit.torch.sac.policies import TanhGaussianPolicy
from rlkit.torch.networks import FlattenMlp, MlpEncoder, RecurrentEncoder, VRNNEncoder
from rlkit.torch.sac.sac import PEARLSoftActorCritic
from rlkit.torch.sac.agent import PEARLAgent
from rlkit.launchers.launcher_util import setup_logger
import rlkit.torch.pytorch_util as ptu
from configs.default import default_config
import random
import pickle
import pdb
import sys
def set_global_seeds(i):
np.random.seed(i)
random.seed(i)
torch.manual_seed(i)
torch.backends.cudnn.deterministic = True
torch.backends.cudnn.benchmark = False
def read_dim(s):
a, b, c, d, e = s.split('.')
return [int(a), int(b), int(c), int(d), int(e)]
def gpu_optimizer(optimizer):
for state in optimizer.state.values():
for k, v in state.items():
if torch.is_tensor(v):
state[k] = v.cuda()
def experiment(variant):
print (variant['env_name'])
print (variant['env_params'])
env = NormalizedBoxEnv(ENVS[variant['env_name']](**variant['env_params']))
tasks = env.get_all_task_idx()
obs_dim = int(np.prod(env.observation_space.shape))
action_dim = int(np.prod(env.action_space.shape))
cont_latent_dim, num_cat, latent_dim, num_dir, dir_latent_dim = read_dim(variant['global_latent'])
r_cont_dim, r_n_cat, r_cat_dim, r_n_dir, r_dir_dim = read_dim(variant['vrnn_latent'])
reward_dim = 1
net_size = variant['net_size']
recurrent = variant['algo_params']['recurrent']
glob = variant['algo_params']['glob']
rnn = variant['rnn']
vrnn_latent = variant['vrnn_latent']
encoder_model = MlpEncoder
if recurrent:
if variant['vrnn_constraint'] == 'logitnormal':
output_size = r_cont_dim * 2 + r_n_cat * r_cat_dim + r_n_dir * r_dir_dim * 2
else:
output_size = r_cont_dim * 2 + r_n_cat * r_cat_dim + r_n_dir * r_dir_dim
if variant['rnn_sample'] == 'batch_sampling':
if variant['algo_params']['use_next_obs']:
input_size = (2 * obs_dim + action_dim + reward_dim) * variant['temp_res']
else:
input_size = (obs_dim + action_dim + reward_dim) * variant['temp_res']
else:
if variant['algo_params']['use_next_obs']:
input_size = (2 * obs_dim + action_dim + reward_dim)
else:
input_size = (obs_dim + action_dim + reward_dim)
if rnn == 'rnn':
recurrent_model = RecurrentEncoder
recurrent_context_encoder = recurrent_model(
hidden_sizes=[net_size, net_size, net_size],
input_size=input_size,
output_size = output_size
)
elif rnn == 'vrnn':
recurrent_model = VRNNEncoder
recurrent_context_encoder = recurrent_model(
hidden_sizes=[net_size, net_size, net_size],
input_size=input_size,
output_size=output_size,
temperature=variant['temperature'],
vrnn_latent=variant['vrnn_latent'],
vrnn_constraint=variant['vrnn_constraint'],
r_alpha=variant['vrnn_alpha'],
r_var=variant['vrnn_var'],
)
else:
recurrent_context_encoder = None
ptu.set_gpu_mode(variant['util_params']['use_gpu'], variant['util_params']['gpu_id'])
if glob:
if dir_latent_dim > 0 and variant['constraint'] == 'logitnormal':
output_size = cont_latent_dim * 2 + num_cat * latent_dim + num_dir * dir_latent_dim * 2
else:
output_size = cont_latent_dim * 2 + num_cat * latent_dim + num_dir * dir_latent_dim
if variant['algo_params']['use_next_obs']:
input_size = 2 * obs_dim + action_dim + reward_dim
else:
input_size = obs_dim + action_dim + reward_dim
global_context_encoder = encoder_model(
hidden_sizes=[net_size, net_size, net_size],
input_size=input_size,
output_size=output_size,
)
else:
global_context_encoder = None
qf1 = FlattenMlp(
hidden_sizes=[net_size, net_size, net_size],
input_size=obs_dim + action_dim + latent_dim*num_cat + cont_latent_dim + dir_latent_dim*num_dir \
+ r_n_cat * r_cat_dim + r_cont_dim + r_n_dir * r_dir_dim,
output_size=1,
)
qf2 = FlattenMlp(
hidden_sizes=[net_size, net_size, net_size],
input_size=obs_dim + action_dim + latent_dim*num_cat + cont_latent_dim + dir_latent_dim*num_dir \
+ r_n_cat * r_cat_dim + r_cont_dim + r_n_dir * r_dir_dim,
output_size=1,
)
target_qf1 = FlattenMlp(
hidden_sizes=[net_size, net_size, net_size],
input_size=obs_dim + action_dim + latent_dim*num_cat + cont_latent_dim + dir_latent_dim*num_dir \
+ r_n_cat * r_cat_dim + r_cont_dim + r_n_dir * r_dir_dim,
output_size=1,
)
target_qf2 = FlattenMlp(
hidden_sizes=[net_size, net_size, net_size],
input_size=obs_dim + action_dim + latent_dim*num_cat + cont_latent_dim + dir_latent_dim*num_dir \
+ r_n_cat * r_cat_dim + r_cont_dim + r_n_dir * r_dir_dim,
output_size=1,
)
policy = TanhGaussianPolicy(
hidden_sizes=[net_size, net_size, net_size],
obs_dim=obs_dim + latent_dim*num_cat + cont_latent_dim + dir_latent_dim*num_dir \
+ r_n_cat * r_cat_dim + r_cont_dim + r_n_dir * r_dir_dim,
latent_dim=latent_dim*num_cat + cont_latent_dim + dir_latent_dim*num_dir \
+ r_n_cat * r_cat_dim + r_cont_dim + r_n_dir * r_dir_dim,
action_dim=action_dim,
)
agent = PEARLAgent(
global_context_encoder,
recurrent_context_encoder,
variant['global_latent'],
variant['vrnn_latent'],
policy,
variant['temperature'],
variant['unitkl'],
variant['alpha'],
variant['constraint'],
variant['vrnn_constraint'],
variant['var'],
variant['vrnn_alpha'],
variant['vrnn_var'],
rnn,
variant['temp_res'],
variant['rnn_sample'],
variant['weighted_sample'],
**variant['algo_params']
)
if variant['path_to_weights'] is not None:
path = variant['path_to_weights']
with open(os.path.join(path, 'extra_data.pkl'), 'rb') as f:
extra_data = pickle.load(f)
variant['algo_params']['start_epoch'] = extra_data['epoch'] + 1
replay_buffer = extra_data['replay_buffer']
enc_replay_buffer = extra_data['enc_replay_buffer']
variant['algo_params']['_n_train_steps_total'] = extra_data['_n_train_steps_total']
variant['algo_params']['_n_env_steps_total'] = extra_data['_n_env_steps_total']
variant['algo_params']['_n_rollouts_total'] = extra_data['_n_rollouts_total']
else:
replay_buffer=None
enc_replay_buffer=None
algorithm = PEARLSoftActorCritic(
env=env,
train_tasks=list(tasks[:variant['n_train_tasks']]),
eval_tasks=list(tasks[-variant['n_eval_tasks']:]),
nets=[agent, qf1, qf2, target_qf1, target_qf2],
latent_dim=latent_dim,
replay_buffer=replay_buffer,
enc_replay_buffer=enc_replay_buffer,
temp_res=variant['temp_res'],
rnn_sample=variant['rnn_sample'],
**variant['algo_params']
)
if variant['path_to_weights'] is not None:
path = variant['path_to_weights']
if recurrent_context_encoder != None:
recurrent_context_encoder.load_state_dict(torch.load(os.path.join(path, 'recurrent_context_encoder.pth')))
if global_context_encoder != None:
global_context_encoder.load_state_dict(torch.load(os.path.join(path, 'global_context_encoder.pth')))
qf1.load_state_dict(torch.load(os.path.join(path, 'qf1.pth')))
qf2.load_state_dict(torch.load(os.path.join(path, 'qf2.pth')))
target_qf1.load_state_dict(torch.load(os.path.join(path, 'target_qf1.pth')))
target_qf2.load_state_dict(torch.load(os.path.join(path, 'target_qf2.pth')))
policy.load_state_dict(torch.load(os.path.join(path, 'policy.pth')))
if ptu.gpu_enabled():
algorithm.to()
DEBUG = variant['util_params']['debug']
os.environ['DEBUG'] = str(int(DEBUG))
exp_id = 'debug' if DEBUG else None
if variant.get('log_name', "") == "":
log_name = variant['env_name']
else:
log_name = variant['log_name']
experiment_log_dir = setup_logger(log_name, \
variant=variant, \
exp_id=exp_id, \
base_log_dir=variant['util_params']['base_log_dir'], \
config_log_dir=variant['util_params']['config_log_dir'], \
log_dir=variant['util_params']['log_dir'])
if variant['algo_params']['dump_eval_paths']:
pickle_dir = experiment_log_dir + '/eval_trajectories'
pathlib.Path(pickle_dir).mkdir(parents=True, exist_ok=True)
env.save_all_tasks(experiment_log_dir)
if variant['eval']:
algorithm._try_to_eval(0, eval_all=True, eval_train_offline=False, animated=True)
else:
algorithm.train()
def deep_update_dict(fr, to):
for k, v in fr.items():
if type(v) is dict:
deep_update_dict(v, to[k])
else:
to[k] = v
return to
@click.command()
@click.argument('config', default=None)
@click.option('--gpu', default=0)
@click.option('--debug', is_flag=True, default=False)
@click.option('--seed', default=0)
@click.option('--kl_anneal', default="none", help="none or mono or cycle, annealing of the KL loss")
@click.option('--temperature', default=0.33, help="temperature as in gumbel softmax")
@click.option('--logdir', default='output', help="prefix of the log dir")
@click.option('--kl_lambda', default=1., help="weight of the KL loss")
@click.option('--unitkl', is_flag=True, default=False, help="KL loss on the accumulated global context variables or on each single global context variable")
@click.option('--n_iteration', default=500, help='number of iterations')
@click.option('--alpha', default=0.7)
@click.option('--constraint', default='dirichlet', help='dirichlet or logitnormal for global encoder')
@click.option('--var', default=2.5)
@click.option('--eval', is_flag=True, default=False)
@click.option('--path_to_weights', default=None)
@click.option('--recurrent', is_flag=True, default=False, help='Use local encoder or not')
@click.option('--vrnn_latent', default='2.0.0.2.4', help='gaus-dim.num-cat.cat-dim.num-dir.dir-dim')
@click.option('--global_latent', default='2.0.0.2.4', help='gaus-dim.num-cat.cat-dim.num-dir.dir-dim')
@click.option('--rnn', default='rnn', help='rnn or vrnn or None, architecture of the local encoder')
@click.option('--traj_batch_size', default=16, help='Number of trajectories sampled in one update')
@click.option('--vrnn_constraint', default='dirichlet', help="logitnormal or dirichlet")
@click.option('--vrnn_alpha', default=0.7)
@click.option('--vrnn_var', default=2.5)
@click.option('--temp_res', default=10, help="Temporal resolution")
@click.option('--rnn_sample', default="full", help="full or full_wo_sampling or single_sampling or batch_sampling") # full: infer posterior and update local variable each step; full_wo_sampling: infer posterior each step but update local variable every temp_res steps; single_sampling: infer posterior every temp_res steps using only the contexts in the previous step; batch_sampling: infer posterior every temp_res steps using previous temp_res contexts.
@click.option('--resample_in_traj', is_flag=True, default=False, help='resample global context variable in one trajectory, can be used in model with only global encoder')
@click.option('--weighted_sample', is_flag=True, default=False, help="When calculating the global posterior, do we use weighted product of the PDF or not")
@click.option('--use_next_obs', is_flag=True, default=False, help='Context uses next observation or not')
def main(config, gpu, debug, seed, kl_anneal, temperature, logdir, kl_lambda,
unitkl,
n_iteration, alpha,
constraint,
var,
eval, path_to_weights,
recurrent, vrnn_latent, global_latent, rnn, traj_batch_size, vrnn_constraint,
vrnn_alpha, vrnn_var, temp_res, rnn_sample, resample_in_traj,
weighted_sample, use_next_obs,
):
cont_latent_size, num_cat, latent_size, num_dir, dir_latent_size = read_dim(global_latent)
glob = latent_size * num_cat + cont_latent_size + dir_latent_size * num_dir > 0
if resample_in_traj:
assert glob
assert kl_anneal in ['none', 'mono', 'cycle']
if not recurrent:
vrnn_latent = '0.0.0.0.0'
rnn = 'None'
traj_batch_size = -1
vrnn_constraint = None
vrnn_alpha = None
vrnn_var = None
if not resample_in_traj:
temp_res = None
rnn_sample = None
r_cont_dim, r_n_cat, r_cat_dim, r_n_dir, r_dir_dim = read_dim(vrnn_latent)
if recurrent:
temp_res = int(temp_res)
assert rnn_sample in ["full", "full_wo_sampling", "single_sampling", "batch_sampling"]
if rnn_sample == 'full':
temp_res = 1
if r_dir_dim > 0:
assert vrnn_constraint in ['logitnormal', 'dirichlet']
if vrnn_constraint == 'logitnormal':
vrnn_alpha = None
else:
vrnn_var = None
else:
vrnn_alpha = None
vrnn_var = None
vrnn_constraint = None
if resample_in_traj:
temp_res = int(temp_res)
if latent_size == 0:
num_cat = 0
if dir_latent_size == 0:
num_dir = 0
if dir_latent_size > 0:
assert constraint in ['dirichlet', 'logitnormal']
if constraint == 'logitnormal':
alpha = None
else:
var = None
else:
constraint = None
alpha = None
var = None
set_global_seeds(seed)
variant = default_config
if config:
with open(os.path.join(config)) as f:
exp_params = json.load(f)
variant = deep_update_dict(exp_params, variant)
if gpu != -1:
variant['util_params']['gpu_id'] = gpu
else:
variant['util_params']['use_gpu'] = False
variant['seed'] = seed
variant['temperature'] = temperature
variant['env_params']['seed'] = seed
variant['algo_params']['kl_anneal'] = kl_anneal
variant['util_params']['base_log_dir'] = logdir
variant["algo_params"]['kl_lambda'] = kl_lambda
variant['algo_params']['use_next_obs'] = use_next_obs
variant['unitkl'] = unitkl
variant['alpha'] = alpha
variant['var'] = var
variant['constraint'] = constraint
variant['eval'] = eval
variant['path_to_weights'] = path_to_weights
variant['algo_params']['recurrent'] = recurrent
variant['algo_params']['glob'] = glob
variant['vrnn_latent'] = vrnn_latent
variant['global_latent'] = global_latent
variant['rnn'] = rnn
variant['algo_params']['traj_batch_size'] = traj_batch_size
variant['vrnn_constraint'] = vrnn_constraint
variant['vrnn_alpha'] = vrnn_alpha
variant['vrnn_var'] = vrnn_var
variant['weighted_sample'] = weighted_sample
variant['util_params']['log_dir'] = None
variant['temp_res'] = temp_res
variant['rnn_sample'] = rnn_sample
variant['algo_params']['resample_in_traj'] = resample_in_traj
save_dir = 'global-{}-local-{}-{}-{}-temp-{}-{}'.format(global_latent, vrnn_latent, vrnn_constraint, rnn, temp_res, rnn_sample)
if resample_in_traj:
save_dir += '-resample'
if weighted_sample:
save_dir += '-ws'
save_dir = os.path.join(save_dir, 'seed-%s'%seed)
variant['util_params']['config_log_dir'] = save_dir
variant['meta'] = None
if eval:
variant['util_params']['config_log_dir'] = os.path.join('eval', variant['util_params']['config_log_dir'])
variant['util_params']['debug'] = debug
variant['algo_params']['num_iterations'] = int(n_iteration)
variant['algo_params']['debug'] = debug
experiment(variant)
if __name__ == "__main__":
main()