-
Notifications
You must be signed in to change notification settings - Fork 11
/
train_command_generation.py
151 lines (130 loc) · 7.14 KB
/
train_command_generation.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
import datetime
import os
import time
import json
import math
import numpy as np
from os.path import join as pjoin
from command_generation_dataset import CommandGenerationData
from agent import Agent
import generic
import evaluate
def train():
time_1 = datetime.datetime.now()
config = generic.load_config()
env = CommandGenerationData(config)
env.split_reset("train")
agent = Agent(config)
agent.zero_noise()
ave_train_loss = generic.HistoryScoreCache(capacity=500)
# visdom
if config["general"]["visdom"]:
import visdom
viz = visdom.Visdom()
plt_win = None
eval_plt_win = None
viz_loss, viz_eval_exact_f1, viz_eval_soft_f1 = [], [], []
episode_no = 0
batch_no = 0
output_dir = "."
data_dir = "."
json_file_name = agent.experiment_tag.replace(" ", "_")
best_eval_exact_f1_so_far, best_eval_soft_f1_so_far, best_training_loss_so_far = 0.0, 0.0, 10000.0
# load model from checkpoint
if agent.load_pretrained:
if os.path.exists(output_dir + "/" + agent.experiment_tag + "_model.pt"):
agent.load_pretrained_model(output_dir + "/" + agent.experiment_tag + "_model.pt", load_partial_graph=False)
elif os.path.exists(data_dir + "/" + agent.load_graph_update_model_from_tag + ".pt"):
agent.load_pretrained_model(data_dir + "/" + agent.load_graph_update_model_from_tag + ".pt", load_partial_graph=False)
try:
while(True):
if episode_no > agent.max_episode:
break
agent.train()
observation_strings, triplets, target_strings = env.get_batch()
curr_batch_size = len(observation_strings)
_, loss = agent.command_generation_teacher_force(observation_strings, triplets, target_strings)
ave_train_loss.push(loss)
# lr schedule
# learning_rate = 1.0 * (generic.power(agent.model.block_hidden_dim, -0.5) * min(generic.power(batch_no, -0.5), batch_no * generic.power(agent.learning_rate_warmup_until, -1.5)))
if batch_no < agent.learning_rate_warmup_until:
cr = agent.init_learning_rate / math.log2(agent.learning_rate_warmup_until)
learning_rate = cr * math.log2(batch_no + 1)
else:
learning_rate = agent.init_learning_rate
for param_group in agent.optimizer.param_groups:
param_group['lr'] = learning_rate
episode_no += curr_batch_size
batch_no += 1
if agent.report_frequency == 0 or (episode_no % agent.report_frequency > (episode_no - curr_batch_size) % agent.report_frequency):
continue
eval_f1_exact, eval_f1_soft = 0.0, 0.0
if episode_no % agent.report_frequency <= (episode_no - curr_batch_size) % agent.report_frequency:
if agent.run_eval:
eval_f1_exact, eval_f1_soft = evaluate.evaluate_pretrained_command_generation(env, agent, "valid")
env.split_reset("train")
# if run eval, then save model by eval accuracy
if eval_f1_exact > best_eval_exact_f1_so_far:
best_eval_exact_f1_so_far = eval_f1_exact
best_eval_soft_f1_so_far = eval_f1_soft
agent.save_model_to_path(output_dir + "/" + agent.experiment_tag + "_model.pt")
elif eval_f1_exact == best_eval_exact_f1_so_far and eval_f1_soft > best_eval_soft_f1_so_far:
best_eval_soft_f1_so_far = eval_f1_soft
agent.save_model_to_path(output_dir + "/" + agent.experiment_tag + "_model.pt")
else:
if loss < best_training_loss_so_far:
best_training_loss_so_far = loss
agent.save_model_to_path(output_dir + "/" + agent.experiment_tag + "_model.pt")
time_2 = datetime.datetime.now()
print("Episode: {:3d} | time spent: {:s} | loss: {:2.3f} | valid exact f1: {:2.3f} | valid soft f1: {:2.3f}".format(episode_no, str(time_2 - time_1).rsplit(".")[0], loss, eval_f1_exact, eval_f1_soft))
# plot using visdom
if config["general"]["visdom"]:
viz_loss.append(ave_train_loss.get_avg())
viz_eval_exact_f1.append(eval_f1_exact)
viz_eval_soft_f1.append(eval_f1_soft)
viz_x = np.arange(len(viz_loss)).tolist()
viz_eval_x = np.arange(len(viz_eval_exact_f1)).tolist()
if plt_win is None:
plt_win = viz.line(X=viz_x, Y=viz_loss,
opts=dict(title=agent.experiment_tag + "_loss"),
name="training loss")
else:
viz.line(X=[len(viz_loss) - 1], Y=[viz_loss[-1]],
opts=dict(title=agent.experiment_tag + "_loss"),
win=plt_win,
update='append', name="training loss")
if eval_plt_win is None:
eval_plt_win = viz.line(X=viz_eval_x, Y=viz_eval_exact_f1,
opts=dict(title=agent.experiment_tag + "_exact_f1"),
name="eval exact f1")
viz.line(X=viz_eval_x, Y=viz_eval_soft_f1,
opts=dict(title=agent.experiment_tag + "_soft_f1"),
win=eval_plt_win, update='append', name="eval soft f1")
else:
viz.line(X=[len(viz_eval_exact_f1) - 1], Y=[viz_eval_exact_f1[-1]],
opts=dict(title=agent.experiment_tag + "_exact_f1"),
win=eval_plt_win,
update='append', name="eval exact f1")
viz.line(X=[len(viz_eval_soft_f1) - 1], Y=[viz_eval_soft_f1[-1]],
opts=dict(title=agent.experiment_tag + "_soft_f1"),
win=eval_plt_win,
update='append', name="eval soft f1")
# write accuracies down into file
_s = json.dumps({"time spent": str(time_2 - time_1).rsplit(".")[0],
"loss": str(ave_train_loss.get_avg()),
"eval exact f1": str(eval_f1_exact),
"eval soft f1": str(eval_f1_soft)})
with open(output_dir + "/" + json_file_name + '.json', 'a+') as outfile:
outfile.write(_s + '\n')
outfile.flush()
# At any point you can hit Ctrl + C to break out of training early.
except KeyboardInterrupt:
print('--------------------------------------------')
print('Exiting from training early...')
if agent.run_eval:
if os.path.exists(output_dir + "/" + agent.experiment_tag + "_model.pt"):
print('Evaluating on test set and saving log...')
agent.load_pretrained_model(output_dir + "/" + agent.experiment_tag + "_model.pt", load_partial_graph=False)
_, _ = evaluate.evaluate_pretrained_command_generation(env, agent, "test", verbose=True)
if __name__ == '__main__':
train()