def get_all_metrics(smiles):
mols = [MolFromSmiles(s) for s in smiles]
scorer = NormalizedScorer()
scores, norm_scores = scorer.get_scores_from_mols(mols)
arom_rings = np.array([Descriptors.NumAromaticRings(m) for m in mols])
metrics = np.concatenate([scores.sum(axis=1)[:, None],
norm_scores.sum(axis=1)[:, None],
scores[:, 1][:, None],
norm_scores[:, 1][:, None],
arom_rings[:, None]],
axis=1)
return (smiles, metrics)
A simple reward to encourage larger molecule length
:param smiles: list of strings
:return: reward, list of float
'''
return np.array(
[-1 if num is None else num for num in num_aliphatic_rings(smiles)])
batch_size = 40
drop_rate = 0.3
molecules = True
grammar = 'new' #True#
settings = get_settings(molecules, grammar)
invalid_value = -3 * 3.5
scorer = NormalizedScorer(invalid_value=invalid_value,
sa_mult=0.0,
sa_thresh=-1.5,
normalize_scores=False)
max_steps = 277 #settings['max_seq_length']
def second_score(smiles):
pre_scores = 3 * 2.5 + scorer.get_scores(smiles)[0]
score = np.power(pre_scores.prod(1), 0.333)
for i in range(len(score)):
if np.isnan(score[i]):
score[i] = -1
return score
reward_fun = lambda x: 3 * 2.5 + scorer(
x
'''
if not len(smiles):
return -1 # an empty string is invalid for our purposes
atoms = num_aromatic_rings(smiles)
return [-1 if num is None else num + 0.5 for num in atoms]
batch_size = 20 # 20
drop_rate = 0.5
molecules = True
grammar_cache = 'hyper_grammar.pickle'
grammar = 'hypergraph:' + grammar_cache
settings = get_settings(molecules, grammar)
# max_steps = 277 # settings['max_seq_length']
invalid_value = -3.5
scorer = NormalizedScorer(invalid_value=invalid_value)
reward_fun = scorer #lambda x: np.ones(len(x)) # lambda x: reward_aromatic_rings(x)#
# later will run this ahead of time
gi = GrammarInitializer(grammar_cache)
# if True:
# gi.delete_cache()
# gi = GrammarInitializer(grammar_cache)
# max_steps_smiles = gi.init_grammar(1000)
max_steps = 30
model, gen_fitter, disc_fitter = train_policy_gradient(
molecules,
grammar,
EPOCHS=100,
BATCH_SIZE=batch_size,
reward_fun_on=reward_fun,
A simple reward to encourage larger molecule length
:param smiles: list of strings
:return: reward, list of float
'''
return np.array(
[-1 if num is None else num for num in num_aliphatic_rings(smiles)])
batch_size = 40
drop_rate = 0.3
molecules = True
grammar = 'new' #True#
settings = get_settings(molecules, grammar)
invalid_value = -3 * 3.5
scorer = NormalizedScorer(invalid_value=invalid_value,
sa_mult=20,
sa_thresh=0,
normalize_scores=True)
max_steps = 277 #settings['max_seq_length']
def second_score(smiles):
pre_scores = 3 * 2.5 + scorer.get_scores(smiles)[0]
score = np.power(pre_scores.prod(1), 0.333)
for i in range(len(score)):
if np.isnan(score[i]):
score[i] = -1
return score
reward_fun = lambda x: scorer(x) + np.array(
[-5 * max(x - 5, 0) for x in num_aromatic_rings(x)]
atoms = num_atoms(smiles)
return [-1 if num is None else num for num in atoms]
batch_size = 100
drop_rate = 0.3
molecules = True
grammar = True
settings = get_settings(molecules, grammar)
max_steps = 50 #settings['max_seq_length']
invalid_value = -7.0
task = SequenceGenerationTask(molecules = molecules,
grammar = grammar,
reward_fun = NormalizedScorer(settings['data_path'],
invalid_value=invalid_value),
batch_size = batch_size,
max_steps=50)
if grammar:
mask_gen = GrammarMaskGenerator(task.env._max_episode_steps, grammar=settings['grammar'])
else:
mask_gen = None
def a2c_sequence(name = 'a2c_sequence', task=None, body=None):
config = Config()
config.num_workers = batch_size # same thing as batch size
config.task_fn = lambda: task
config.optimizer_fn = lambda params: torch.optim.RMSprop(params, lr=0.0007)
import torch
import numpy as np
from rdkit import Chem
from rdkit.Chem import rdMolDescriptors as desc
from rdkit.Chem.Draw import MolToFile
import os, inspect
from generative_playground.molecules.rdkit_utils.rdkit_utils import NormalizedScorer
import numpy as np
scorer = NormalizedScorer()
root_location = os.path.dirname(os.path.abspath(inspect.getfile(inspect.currentframe())))
def model_process_fun(model_out, visdom, n):
# TODO: rephrase this to return a dict, instead of calling visdom directly
from rdkit import Chem
from rdkit.Chem.Draw import MolToFile
# actions, logits, rewards, terminals, info = model_out
smiles, valid = model_out['info']
valid = to_numpy(valid)
total_rewards = to_numpy(model_out['rewards'])
if len(total_rewards.shape) > 1:
total_rewards = total_rewards.sum(1)
best_ind = np.argmax(total_rewards)
this_smile = smiles[best_ind]
mol = Chem.MolFromSmiles(this_smile)
pic_save_path = os.path.realpath(root_location + '/images/' + 'tmp.svg')
if mol is not None:
try:
MolToFile(mol, pic_save_path, imageType='svg')
def train_policy_gradient_ppo(molecules=True,
grammar=True,
smiles_source='ZINC',
EPOCHS=None,
BATCH_SIZE=None,
reward_fun_on=None,
reward_fun_off=None,
max_steps=277,
lr_on=2e-4,
lr_discrim=1e-4,
discrim_wt=2,
p_thresh=0.5,
drop_rate=0.0,
plot_ignore_initial=0,
randomize_reward=False,
save_file_root_name=None,
reward_sm=0.0,
preload_file_root_name=None,
anchor_file=None,
anchor_weight=0.0,
decoder_type='action',
plot_prefix='',
dashboard='policy gradient',
smiles_save_file=None,
on_policy_loss_type='best',
priors=True,
node_temperature_schedule=lambda x: 1.0,
rule_temperature_schedule=lambda x: 1.0,
eps=0.0,
half_float=False,
extra_repetition_penalty=0.0):
root_location = os.path.dirname(
os.path.abspath(inspect.getfile(inspect.currentframe())))
root_location = root_location + '/../../'
def full_path(x):
return os.path.realpath(root_location + 'pretrained/' + x)
if save_file_root_name is not None:
gen_save_file = save_file_root_name + '_gen.h5'
disc_save_file = save_file_root_name + '_disc.h5'
if preload_file_root_name is not None:
gen_preload_file = preload_file_root_name + '_gen.h5'
disc_preload_file = preload_file_root_name + '_disc.h5'
settings = get_settings(molecules=molecules, grammar=grammar)
codec = get_codec(molecules, grammar, settings['max_seq_length'])
discrim_model = GraphDiscriminator(codec.grammar, drop_rate=drop_rate)
if False and preload_file_root_name is not None:
try:
preload_path = full_path(disc_preload_file)
discrim_model.load_state_dict(torch.load(preload_path),
strict=False)
print('Discriminator weights loaded successfully!')
except Exception as e:
print('failed to load discriminator weights ' + str(e))
zinc_data = get_smiles_from_database(source=smiles_source)
zinc_set = set(zinc_data)
lookbacks = [BATCH_SIZE, 10 * BATCH_SIZE, 100 * BATCH_SIZE]
history_data = [deque(['O'], maxlen=lb) for lb in lookbacks]
def originality_mult(smiles_list):
out = []
for s in smiles_list:
if s in zinc_set:
out.append(0.5)
elif s in history_data[0]:
out.append(0.5)
elif s in history_data[1]:
out.append(0.70)
elif s in history_data[2]:
out.append(0.85)
else:
out.append(1.0)
return np.array(out)
def sigmoid(x):
return 1 / (1 + np.exp(-x))
def discriminator_reward_mult(smiles_list):
orig_state = discrim_model.training
discrim_model.eval()
discrim_out_logits = discrim_model(smiles_list)['p_zinc']
discrim_probs = F.softmax(discrim_out_logits, dim=1)
prob_zinc = discrim_probs[:, 1].detach().cpu().numpy()
if orig_state:
discrim_model.train()
return prob_zinc
def apply_originality_penalty(x, orig_mult):
assert x <= 1, "Reward must be no greater than 0"
if x > 0.5: # want to punish nearly-perfect scores less and less
out = math.pow(x, 1 / orig_mult)
else: # continuous join at 0.5
penalty = math.pow(0.5, 1 / orig_mult) - 0.5
out = x + penalty
out -= extra_repetition_penalty * (1 - 1 / orig_mult)
return out
def adj_reward(x):
if discrim_wt > 1e-5:
p = discriminator_reward_mult(x)
else:
p = 0
rwd = np.array(reward_fun_on(x))
orig_mult = originality_mult(x)
# we assume the reward is <=1, first term will dominate for reward <0, second for 0 < reward < 1
# reward = np.minimum(rwd/orig_mult, np.power(np.abs(rwd),1/orig_mult))
reward = np.array([
apply_originality_penalty(x, om) for x, om in zip(rwd, orig_mult)
])
out = reward + discrim_wt * p * orig_mult
return out
def adj_reward_old(x):
p = discriminator_reward_mult(x)
w = sigmoid(-(p - p_thresh) / 0.01)
if randomize_reward:
rand = np.random.uniform(size=p.shape)
w *= rand
reward = np.maximum(reward_fun_on(x), p_thresh)
weighted_reward = w * p + (1 - w) * reward
out = weighted_reward * originality_mult(x) #
return out
if EPOCHS is not None:
settings['EPOCHS'] = EPOCHS
if BATCH_SIZE is not None:
settings['BATCH_SIZE'] = BATCH_SIZE
task = SequenceGenerationTask(molecules=molecules,
grammar=grammar,
reward_fun=adj_reward,
batch_size=BATCH_SIZE,
max_steps=max_steps,
save_dataset=None)
node_policy = SoftmaxRandomSamplePolicy(temperature=torch.tensor(1.0),
eps=eps)
rule_policy = SoftmaxRandomSamplePolicy(temperature=torch.tensor(1.0),
eps=eps)
model = get_decoder(molecules,
grammar,
z_size=settings['z_size'],
decoder_hidden_n=200,
feature_len=codec.feature_len(),
max_seq_length=max_steps,
drop_rate=drop_rate,
batch_size=BATCH_SIZE,
decoder_type=decoder_type,
reward_fun=adj_reward,
task=task,
node_policy=node_policy,
rule_policy=rule_policy,
priors=priors)[0]
if preload_file_root_name is not None:
try:
preload_path = full_path(gen_preload_file)
model.load_state_dict(torch.load(preload_path, map_location='cpu'),
strict=False)
print('Generator weights loaded successfully!')
except Exception as e:
print('failed to load generator weights ' + str(e))
anchor_model = None
from generative_playground.molecules.rdkit_utils.rdkit_utils import NormalizedScorer
import numpy as np
scorer = NormalizedScorer()
if reward_fun_off is None:
reward_fun_off = reward_fun_on
# construct the loader to feed the discriminator
def make_callback(data):
def hc(inputs, model, outputs, loss_fn, loss):
graphs = outputs['graphs']
smiles = [g.to_smiles() for g in graphs]
for s in smiles: # only store unique instances of molecules so discriminator can't guess on frequency
if s not in data:
data.append(s)
return hc
class TemperatureCallback:
def __init__(self, policy, temperature_function):
self.policy = policy
self.counter = 0
self.temp_fun = temperature_function
def __call__(self, inputs, model, outputs, loss_fn, loss):
self.counter += 1
target_temp = self.temp_fun(self.counter)
self.policy.set_temperature(target_temp)
# need to have something there to begin with, else the DataLoader constructor barfs
def get_rl_fitter(model,
loss_obj,
train_gen,
save_path,
fit_plot_prefix='',
model_process_fun=None,
lr=None,
extra_callbacks=[],
loss_display_cap=float('inf'),
anchor_model=None,
anchor_weight=0):
nice_params = filter(lambda p: p.requires_grad, model.parameters())
optimizer = optim.Adam(nice_params, lr=lr, eps=1e-4)
scheduler = lr_scheduler.StepLR(optimizer, step_size=100, gamma=0.99)
if dashboard is not None:
metric_monitor = MetricPlotter(
plot_prefix=fit_plot_prefix,
loss_display_cap=loss_display_cap,
dashboard_name=dashboard,
plot_ignore_initial=plot_ignore_initial,
process_model_fun=model_process_fun,
smooth_weight=reward_sm)
else:
metric_monitor = None
checkpointer = Checkpointer(valid_batches_to_checkpoint=1,
save_path=save_path,
save_always=True)
fitter = fit_rl(train_gen=train_gen,
model=model,
optimizer=optimizer,
scheduler=scheduler,
epochs=EPOCHS,
loss_fn=loss_obj,
grad_clip=5,
half_float=half_float,
anchor_model=anchor_model,
anchor_weight=anchor_weight,
callbacks=[metric_monitor, checkpointer] +
extra_callbacks)
return fitter
def my_gen():
for _ in range(1000):
yield to_gpu(torch.zeros(BATCH_SIZE, settings['z_size']))
# the on-policy fitter
gen_extra_callbacks = [make_callback(d) for d in history_data]
if smiles_save_file is not None:
smiles_save_path = os.path.realpath(root_location + 'pretrained/' +
smiles_save_file)
gen_extra_callbacks.append(MoleculeSaver(smiles_save_path, gzip=True))
print('Saved SMILES to {}'.format(smiles_save_file))
if node_temperature_schedule is not None:
gen_extra_callbacks.append(
TemperatureCallback(node_policy, node_temperature_schedule))
if rule_temperature_schedule is not None:
gen_extra_callbacks.append(
TemperatureCallback(rule_policy, rule_temperature_schedule))
fitter1 = get_rl_fitter(
model,
PolicyGradientLoss(on_policy_loss_type), # last_reward_wgt=reward_sm),
GeneratorToIterable(my_gen),
full_path(gen_save_file),
plot_prefix + 'on-policy',
model_process_fun=model_process_fun,
lr=lr_on,
extra_callbacks=gen_extra_callbacks,
anchor_model=anchor_model,
anchor_weight=anchor_weight)
#
# # get existing molecule data to add training
pre_dataset = EvenlyBlendedDataset(
2 * [history_data[0]] + history_data[1:],
labels=False) # a blend of 3 time horizons
dataset = EvenlyBlendedDataset([pre_dataset, zinc_data], labels=True)
discrim_loader = DataLoader(dataset, shuffle=True, batch_size=50)
class MyLoss(nn.Module):
def __init__(self):
super().__init__()
self.celoss = nn.CrossEntropyLoss()
def forward(self, x):
# tmp = discriminator_reward_mult(x['smiles'])
# tmp2 = F.softmax(x['p_zinc'], dim=1)[:,1].detach().cpu().numpy()
# import numpy as np
# assert np.max(np.abs(tmp-tmp2)) < 1e-6
return self.celoss(x['p_zinc'].to(device),
x['dataset_index'].to(device))
fitter2 = get_rl_fitter(discrim_model,
MyLoss(),
IterableTransform(
discrim_loader, lambda x: {
'smiles': x['X'],
'dataset_index': x['dataset_index']
}),
full_path(disc_save_file),
plot_prefix + ' discriminator',
lr=lr_discrim,
model_process_fun=None)
def on_policy_gen(fitter, model):
while True:
# model.policy = SoftmaxRandomSamplePolicy()#bias=codec.grammar.get_log_frequencies())
yield next(fitter)
return model, fitter1, fitter2 #,on_policy_gen(fitter1, model)
from generative_playground.molecules.model_settings import get_settings
from generative_playground.molecules.train.pg.hypergraph.main_train_policy_gradient_minimal import train_policy_gradient
from generative_playground.codec.hypergraph_grammar import GrammarInitializer
batch_size = 8# 20
drop_rate = 0.5
molecules = True
grammar_cache = 'hyper_grammar_guac_10k_with_clique_collapse.pickle'#'hyper_grammar.pickle'
grammar = 'hypergraph:' + grammar_cache
# settings = get_settings(molecules, grammar)
# max_steps = 277 # settings['max_seq_length']
invalid_value = -3.5
atom_penalty = lambda x: -0.05*(np.maximum(37, np.array(num_atoms(x)))-37)
scorer = NormalizedScorer(invalid_value=invalid_value, normalize_scores=True)
reward_fun = lambda x: np.tanh(0.1*scorer(x)) + atom_penalty(x) # lambda x: reward_aromatic_rings(x)#
# later will run this ahead of time
# gi = GrammarInitializer(grammar_cache)
max_steps = 60
root_name = 'classic_logP'
model, gen_fitter, disc_fitter = train_policy_gradient(molecules,
grammar,
EPOCHS=100,
BATCH_SIZE=batch_size,
reward_fun_on=reward_fun,
max_steps=max_steps,
lr_on=3e-5,
lr_discrim=5e-4,
discrim_wt=0.3,
def reward_aliphatic_rings(smiles):
'''
A simple reward to encourage larger molecule length
:param smiles: list of strings
:return: reward, list of float
'''
return np.array([-1 if num is None else num for num in num_aliphatic_rings(smiles)])
batch_size = 40
drop_rate = 0.3
molecules = True
grammar = 'new'#True#
settings = get_settings(molecules, grammar)
invalid_value = -10
scorer = NormalizedScorer(invalid_value=invalid_value, sa_mult=0)
max_steps = 277 #settings['max_seq_length']
def second_score(smiles):
pre_scores = 2.5 + scorer.get_scores(smiles)[0]
score = np.power(pre_scores.prod(1), 0.333)
for i in range(len(score)):
if np.isnan(score[i]):
score[i] = -1
return score
reward_fun = lambda x: np.array([1 for _ in x])
model, fitter1, fitter2 = train_policy_gradient(molecules,
grammar,
EPOCHS=100,
def train_policy_gradient(molecules=True,
grammar=True,
EPOCHS=None,
BATCH_SIZE=None,
reward_fun_on=None,
reward_fun_off=None,
max_steps=277,
lr_on=2e-4,
lr_off=1e-4,
drop_rate=0.0,
plot_ignore_initial=0,
save_file=None,
preload_file=None,
anchor_file=None,
anchor_weight=0.0,
decoder_type='action',
plot_prefix='',
dashboard='policy gradient',
smiles_save_file=None,
on_policy_loss_type='best',
off_policy_loss_type='mean',
sanity_checks=True):
root_location = os.path.dirname(
os.path.abspath(inspect.getfile(inspect.currentframe())))
root_location = root_location + '/../'
save_path = root_location + 'pretrained/' + save_file
smiles_save_path = root_location + 'pretrained/' + smiles_save_file
settings = get_settings(molecules=molecules, grammar=grammar)
if EPOCHS is not None:
settings['EPOCHS'] = EPOCHS
if BATCH_SIZE is not None:
settings['BATCH_SIZE'] = BATCH_SIZE
save_dataset = IncrementingHDF5Dataset(smiles_save_path)
task = SequenceGenerationTask(molecules=molecules,
grammar=grammar,
reward_fun=reward_fun_on,
batch_size=BATCH_SIZE,
max_steps=max_steps,
save_dataset=save_dataset)
def get_model(sanity_checks=sanity_checks):
return get_decoder(
molecules,
grammar,
z_size=settings['z_size'],
decoder_hidden_n=200,
feature_len=None, # made redundant, need to factor it out
max_seq_length=max_steps,
drop_rate=drop_rate,
decoder_type=decoder_type,
task=task)[0]
model = get_model()
if preload_file is not None:
try:
preload_path = root_location + 'pretrained/' + preload_file
model.load_state_dict(torch.load(preload_path))
except:
pass
anchor_model = None
if anchor_file is not None:
anchor_model = get_model()
try:
anchor_path = root_location + 'pretrained/' + anchor_file
anchor_model.load_state_dict(torch.load(anchor_path))
except:
anchor_model = None
from generative_playground.molecules.rdkit_utils.rdkit_utils import NormalizedScorer
import rdkit.Chem.rdMolDescriptors as desc
import numpy as np
scorer = NormalizedScorer()
def model_process_fun(model_out, visdom, n):
from rdkit import Chem
from rdkit.Chem.Draw import MolToFile
actions, logits, rewards, terminals, info = model_out
smiles, valid = info
total_rewards = rewards.sum(1)
best_ind = torch.argmax(total_rewards).data.item()
this_smile = smiles[best_ind]
mol = Chem.MolFromSmiles(this_smile)
pic_save_path = root_location + 'images/' + 'test.svg'
if mol is not None:
try:
MolToFile(mol, pic_save_path, imageType='svg')
with open(pic_save_path, 'r') as myfile:
data = myfile.read()
data = data.replace('svg:', '')
visdom.append('best molecule of batch', 'svg', svgstr=data)
except:
pass
scores, norm_scores = scorer.get_scores([this_smile])
visdom.append(
'score component',
'line',
X=np.array([n]),
Y=np.array(
[[x for x in norm_scores[0]] + [norm_scores[0].sum()] +
[scores[0].sum()] + [desc.CalcNumAromaticRings(mol)]]),
opts={
'legend': [
'logP', 'SA', 'cycle', 'norm_reward', 'reward',
'Aromatic rings'
]
})
visdom.append('fraction valid',
'line',
X=np.array([n]),
Y=np.array([valid.mean().data.item()]))
if reward_fun_off is None:
reward_fun_off = reward_fun_on
def get_fitter(model,
loss_obj,
fit_plot_prefix='',
model_process_fun=None,
lr=None,
loss_display_cap=float('inf'),
anchor_model=None,
anchor_weight=0):
nice_params = filter(lambda p: p.requires_grad, model.parameters())
optimizer = optim.Adam(nice_params, lr=lr)
scheduler = lr_scheduler.StepLR(optimizer, step_size=100, gamma=0.99)
if dashboard is not None:
metric_monitor = MetricPlotter(
plot_prefix=fit_plot_prefix,
loss_display_cap=loss_display_cap,
dashboard_name=dashboard,
plot_ignore_initial=plot_ignore_initial,
process_model_fun=model_process_fun)
else:
metric_monitor = None
checkpointer = Checkpointer(valid_batches_to_checkpoint=1,
save_path=save_path,
save_always=True)
def my_gen():
for _ in range(1000):
yield to_gpu(torch.zeros(BATCH_SIZE, settings['z_size']))
fitter = fit_rl(train_gen=my_gen,
model=model,
optimizer=optimizer,
scheduler=scheduler,
epochs=EPOCHS,
loss_fn=loss_obj,
grad_clip=5,
anchor_model=anchor_model,
anchor_weight=anchor_weight,
metric_monitor=metric_monitor,
checkpointer=checkpointer)
return fitter
# the on-policy fitter
fitter1 = get_fitter(model,
PolicyGradientLoss(on_policy_loss_type),
plot_prefix + 'on-policy',
model_process_fun=model_process_fun,
lr=lr_on,
anchor_model=anchor_model,
anchor_weight=anchor_weight)
# get existing molecule data to add training
main_dataset = DatasetFromHDF5(settings['data_path'], 'actions')
# TODO change call to a simple DataLoader, no validation
train_loader, valid_loader = train_valid_loaders(main_dataset,
valid_fraction=0.1,
batch_size=BATCH_SIZE,
pin_memory=use_gpu)
fitter2 = get_fitter(model,
PolicyGradientLoss(off_policy_loss_type),
plot_prefix + ' off-policy',
lr=lr_off,
model_process_fun=model_process_fun,
loss_display_cap=125)
def on_policy_gen(fitter, model):
while True:
model.policy = SoftmaxRandomSamplePolicy()
yield next(fitter)
def off_policy_gen(fitter, data_gen, model):
while True:
data_iter = data_gen.__iter__()
try:
x_actions = next(data_iter).to(torch.int64)
model.policy = PolicyFromTarget(x_actions)
yield next(fitter)
except StopIteration:
data_iter = data_gen.__iter__()
return model, on_policy_gen(fitter1,
model), off_policy_gen(fitter2, train_loader,
model)
'''
if not len(smiles):
return -1 # an empty string is invalid for our purposes
atoms = num_aromatic_rings(smiles)
return [-1 if num is None else num + 0.5 for num in atoms]
batch_size = 40
drop_rate = 0.3
molecules = True
grammar = True
settings = get_settings(molecules, grammar)
max_steps = 277 #settings['max_seq_length']
invalid_value = -3.5
reward_fun = lambda x: 2.5 + NormalizedScorer(
settings['data_path'], invalid_value=invalid_value)(
x) #lambda x: reward_aromatic_rings(x)#
model, fitter1, fitter2 = train_policy_gradient(
molecules,
grammar,
EPOCHS=100,
BATCH_SIZE=batch_size,
reward_fun_on=reward_fun,
max_steps=max_steps,
lr=1e-4,
drop_rate=drop_rate,
decoder_type='attention',
plot_prefix='rings ',
dashboard='policy gradient',
save_file='policy_gradient_rings.h5',
def train_policy_gradient(molecules=True,
grammar=True,
EPOCHS=None,
BATCH_SIZE=None,
reward_fun_on=None,
reward_fun_off=None,
max_steps=277,
lr_on=2e-4,
lr_discrim=1e-4,
p_thresh=0.5,
drop_rate=0.0,
plot_ignore_initial=0,
randomize_reward=False,
save_file=None,
reward_sm=0.0,
preload_file=None,
anchor_file=None,
anchor_weight=0.0,
decoder_type='action',
plot_prefix='',
dashboard='policy gradient',
smiles_save_file=None,
on_policy_loss_type='best',
off_policy_loss_type='mean',
sanity_checks=True):
root_location = os.path.dirname(
os.path.abspath(inspect.getfile(inspect.currentframe())))
root_location = root_location + '/../../'
gen_save_path = root_location + 'pretrained/gen_' + save_file
disc_save_path = root_location + 'pretrained/disc_' + save_file
if smiles_save_file is not None:
smiles_save_path = root_location + 'pretrained/' + smiles_save_file
save_dataset = IncrementingHDF5Dataset(smiles_save_path)
else:
save_dataset = None
settings = get_settings(molecules=molecules, grammar=grammar)
codec = get_codec(molecules, grammar, settings['max_seq_length'])
discrim_model = GraphDiscriminator(codec.grammar, drop_rate=drop_rate)
zinc_data = get_zinc_smiles()
zinc_set = set(zinc_data)
lookbacks = [BATCH_SIZE, 10 * BATCH_SIZE, 100 * BATCH_SIZE]
history_data = [deque(['O'], maxlen=lb) for lb in lookbacks]
def originality_mult(smiles_list):
out = []
for s in smiles_list:
if s in zinc_set:
out.append(0.5)
elif s in history_data[0]:
out.append(0.5)
elif s in history_data[1]:
out.append(0.70)
elif s in history_data[2]:
out.append(0.85)
else:
out.append(1.0)
return np.array(out)
def sigmoid(x):
tmp = -x #(
return 1 / (1 + np.exp(-x))
def discriminator_reward_mult(smiles_list):
orig_state = discrim_model.training
discrim_model.eval()
discrim_out_logits = discrim_model(smiles_list)['p_zinc']
discrim_probs = F.softmax(discrim_out_logits, dim=1)
prob_zinc = discrim_probs[:, 1].detach().cpu().numpy()
if orig_state:
discrim_model.train()
return prob_zinc
def adj_reward(x):
p = discriminator_reward_mult(x)
reward = np.maximum(reward_fun_on(x), 0)
out = reward * originality_mult(x) + 2 * p
return out
def adj_reward_old(x):
p = discriminator_reward_mult(x)
w = sigmoid(-(p - p_thresh) / 0.01)
if randomize_reward:
rand = np.random.uniform(size=p.shape)
w *= rand
reward = np.maximum(reward_fun_on(x), p_thresh)
weighted_reward = w * p + (1 - w) * reward
out = weighted_reward * originality_mult(x) #
return out
if EPOCHS is not None:
settings['EPOCHS'] = EPOCHS
if BATCH_SIZE is not None:
settings['BATCH_SIZE'] = BATCH_SIZE
task = SequenceGenerationTask(molecules=molecules,
grammar=grammar,
reward_fun=adj_reward,
batch_size=BATCH_SIZE,
max_steps=max_steps,
save_dataset=save_dataset)
model = get_decoder(molecules,
grammar,
z_size=settings['z_size'],
decoder_hidden_n=200,
feature_len=codec.feature_len(),
max_seq_length=max_steps,
drop_rate=drop_rate,
decoder_type=decoder_type,
task=task)[0]
# TODO: really ugly, refactor! In fact this model doesn't need a MaskingHead at all!
model.stepper.model.mask_gen.priors = True #'conditional' # use empirical priors for the mask gen
# if preload_file is not None:
# try:
# preload_path = root_location + 'pretrained/' + preload_file
# model.load_state_dict(torch.load(preload_path))
# except:
# pass
anchor_model = None
from generative_playground.molecules.rdkit_utils.rdkit_utils import NormalizedScorer
import rdkit.Chem.rdMolDescriptors as desc
import numpy as np
scorer = NormalizedScorer()
def model_process_fun(model_out, visdom, n):
# TODO: rephrase this to return a dict, instead of calling visdom directly
from rdkit import Chem
from rdkit.Chem.Draw import MolToFile
# actions, logits, rewards, terminals, info = model_out
smiles, valid = model_out['info']
total_rewards = model_out['rewards'].sum(1)
best_ind = torch.argmax(total_rewards).data.item()
this_smile = smiles[best_ind]
mol = Chem.MolFromSmiles(this_smile)
pic_save_path = root_location + 'images/' + 'tmp.svg'
if mol is not None:
try:
MolToFile(mol, pic_save_path, imageType='svg')
with open(pic_save_path, 'r') as myfile:
data = myfile.read()
data = data.replace('svg:', '')
visdom.append('best molecule of batch', 'svg', svgstr=data)
except Exception as e:
print(e)
scores, norm_scores = scorer.get_scores([this_smile])
visdom.append(
'score component',
'line',
X=np.array([n]),
Y=np.array(
[[x for x in norm_scores[0]] + [norm_scores[0].sum()] +
[scores[0].sum()] + [desc.CalcNumAromaticRings(mol)] +
[total_rewards[best_ind].item()]]),
opts={
'legend': [
'logP', 'SA', 'cycle', 'norm_reward', 'reward',
'Aromatic rings', 'eff_reward'
]
})
visdom.append('fraction valid',
'line',
X=np.array([n]),
Y=np.array([valid.mean().data.item()]))
if reward_fun_off is None:
reward_fun_off = reward_fun_on
# construct the loader to feed the discriminator
def make_callback(data):
def hc(inputs, model, outputs, loss_fn, loss):
graphs = outputs['graphs']
smiles = [g.to_smiles() for g in graphs]
for s in smiles: # only store unique instances of molecules so discriminator can't guess on frequency
if s not in data:
data.append(s)
return hc
# need to have something there to begin with, else the DataLoader constructor barfs
def get_rl_fitter(model,
loss_obj,
train_gen,
save_path,
fit_plot_prefix='',
model_process_fun=None,
lr=None,
extra_callbacks=[],
loss_display_cap=float('inf'),
anchor_model=None,
anchor_weight=0):
nice_params = filter(lambda p: p.requires_grad, model.parameters())
optimizer = optim.Adam(nice_params, lr=lr)
scheduler = lr_scheduler.StepLR(optimizer, step_size=100, gamma=0.99)
if dashboard is not None:
metric_monitor = MetricPlotter(
plot_prefix=fit_plot_prefix,
loss_display_cap=loss_display_cap,
dashboard_name=dashboard,
plot_ignore_initial=plot_ignore_initial,
process_model_fun=model_process_fun,
smooth_weight=0.9)
else:
metric_monitor = None
checkpointer = Checkpointer(valid_batches_to_checkpoint=1,
save_path=save_path,
save_always=True)
fitter = fit_rl(train_gen=train_gen,
model=model,
optimizer=optimizer,
scheduler=scheduler,
epochs=EPOCHS,
loss_fn=loss_obj,
grad_clip=5,
anchor_model=anchor_model,
anchor_weight=anchor_weight,
callbacks=[metric_monitor, checkpointer] +
extra_callbacks)
return fitter
class GeneratorToIterable:
def __init__(self, gen):
self.gen = gen
# we assume the generator is finite
self.len = 0
for _ in gen():
self.len += 1
def __len__(self):
return self.len
def __iter__(self):
return self.gen()
def my_gen():
for _ in range(1000):
yield to_gpu(torch.zeros(BATCH_SIZE, settings['z_size']))
# the on-policy fitter
history_callbacks = [make_callback(d) for d in history_data]
fitter1 = get_rl_fitter(model,
PolicyGradientLoss(on_policy_loss_type,
last_reward_wgt=reward_sm),
GeneratorToIterable(my_gen),
gen_save_path,
plot_prefix + 'on-policy',
model_process_fun=model_process_fun,
lr=lr_on,
extra_callbacks=history_callbacks,
anchor_model=anchor_model,
anchor_weight=anchor_weight)
#
# # get existing molecule data to add training
pre_dataset = EvenlyBlendedDataset(
2 * [history_data[0]] + history_data[1:],
labels=False) # a blend of 3 time horizons
dataset = EvenlyBlendedDataset([pre_dataset, zinc_data], labels=True)
discrim_loader = DataLoader(dataset, shuffle=True, batch_size=50)
celoss = nn.CrossEntropyLoss()
def my_loss(x):
# tmp = discriminator_reward_mult(x['smiles'])
# tmp2 = F.softmax(x['p_zinc'], dim=1)[:,1].detach().cpu().numpy()
# import numpy as np
# assert np.max(np.abs(tmp-tmp2)) < 1e-6
return celoss(x['p_zinc'].to(device), x['dataset_index'].to(device))
fitter2 = get_rl_fitter(discrim_model,
my_loss,
IterableTransform(
discrim_loader, lambda x: {
'smiles': x['X'],
'dataset_index': x['dataset_index']
}),
disc_save_path,
plot_prefix + ' discriminator',
lr=lr_discrim,
model_process_fun=None)
def on_policy_gen(fitter, model):
while True:
model.policy = SoftmaxRandomSamplePolicy(
) #bias=codec.grammar.get_log_frequencies())
yield next(fitter)
return model, on_policy_gen(fitter1, model), fitter2
