BranchSBM / train.py
Sophia Tang
Initial commit
b55bace
Raw
History Blame Contribute Delete
18.9 kB
import sys
import os
import argparse
import copy
import time
import json
import torch.nn as nn
import wandb
from pytorch_lightning import Trainer
from pytorch_lightning.loggers import WandbLogger
from torchcfm.optimal_transport import OTPlanSampler
from parsers import parse_args
from train_utils import load_config, merge_config, generate_group_string, dataset_name2datapath, create_callbacks
from src.branchsbm import BranchSBM
from src.branch_flow_net_train import FlowNetTrainCell, FlowNetTrainLidar
from src.branch_flow_net_test import (
FlowNetTestLidar, FlowNetTestMouse, FlowNetTestClonidine, FlowNetTestTrametinib, FlowNetTestVeres
)
from src.branch_interpolant_train import BranchInterpolantTrain
from src.branch_growth_net_train import GrowthNetTrain, GrowthNetTrainCell, GrowthNetTrainLidar, SequentialGrowthNetTrain
from src.networks.flow_mlp import VelocityNet
from src.networks.growth_mlp import GrowthNet
from src.networks.interpolant_mlp import GeoPathMLP
from src.utils import set_seed
from src.ema import EMA
from src.geo_metrics.metric_factory import DataManifoldMetric
from dataloaders.mouse_data import WeightedBranchedCellDataModule, SingleBranchCellDataModule
from dataloaders.three_branch_data import ThreeBranchTahoeDataModule
from dataloaders.clonidine_v2_data import ClonidineV2DataModule
from dataloaders.clonidine_single_branch import ClonidineSingleBranchDataModule
from dataloaders.trametinib_single import TrametinibSingleBranchDataModule
from dataloaders.lidar_data import WeightedBranchedLidarDataModule
from dataloaders.lidar_data_single import LidarSingleDataModule
from dataloaders.veres_leiden_data import WeightedBranchedVeresDataModule
def main(args: argparse.Namespace, seed: int, t_exclude: int) -> None:
set_seed(seed)
branches = args.branches
skipped_time_points = [t_exclude] if t_exclude else []
print("config path:")
print(args.config_path)
print("whiten")
print(args.whiten)
# Add date and time prefix to run name for distinguishable results
current_datetime = time.strftime("%m_%d_%H%M", time.localtime())
run_name_with_datetime = f"{current_datetime}_{args.run_name}"
# Update args.run_name so test classes use the dated name
args.run_name = run_name_with_datetime
### DATAMODULES
### DATAMODULES ###
if args.data_name == "lidar":
datamodule = WeightedBranchedLidarDataModule(args=args)
elif args.data_name == "lidarsingle":
datamodule = LidarSingleDataModule(args=args)
elif args.data_name == "mouse":
datamodule = WeightedBranchedCellDataModule(args=args)
elif args.data_name == "mousesingle":
datamodule = SingleBranchCellDataModule(args=args)
elif args.data_name in ["clonidine50D", "clonidine100D", "clonidine150D"]:
datamodule = ClonidineV2DataModule(args=args)
elif args.data_name == "clonidine50Dsingle":
datamodule = ClonidineSingleBranchDataModule(args=args)
elif args.data_name == "trametinib":
datamodule = ThreeBranchTahoeDataModule(args=args)
elif args.data_name == "trametinibsingle":
datamodule = TrametinibSingleBranchDataModule(args=args)
elif args.data_name == "veres":
datamodule = WeightedBranchedVeresDataModule(args=args)
branches = datamodule.num_branches
print("number of branches:", branches)
flow_nets = nn.ModuleList()
geopath_nets = nn.ModuleList()
growth_nets = nn.ModuleList()
##### initialize branched flow and growth networks #####
for i in range(branches):
flow_net = VelocityNet(
dim=args.dim,
hidden_dims=args.hidden_dims_flow,
activation=args.activation_flow,
batch_norm=False,
)
geopath_net = GeoPathMLP(
input_dim=args.dim,
hidden_dims=args.hidden_dims_geopath,
time_geopath=args.time_geopath,
activation=args.activation_geopath,
batch_norm=False,
)
if i == 0:
growth_net = GrowthNet(
dim=args.dim,
hidden_dims=args.hidden_dims_growth,
activation=args.activation_growth,
batch_norm=False,
negative=True
)
else:
growth_net = GrowthNet(
dim=args.dim,
hidden_dims=args.hidden_dims_growth,
activation=args.activation_growth,
batch_norm=False,
negative=False
)
if args.ema_decay is not None:
flow_net = EMA(model=flow_net, decay=args.ema_decay)
geopath_net = EMA(model=geopath_net, decay=args.ema_decay)
growth_net = EMA(model=growth_net, decay=args.ema_decay)
flow_nets.append(flow_net)
geopath_nets.append(geopath_net)
growth_nets.append(growth_net)
ot_sampler = (
OTPlanSampler(method=args.optimal_transport_method)
if args.optimal_transport_method != "None"
else None
)
wandb.init(
project="branchsbm",
name=run_name_with_datetime,
config=vars(args),
dir=args.working_dir,
)
flow_matcher_base = BranchSBM(
geopath_nets=geopath_nets,
sigma=args.sigma,
alpha=int(args.branchsbm),
)
##### STAGE 1: Training of Geodesic Interpolants Beginning #####
geopath_callbacks = create_callbacks(
args, phase="geopath", data_type=args.data_type, run_id=wandb.run.id
)
# define state cost
data_manifold_metric = DataManifoldMetric(
args=args,
skipped_time_points=skipped_time_points,
datamodule=datamodule,
)
geopath_model = BranchInterpolantTrain(
flow_matcher=flow_matcher_base,
skipped_time_points=skipped_time_points,
ot_sampler=ot_sampler,
args=args,
data_manifold_metric=data_manifold_metric
)
wandb_logger = WandbLogger(version=run_name_with_datetime)
trainer = Trainer(
max_epochs=args.epochs,
callbacks=geopath_callbacks,
accelerator=args.accelerator,
logger=wandb_logger,
num_sanity_val_steps=0,
default_root_dir=args.working_dir,
gradient_clip_val=(1.0 if args.data_type == "image" else None),
)
if args.load_geopath_model_ckpt:
best_model_path = args.load_geopath_model_ckpt
else:
trainer.fit(
geopath_model,
datamodule=datamodule,
)
best_model_path = geopath_callbacks[0].best_model_path
geopath_model = BranchInterpolantTrain.load_from_checkpoint(best_model_path)
flow_matcher_base.geopath_nets = geopath_model.geopath_nets
##### STAGE 1: Training of Geodesic Interpolants End #####
##### STAGE 2: Flow Matching Beginning #####
flow_callbacks = create_callbacks(
args,
phase="flow",
data_type=args.data_type,
run_id=wandb.run.id,
datamodule=datamodule,
)
if args.data_type == "lidar":
FlowNetTrain = FlowNetTrainLidar
else:
FlowNetTrain = FlowNetTrainCell
flow_train = FlowNetTrain(
flow_matcher=flow_matcher_base,
flow_nets=flow_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
)
# Reuse existing wandb run from Stage 1
wandb_logger = WandbLogger(version=run_name_with_datetime)
trainer = Trainer(
max_epochs=args.epochs,
callbacks=flow_callbacks,
check_val_every_n_epoch=args.check_val_every_n_epoch,
accelerator=args.accelerator,
logger=wandb_logger,
default_root_dir=args.working_dir,
gradient_clip_val=(1.0 if args.data_type == "image" else None),
num_sanity_val_steps=(0 if args.data_type == "image" else None),
)
trainer.fit(
flow_train, datamodule=datamodule, ckpt_path=args.resume_flow_model_ckpt
)
if args.data_type == "lidar":
trainer.test(flow_train, datamodule=datamodule)
##### STAGE 2: Flow Matching End #####
##### STAGE 3: Training Growth Networks Beginning ####
flow_nets = flow_train.flow_nets
growth_callbacks = create_callbacks(
args,
phase="growth",
data_type=args.data_type,
run_id=wandb.run.id,
datamodule=datamodule,
)
if args.data_type == "lidar":
GrowthNetTrainClass = GrowthNetTrainLidar
else:
GrowthNetTrainClass = GrowthNetTrainCell
growth_train = GrowthNetTrainClass(
flow_nets = flow_nets,
growth_nets = growth_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
data_manifold_metric=data_manifold_metric,
joint = False
)
# Reuse existing wandb run
wandb_logger = WandbLogger(version=run_name_with_datetime)
trainer = Trainer(
max_epochs=args.epochs,
callbacks=growth_callbacks,
check_val_every_n_epoch=args.check_val_every_n_epoch,
accelerator=args.accelerator,
logger=wandb_logger,
default_root_dir=args.working_dir,
gradient_clip_val=(1.0 if args.data_type == "image" else None),
num_sanity_val_steps=(0 if args.data_type == "image" else None),
)
trainer.fit(
growth_train, datamodule=datamodule, ckpt_path=None
)
# Load best checkpoint for testing
best_growth_path = growth_callbacks[0].best_model_path
if best_growth_path:
print(f"Loading best growth model from: {best_growth_path}")
if args.sequential:
growth_train = SequentialGrowthNetTrain.load_from_checkpoint(
best_growth_path,
flow_nets=flow_nets,
growth_nets=growth_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
data_manifold_metric=data_manifold_metric,
joint=False
)
else:
growth_train = GrowthNetTrainClass.load_from_checkpoint(
best_growth_path,
flow_nets=flow_nets,
growth_nets=growth_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
data_manifold_metric=data_manifold_metric,
joint=False
)
# Extract the trained flow_nets from the loaded checkpoint
flow_nets = growth_train.flow_nets
# Ensure flow_nets and growth_nets are ModuleList (not tuple)
if isinstance(flow_nets, tuple):
flow_nets = nn.ModuleList(flow_nets)
if isinstance(growth_nets, tuple):
growth_nets = nn.ModuleList(growth_nets)
# Use appropriate test class based on data type
if "lidar" in args.data_name.lower():
test_model = FlowNetTestLidar(
flow_nets = flow_nets,
growth_nets = growth_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
data_manifold_metric=data_manifold_metric,
joint = False
)
elif "mouse" in args.data_name.lower():
test_model = FlowNetTestMouse(
flow_nets = flow_nets,
growth_nets = growth_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
data_manifold_metric=data_manifold_metric,
joint = False
)
elif "clonidine" in args.data_name.lower():
test_model = FlowNetTestClonidine(
flow_matcher=flow_matcher_base,
flow_nets=flow_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
)
elif "trametinib" in args.data_name.lower():
test_model = FlowNetTestTrametinib(
flow_matcher=flow_matcher_base,
flow_nets=flow_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
)
elif "veres" in args.data_name.lower():
test_model = FlowNetTestVeres(
flow_nets = flow_nets,
growth_nets = growth_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
data_manifold_metric=data_manifold_metric,
joint = False
)
else:
# Default to growth_train test
test_model = growth_train
trainer.test(test_model, datamodule=datamodule)
##### STAGE 3: Training Growth Networks End ####
##### STAGE 4: Joint Training Beginning ####
growth_nets = growth_train.growth_nets
joint_callbacks = create_callbacks(
args,
phase="joint",
data_type=args.data_type,
run_id=wandb.run.id,
datamodule=datamodule,
)
if args.sequential:
joint_train = SequentialGrowthNetTrain(
flow_nets = flow_nets,
growth_nets = growth_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
data_manifold_metric=data_manifold_metric,
joint = True
)
else:
if args.data_type == "lidar":
GrowthNetTrainClass = GrowthNetTrainLidar
else:
GrowthNetTrainClass = GrowthNetTrainCell
joint_train = GrowthNetTrainClass(
flow_nets = flow_nets,
growth_nets = growth_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
data_manifold_metric=data_manifold_metric,
joint = True
)
# Reuse existing wandb run
wandb_logger = WandbLogger(version=run_name_with_datetime)
trainer = Trainer(
max_epochs=args.epochs,
callbacks=joint_callbacks,
check_val_every_n_epoch=args.check_val_every_n_epoch,
accelerator=args.accelerator,
logger=wandb_logger,
default_root_dir=args.working_dir,
gradient_clip_val=(1.0 if args.data_type == "image" else None),
num_sanity_val_steps=(0 if args.data_type == "image" else None),
)
trainer.fit(
joint_train, datamodule=datamodule, ckpt_path=None
)
# Load best checkpoint for testing
best_joint_path = joint_callbacks[0].best_model_path
if best_joint_path:
print(f"Loading best joint model from: {best_joint_path}")
if args.sequential:
joint_train = SequentialGrowthNetTrain.load_from_checkpoint(
best_joint_path,
flow_nets=flow_nets,
growth_nets=growth_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
data_manifold_metric=data_manifold_metric,
joint=True
)
else:
joint_train = GrowthNetTrainClass.load_from_checkpoint(
best_joint_path,
flow_nets=flow_nets,
growth_nets=growth_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
data_manifold_metric=data_manifold_metric,
joint=True
)
# Extract the trained flow_nets and growth_nets from the loaded checkpoint
flow_nets = joint_train.flow_nets
growth_nets = joint_train.growth_nets
# Ensure flow_nets and growth_nets are ModuleList (not tuple)
if isinstance(flow_nets, tuple):
flow_nets = nn.ModuleList(flow_nets)
if isinstance(growth_nets, tuple):
growth_nets = nn.ModuleList(growth_nets)
# Use appropriate test class based on data type
if "lidar" in args.data_name.lower():
test_model = FlowNetTestLidar(
flow_nets = flow_nets,
growth_nets = growth_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
data_manifold_metric=data_manifold_metric,
joint = True
)
elif "mouse" in args.data_name.lower():
test_model = FlowNetTestMouse(
flow_nets = flow_nets,
growth_nets = growth_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
data_manifold_metric=data_manifold_metric,
joint = True
)
elif "clonidine" in args.data_name.lower():
test_model = FlowNetTestClonidine(
flow_matcher=flow_matcher_base,
flow_nets=flow_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
)
elif "trametinib" in args.data_name.lower():
test_model = FlowNetTestTrametinib(
flow_matcher=flow_matcher_base,
flow_nets=flow_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
)
elif "veres" in args.data_name.lower():
test_model = FlowNetTestVeres(
flow_nets = flow_nets,
growth_nets = growth_nets,
ot_sampler=ot_sampler,
skipped_time_points=skipped_time_points,
args=args,
data_manifold_metric=data_manifold_metric,
joint = True
)
else:
test_model = joint_train
test_model = joint_train
trainer.test(test_model, datamodule=datamodule)
##### STAGE 4: Joint Training End ####
wandb.finish()
if __name__ == "__main__":
args = parse_args()
updated_args = copy.deepcopy(args)
if args.config_path:
config = load_config(args.config_path)
updated_args = merge_config(updated_args, config)
updated_args.group_name = generate_group_string()
updated_args.data_path = dataset_name2datapath(
updated_args.data_name, updated_args.working_dir
)
for seed in updated_args.seeds:
if updated_args.t_exclude:
for i, t_exclude in enumerate(updated_args.t_exclude):
updated_args.t_exclude_current = t_exclude
updated_args.seed_current = seed
updated_args.gamma_current = updated_args.gammas[i]
main(updated_args, seed=seed, t_exclude=t_exclude)
else:
updated_args.seed_current = seed
updated_args.gamma_current = updated_args.gammas[0]
main(updated_args, seed=seed, t_exclude=None)