import copy import functools import os import time from types import SimpleNamespace import numpy as np import blobfile as bf import torch from torch.optim import AdamW from diffusion import logger from utils import dist_util from diffusion.fp16_util import MixedPrecisionTrainer from diffusion.resample import LossAwareSampler, UniformSampler from tqdm import tqdm from diffusion.resample import create_named_schedule_sampler from data_loaders.humanml.networks.evaluator_wrapper import EvaluatorMDMWrapper from eval import eval_humanml, eval_humanact12_uestc from data_loaders.get_data import get_dataset_loader # For ImageNet experiments, this was a good default value. # We found that the lg_loss_scale quickly climbed to # 20-21 within the first ~1K steps of training. ## intial log loss scale ## INITIAL_LOG_LOSS_SCALE = 20.0 # do not # class TrainLoop: def __init__(self, args, train_platform, model, diffusion, data): self.args = args self.dataset = args.dataset self.train_platform = train_platform self.model = model self.diffusion = diffusion self.cond_mode = model.cond_mode self.data = data self.batch_size = args.batch_size self.microbatch = args.batch_size # deprecating this option self.lr = args.lr self.log_interval = args.log_interval self.save_interval = args.save_interval self.resume_checkpoint = args.resume_checkpoint self.use_fp16 = False # deprecating this option self.fp16_scale_growth = 1e-3 # deprecating this option self.weight_decay = args.weight_decay self.lr_anneal_steps = args.lr_anneal_steps self.step = 0 self.resume_step = 0 self.global_batch = self.batch_size # * dist.get_world_size() self.num_steps = args.num_steps self.num_epochs = self.num_steps // len(self.data) + 1 self.sync_cuda = torch.cuda.is_available() self._load_and_sync_parameters() self.mp_trainer = MixedPrecisionTrainer( model=self.model, # use_fp16=self.use_fp16, fp16_scale_growth=self.fp16_scale_growth, ) self.save_dir = args.save_dir self.overwrite = args.overwrite self.opt = AdamW( self.mp_trainer.master_params, lr=self.lr, weight_decay=self.weight_decay ) if self.resume_step: self._load_optimizer_state() # Model was resumed, either due to a restart or a checkpoint # being specified at the command line. self.device = torch.device("cpu") if torch.cuda.is_available() and dist_util.dev() != 'cpu': self.device = torch.device(dist_util.dev()) self.schedule_sampler_type = 'uniform' self.schedule_sampler = create_named_schedule_sampler(self.schedule_sampler_type, diffusion) self.eval_wrapper, self.eval_data, self.eval_gt_data = None, None, None if args.dataset in ['kit', 'humanml', 'motion_ours'] and args.eval_during_training: mm_num_samples = 0 # mm is super slow hence we won't run it during training mm_num_repeats = 0 # mm is super slow hence we won't run it during training gen_loader = get_dataset_loader(name=args.dataset, batch_size=args.eval_batch_size, num_frames=None, split=args.eval_split, hml_mode='eval') self.eval_gt_data = get_dataset_loader(name=args.dataset, batch_size=args.eval_batch_size, num_frames=None, split=args.eval_split, hml_mode='gt') self.eval_wrapper = EvaluatorMDMWrapper(args.dataset, dist_util.dev()) self.eval_data = { 'test': lambda: eval_humanml.get_mdm_loader( model, diffusion, args.eval_batch_size, gen_loader, mm_num_samples, mm_num_repeats, gen_loader.dataset.opt.max_motion_length, args.eval_num_samples, scale=1., ) } self.use_ddp = False self.ddp_model = self.model def _load_and_sync_parameters(self): resume_checkpoint = find_resume_checkpoint() or self.resume_checkpoint if resume_checkpoint: self.resume_step = parse_resume_step_from_filename(resume_checkpoint) logger.log(f"loading model from checkpoint: {resume_checkpoint}...") self.model.load_state_dict( dist_util.load_state_dict( resume_checkpoint, map_location=dist_util.dev() ) ) def _load_optimizer_state(self): main_checkpoint = find_resume_checkpoint() or self.resume_checkpoint opt_checkpoint = bf.join( bf.dirname(main_checkpoint), f"opt{self.resume_step:09}.pt" ) if bf.exists(opt_checkpoint): logger.log(f"loading optimizer state from checkpoint: {opt_checkpoint}") state_dict = dist_util.load_state_dict( opt_checkpoint, map_location=dist_util.dev() ) self.opt.load_state_dict(state_dict) def run_loop(self): for epoch in range(self.num_epochs): print(f'Starting epoch {epoch}') for motion, cond in tqdm(self.data): ## motion; cond; data ## if not (not self.lr_anneal_steps or self.step + self.resume_step < self.lr_anneal_steps): break # print(f"motion: {motion.size()}, ") ## motion.to(self.device) motion = motion.to(self.device) cond['y'] = {key: val.to(self.device) if torch.is_tensor(val) else val for key, val in cond['y'].items()} ## run current motion and cond ## ## run step ##s self.run_step(motion, cond) ## run step for the motion and cond ## if self.step % self.log_interval == 0: for k,v in logger.get_current().name2val.items(): if k == 'loss': print('step[{}]: loss[{:0.5f}]'.format(self.step+self.resume_step, v)) if k in ['step', 'samples'] or '_q' in k: continue else: self.train_platform.report_scalar(name=k, value=v, iteration=self.step, group_name='Loss') if self.step % self.save_interval == 0: ## save; model.eval; self.save() self.model.eval() self.evaluate() self.model.train() # Run for a finite amount of time in integration tests. if os.environ.get("DIFFUSION_TRAINING_TEST", "") and self.step > 0: return self.step += 1 if not (not self.lr_anneal_steps or self.step + self.resume_step < self.lr_anneal_steps): break # Save the last checkpoint if it wasn't already saved. if (self.step - 1) % self.save_interval != 0: self.save() self.evaluate() def evaluate(self): if not self.args.eval_during_training: return start_eval = time.time() if self.eval_wrapper is not None: print('Running evaluation loop: [Should take about 90 min]') log_file = os.path.join(self.save_dir, f'eval_humanml_{(self.step + self.resume_step):09d}.log') diversity_times = 300 mm_num_times = 0 # mm is super slow hence we won't run it during training eval_dict = eval_humanml.evaluation( self.eval_wrapper, self.eval_gt_data, self.eval_data, log_file, replication_times=self.args.eval_rep_times, diversity_times=diversity_times, mm_num_times=mm_num_times, run_mm=False) print(eval_dict) for k, v in eval_dict.items(): if k.startswith('R_precision'): for i in range(len(v)): self.train_platform.report_scalar(name=f'top{i + 1}_' + k, value=v[i], iteration=self.step + self.resume_step, group_name='Eval') else: self.train_platform.report_scalar(name=k, value=v, iteration=self.step + self.resume_step, group_name='Eval') elif self.dataset in ['humanact12', 'uestc']: eval_args = SimpleNamespace(num_seeds=self.args.eval_rep_times, num_samples=self.args.eval_num_samples, batch_size=self.args.eval_batch_size, device=self.device, guidance_param = 1, dataset=self.dataset, unconstrained=self.args.unconstrained, model_path=os.path.join(self.save_dir, self.ckpt_file_name())) eval_dict = eval_humanact12_uestc.evaluate(eval_args, model=self.model, diffusion=self.diffusion, data=self.data.dataset) print(f'Evaluation results on {self.dataset}: {sorted(eval_dict["feats"].items())}') for k, v in eval_dict["feats"].items(): if 'unconstrained' not in k: self.train_platform.report_scalar(name=k, value=np.array(v).astype(float).mean(), iteration=self.step, group_name='Eval') else: self.train_platform.report_scalar(name=k, value=np.array(v).astype(float).mean(), iteration=self.step, group_name='Eval Unconstrained') end_eval = time.time() print(f'Evaluation time: {round(end_eval-start_eval)/60}min') def run_step(self, batch, cond): self.forward_backward(batch, cond) ## forward self.mp_trainer.optimize(self.opt) self._anneal_lr() self.log_step() def forward_backward(self, batch, cond): self.mp_trainer.zero_grad() for i in range(0, batch.shape[0], self.microbatch): # Eliminates the microbatch feature assert i == 0 assert self.microbatch == self.batch_size micro = batch micro_cond = cond ## micro-batch last_batch = (i + self.microbatch) >= batch.shape[0] t, weights = self.schedule_sampler.sample(micro.shape[0], dist_util.dev()) compute_losses = functools.partial( self.diffusion.training_losses, self.ddp_model, micro, # [bs, ch, image_size, image_size] t, # [bs](int) sampled timesteps model_kwargs=micro_cond, dataset=self.data.dataset ) if last_batch or not self.use_ddp: losses = compute_losses() ## compute lossses else: with self.ddp_model.no_sync(): losses = compute_losses() if isinstance(self.schedule_sampler, LossAwareSampler): self.schedule_sampler.update_with_local_losses( t, losses["loss"].detach() ) loss = (losses["loss"] * weights).mean() log_loss_dict( self.diffusion, t, {k: v * weights for k, v in losses.items()} ) self.mp_trainer.backward(loss) def predict_single_step(self, batch, cond, use_t=None): # self.mp_trainer.zero_grad() tot_samples = [] tot_targets = [] for i in range(0, batch.shape[0], self.microbatch): # Eliminates the microbatch feature assert i == 0 assert self.microbatch == self.batch_size micro = batch micro_cond = cond ## micro-batch ## last_batch = (i + self.microbatch) >= batch.shape[0] t, weights = self.schedule_sampler.sample(micro.shape[0], dist_util.dev()) if use_t is not None: t = torch.zeros_like(t) + use_t # batch: bsz x nnjoints x 3 x nnframes # ## === original sampling === ## # terms, model_output, target, t = self.diffusion.predict_sample_single_step(self.ddp_model, micro, t, model_kwargs=micro_cond, noise=None, dataset=self.data.dataset) ### use p_sample_loop from the diffusion model ### sample_fn = self.diffusion.p_sample_loop samples = sample_fn( self.ddp_model, (micro.size(0), micro.size(1), micro.size(2), micro.size(3)), clip_denoised=False, model_kwargs=cond, skip_timesteps=0, init_image=micro, # init_image=None, progress=True, dump_steps=None, noise=None, ## noise ## # const_noise=False, # whether to cond on noise ## st_timestep=use_t, ) tot_samples.append(samples) tot_targets.append(micro) model_output = torch.cat(tot_samples, dim=0) target = torch.cat(tot_targets, dim=0) print(f"Returning with model_output; {model_output.size()}, target: {target.size()}") ### returning the samples and tarets ### return model_output, target def predict_from_data(self): # for epoch in range(self.num_epochs): # # print(f'Starting epoch {epoch}') # the ## ==== a single pass for a single sequence ==== ## tot_model_outputs = [] tot_targets = [] tot_st_idxes = [] tot_ed_idxes = [] tot_pert_verts = [] tot_verts = [] ## motion; cond; data ## for motion, cond in tqdm(self.data): # if not (not self.lr_anneal_steps or self.step + self.resume_step < self.lr_anneal_steps): # break # tingyouyiside # print(f"motion: {motion.size()}, ") ## motion.to(self.device) motion = motion.to(self.device) cond['y'] = {key: val.to(self.device) if torch.is_tensor(val) else val for key, val in cond['y'].items()} st_idxes = cond['y']['st_idx'] # st_idxes ed_idxes = cond['y']['ed_idx'] # ed_idxes pert_verts = cond['y']['pert_verts'] verts = cond['y']['verts'] if 'avg_joints' in cond['y']: avg_joints = cond['y']['avg_joints'] std_joints = cond['y']['std_joints'] else: avg_joints = None std_joints = None tot_pert_verts.append(pert_verts) tot_verts.append(verts) ## generative denoising -> we want to use it for the denoising task ## # std_joints: bsz x 1 # avg_joints: bsz x 1 x 3 --> mean of joints for each batch ## predict_single_step ## use_t = 100 model_output, target = self.predict_single_step(motion, cond, use_t=use_t) ## model_output: ([6, 21, 3, 60]), target: torch.Size([6, 21, 3, 60]) if avg_joints is not None: # ### model_output, target ### # model_output = (model_output * std_joints.unsqueeze(-1).unsqueeze(-1)) + avg_joints.unsqueeze(-1) # target = (target * std_joints.unsqueeze(-1).unsqueeze(-1)) + avg_joints.unsqueeze(-1) ## std_joints; ## ### model_output, target ### print(f"std_joints: {std_joints.size()}, avg_joints: {avg_joints.size()}") model_output = (model_output * std_joints.permute(0, 2, 3, 1)) + avg_joints.permute(0, 2, 3, 1) target = (target * std_joints.permute(0, 2, 3, 1)) + avg_joints.permute(0, 2, 3, 1) tot_st_idxes.append(st_idxes) tot_ed_idxes.append(ed_idxes) tot_targets.append(target) tot_model_outputs.append(model_output) tot_st_idxes = torch.cat(tot_st_idxes, dim=0) tot_ed_idxes = torch.cat(tot_ed_idxes, dim=0) tot_targets = torch.cat(tot_targets, dim=0) tot_model_outputs = torch.cat(tot_model_outputs, dim=0) tot_pert_verts = torch.cat(tot_pert_verts, dim=0) tot_verts = torch.cat(tot_verts, dim=0) return tot_targets, tot_model_outputs, tot_st_idxes, tot_ed_idxes, tot_pert_verts, tot_verts # ## run current motion and cond # self.run_step(motion, cond) # if self.step % self.log_interval == 0: # for k,v in logger.get_current().name2val.items(): # if k == 'loss': # print('step[{}]: loss[{:0.5f}]'.format(self.step+self.resume_step, v)) # if k in ['step', 'samples'] or '_q' in k: # continue # else: # self.train_platform.report_scalar(name=k, value=v, iteration=self.step, group_name='Loss') # if self.step % self.save_interval == 0: # ## save; model.eval; # self.save() # self.model.eval() # self.evaluate() # self.model.train() # # Run for a finite amount of time in integration tests. # if os.environ.get("DIFFUSION_TRAINING_TEST", "") and self.step > 0: # return # self.step += 1 # if not (not self.lr_anneal_steps or self.step + self.resume_step < self.lr_anneal_steps): # break # ## predict single ## manifold of valid hand trajectories ## # Save the last checkpoint if it wasn't already saved. # if (self.step - 1) % self.save_interval != 0: # self.save() # self.evaluate() def _anneal_lr(self): if not self.lr_anneal_steps: return frac_done = (self.step + self.resume_step) / self.lr_anneal_steps lr = self.lr * (1 - frac_done) for param_group in self.opt.param_groups: param_group["lr"] = lr def log_step(self): logger.logkv("step", self.step + self.resume_step) logger.logkv("samples", (self.step + self.resume_step + 1) * self.global_batch) def ckpt_file_name(self): return f"model{(self.step+self.resume_step):09d}.pt" def save(self): def save_checkpoint(params): state_dict = self.mp_trainer.master_params_to_state_dict(params) # Do not save CLIP weights clip_weights = [e for e in state_dict.keys() if e.startswith('clip_model.')] for e in clip_weights: del state_dict[e] logger.log(f"saving model...") filename = self.ckpt_file_name() with bf.BlobFile(bf.join(self.save_dir, filename), "wb") as f: torch.save(state_dict, f) save_checkpoint(self.mp_trainer.master_params) with bf.BlobFile( bf.join(self.save_dir, f"opt{(self.step+self.resume_step):09d}.pt"), "wb", ) as f: torch.save(self.opt.state_dict(), f) def parse_resume_step_from_filename(filename): """ Parse filenames of the form path/to/modelNNNNNN.pt, where NNNNNN is the checkpoint's number of steps. """ split = filename.split("model") if len(split) < 2: return 0 split1 = split[-1].split(".")[0] try: return int(split1) except ValueError: return 0 def get_blob_logdir(): # You can change this to be a separate path to save checkpoints to # a blobstore or some external drive. return logger.get_dir() def find_resume_checkpoint(): # On your infrastructure, you may want to override this to automatically # discover the latest checkpoint on your blob storage, etc. return None def log_loss_dict(diffusion, ts, losses): for key, values in losses.items(): logger.logkv_mean(key, values.mean().item()) # Log the quantiles (four quartiles, in particular). for sub_t, sub_loss in zip(ts.cpu().numpy(), values.detach().cpu().numpy()): quartile = int(4 * sub_t / diffusion.num_timesteps) logger.logkv_mean(f"{key}_q{quartile}", sub_loss)