Update

6 files changed, 104 insertions, 73 deletions

diff --git a/train_dreambooth.py b/train_dreambooth.py
index d722e68..48bdcf8 100644
--- a/train_dreambooth.py
+++ b/train_dreambooth.py
@@ -14,8 +14,7 @@ from slugify import slugify
 
 from util import load_config, load_embeddings_from_dir
 from data.csv import VlpnDataModule, keyword_filter
-from training.functional import train, generate_class_images, add_placeholder_tokens, get_models
-from training.strategy.ti import textual_inversion_strategy
+from training.functional import train, get_models
 from training.strategy.dreambooth import dreambooth_strategy
 from training.optimization import get_scheduler
 from training.util import save_args
@@ -610,7 +609,7 @@ def main():
     )
 
     trainer(
-        callbacks_fn=dreambooth_strategy,
+        strategy=dreambooth_strategy,
         project="dreambooth",
         train_dataloader=datamodule.train_dataloader,
         val_dataloader=datamodule.val_dataloader,
diff --git a/train_ti.py b/train_ti.py
index e7aeb23..0891c49 100644
--- a/train_ti.py
+++ b/train_ti.py
@@ -14,7 +14,7 @@ from slugify import slugify
 
 from util import load_config, load_embeddings_from_dir
 from data.csv import VlpnDataModule, keyword_filter
-from training.functional import train, generate_class_images, add_placeholder_tokens, get_models
+from training.functional import train, add_placeholder_tokens, get_models
 from training.strategy.ti import textual_inversion_strategy
 from training.optimization import get_scheduler
 from training.util import save_args
@@ -79,6 +79,10 @@ def parse_args():
         help="Number of vectors per embedding."
     )
     parser.add_argument(
+        "--simultaneous",
+        action="store_true",
+    )
+    parser.add_argument(
         "--num_class_images",
         type=int,
         default=0,
@@ -474,11 +478,12 @@ def parse_args():
     if len(args.placeholder_tokens) != len(args.num_vectors):
         raise ValueError("--placeholder_tokens and --num_vectors must have the same number of items")
 
-    if isinstance(args.train_data_template, str):
-        args.train_data_template = [args.train_data_template] * len(args.placeholder_tokens)
+    if not args.simultaneous:
+        if isinstance(args.train_data_template, str):
+            args.train_data_template = [args.train_data_template] * len(args.placeholder_tokens)
 
-    if len(args.placeholder_tokens) != len(args.train_data_template):
-        raise ValueError("--placeholder_tokens and --train_data_template must have the same number of items")
+        if len(args.placeholder_tokens) != len(args.train_data_template):
+            raise ValueError("--placeholder_tokens and --train_data_template must have the same number of items")
 
     if isinstance(args.collection, str):
         args.collection = [args.collection]
@@ -560,6 +565,8 @@ def main():
     elif args.mixed_precision == "bf16":
         weight_dtype = torch.bfloat16
 
+    checkpoint_output_dir = output_dir.joinpath("checkpoints")
+
     trainer = partial(
         train,
         accelerator=accelerator,
@@ -569,30 +576,50 @@ def main():
         noise_scheduler=noise_scheduler,
         dtype=weight_dtype,
         seed=args.seed,
-        callbacks_fn=textual_inversion_strategy
-    )
-
-    checkpoint_output_dir = output_dir.joinpath("checkpoints")
-
-    for i, placeholder_token, initializer_token, num_vectors, data_template in zip(
-        range(len(args.placeholder_tokens)),
-        args.placeholder_tokens,
-        args.initializer_tokens,
-        args.num_vectors,
-        args.train_data_template
-    ):
-        sample_output_dir = output_dir.joinpath(f"samples_{placeholder_token}")
+        with_prior_preservation=args.num_class_images != 0,
+        prior_loss_weight=args.prior_loss_weight,
+        strategy=textual_inversion_strategy,
+        num_train_epochs=args.num_train_epochs,
+        sample_frequency=args.sample_frequency,
+        checkpoint_frequency=args.checkpoint_frequency,
+        global_step_offset=global_step_offset,
+        # --
+        tokenizer=tokenizer,
+        sample_scheduler=sample_scheduler,
+        checkpoint_output_dir=checkpoint_output_dir,
+        learning_rate=args.learning_rate,
+        gradient_checkpointing=args.gradient_checkpointing,
+        use_emb_decay=args.use_emb_decay,
+        emb_decay_target=args.emb_decay_target,
+        emb_decay_factor=args.emb_decay_factor,
+        emb_decay_start=args.emb_decay_start,
+        use_ema=args.use_ema,
+        ema_inv_gamma=args.ema_inv_gamma,
+        ema_power=args.ema_power,
+        ema_max_decay=args.ema_max_decay,
+        sample_batch_size=args.sample_batch_size,
+        sample_num_batches=args.sample_batches,
+        sample_num_steps=args.sample_steps,
+        sample_image_size=args.sample_image_size,
+    )
+
+    def run(i: int, placeholder_tokens, initializer_tokens, num_vectors, data_template):
+        if len(placeholder_tokens) == 1:
+            sample_output_dir = output_dir.joinpath(f"samples_{placeholder_token[0]}")
+        else:
+            sample_output_dir = output_dir.joinpath("samples")
 
         placeholder_token_ids, initializer_token_ids = add_placeholder_tokens(
             tokenizer=tokenizer,
             embeddings=embeddings,
-            placeholder_tokens=[placeholder_token],
-            initializer_tokens=[initializer_token],
-            num_vectors=[num_vectors]
+            placeholder_tokens=placeholder_tokens,
+            initializer_tokens=initializer_tokens,
+            num_vectors=num_vectors
         )
 
-        print(
-            f"{i + 1}: {placeholder_token}, {placeholder_token_ids[0]} ({initializer_token}, {initializer_token_ids[0]})")
+        stats = list(zip(placeholder_tokens, placeholder_token_ids, initializer_tokens, initializer_token_ids))
+
+        print(f"{i + 1}: {stats})")
 
         datamodule = VlpnDataModule(
             data_file=args.train_data_file,
@@ -612,7 +639,7 @@ def main():
             train_set_pad=args.train_set_pad,
             valid_set_pad=args.valid_set_pad,
             seed=args.seed,
-            filter=partial(keyword_filter, [placeholder_token], args.collection, args.exclude_collections),
+            filter=partial(keyword_filter, placeholder_tokens, args.collection, args.exclude_collections),
             dtype=weight_dtype
         )
         datamodule.setup()
@@ -647,36 +674,24 @@ def main():
             val_dataloader=datamodule.val_dataloader,
             optimizer=optimizer,
             lr_scheduler=lr_scheduler,
-            num_train_epochs=args.num_train_epochs,
-            sample_frequency=args.sample_frequency,
-            checkpoint_frequency=args.checkpoint_frequency,
-            global_step_offset=global_step_offset,
-            with_prior_preservation=args.num_class_images != 0,
-            prior_loss_weight=args.prior_loss_weight,
             # --
-            tokenizer=tokenizer,
-            sample_scheduler=sample_scheduler,
             sample_output_dir=sample_output_dir,
-            checkpoint_output_dir=checkpoint_output_dir,
-            placeholder_tokens=[placeholder_token],
+            placeholder_tokens=placeholder_tokens,
             placeholder_token_ids=placeholder_token_ids,
-            learning_rate=args.learning_rate,
-            gradient_checkpointing=args.gradient_checkpointing,
-            use_emb_decay=args.use_emb_decay,
-            emb_decay_target=args.emb_decay_target,
-            emb_decay_factor=args.emb_decay_factor,
-            emb_decay_start=args.emb_decay_start,
-            use_ema=args.use_ema,
-            ema_inv_gamma=args.ema_inv_gamma,
-            ema_power=args.ema_power,
-            ema_max_decay=args.ema_max_decay,
-            sample_batch_size=args.sample_batch_size,
-            sample_num_batches=args.sample_batches,
-            sample_num_steps=args.sample_steps,
-            sample_image_size=args.sample_image_size,
         )
 
-        embeddings.persist()
+    if args.simultaneous:
+        run(0, args.placeholder_tokens, args.initializer_tokens, args.num_vectors, args.train_data_template)
+    else:
+        for i, placeholder_token, initializer_token, num_vectors, data_template in zip(
+            range(len(args.placeholder_tokens)),
+            args.placeholder_tokens,
+            args.initializer_tokens,
+            args.num_vectors,
+            args.train_data_template
+        ):
+            run(i, [placeholder_token], [initializer_token], [num_vectors], data_template)
+            embeddings.persist()
 
 
 if __name__ == "__main__":
diff --git a/training/functional.py b/training/functional.py
index 3d27380..7a3e821 100644
--- a/training/functional.py
+++ b/training/functional.py
@@ -39,11 +39,18 @@ class TrainingCallbacks():
     on_train: Callable[[int], _GeneratorContextManager] = const(nullcontext())
     on_before_optimize: Callable[[int], None] = const()
     on_after_optimize: Callable[[float], None] = const()
+    on_after_epoch: Callable[[float], None] = const()
     on_eval: Callable[[], _GeneratorContextManager] = const(nullcontext())
     on_sample: Callable[[int], None] = const()
     on_checkpoint: Callable[[int, str], None] = const()
 
 
+@dataclass
+class TrainingStrategy():
+    callbacks: Callable[..., TrainingCallbacks]
+    prepare_unet: bool = False
+
+
 def make_grid(images, rows, cols):
     w, h = images[0].size
     grid = Image.new('RGB', size=(cols*w, rows*h))
@@ -373,6 +380,7 @@ def train_loop(
     on_train = callbacks.on_train
     on_before_optimize = callbacks.on_before_optimize
     on_after_optimize = callbacks.on_after_optimize
+    on_after_epoch = callbacks.on_after_epoch
     on_eval = callbacks.on_eval
     on_sample = callbacks.on_sample
     on_checkpoint = callbacks.on_checkpoint
@@ -434,6 +442,8 @@ def train_loop(
 
             accelerator.wait_for_everyone()
 
+            on_after_epoch(lr_scheduler.get_last_lr()[0])
+
             if val_dataloader is not None:
                 model.eval()
 
@@ -512,8 +522,7 @@ def train(
     val_dataloader: Optional[DataLoader],
     optimizer: torch.optim.Optimizer,
     lr_scheduler: torch.optim.lr_scheduler._LRScheduler,
-    callbacks_fn: Callable[..., TrainingCallbacks],
-    prepare_unet: bool = False,
+    strategy: TrainingStrategy,
     num_train_epochs: int = 100,
     sample_frequency: int = 20,
     checkpoint_frequency: int = 50,
@@ -524,12 +533,12 @@ def train(
 ):
     prep = [text_encoder, optimizer, train_dataloader, val_dataloader, lr_scheduler]
 
-    if prepare_unet:
+    if strategy.prepare_unet:
         prep.append(unet)
 
     prep = accelerator.prepare(*prep)
 
-    if prepare_unet:
+    if strategy.prepare_unet:
         text_encoder, optimizer, train_dataloader, val_dataloader, lr_scheduler, unet = prep
     else:
         text_encoder, optimizer, train_dataloader, val_dataloader, lr_scheduler = prep
@@ -542,7 +551,7 @@ def train(
         model.requires_grad_(False)
         model.eval()
 
-    callbacks = callbacks_fn(
+    callbacks = strategy.callbacks(
         accelerator=accelerator,
         unet=unet,
         text_encoder=text_encoder,
diff --git a/training/strategy/dreambooth.py b/training/strategy/dreambooth.py
index 93c81cb..bc26ee6 100644
--- a/training/strategy/dreambooth.py
+++ b/training/strategy/dreambooth.py
@@ -15,10 +15,10 @@ from diffusers import AutoencoderKL, UNet2DConditionModel, DPMSolverMultistepSch
 from pipelines.stable_diffusion.vlpn_stable_diffusion import VlpnStableDiffusion
 from models.clip.tokenizer import MultiCLIPTokenizer
 from training.util import EMAModel
-from training.functional import TrainingCallbacks, save_samples
+from training.functional import TrainingStrategy, TrainingCallbacks, save_samples
 
 
-def dreambooth_strategy(
+def dreambooth_strategy_callbacks(
     accelerator: Accelerator,
     unet: UNet2DConditionModel,
     text_encoder: CLIPTextModel,
@@ -185,3 +185,9 @@ def dreambooth_strategy(
         on_checkpoint=on_checkpoint,
         on_sample=on_sample,
     )
+
+
+dreambooth_strategy = TrainingStrategy(
+    callbacks=dreambooth_strategy_callbacks,
+    prepare_unet=True
+)
diff --git a/training/strategy/ti.py b/training/strategy/ti.py
index 00f3529..597abd0 100644
--- a/training/strategy/ti.py
+++ b/training/strategy/ti.py
@@ -15,10 +15,10 @@ from slugify import slugify
 
 from models.clip.tokenizer import MultiCLIPTokenizer
 from training.util import EMAModel
-from training.functional import TrainingCallbacks, save_samples
+from training.functional import TrainingStrategy, TrainingCallbacks, save_samples
 
 
-def textual_inversion_strategy(
+def textual_inversion_strategy_callbacks(
     accelerator: Accelerator,
     unet: UNet2DConditionModel,
     text_encoder: CLIPTextModel,
@@ -119,17 +119,18 @@ def textual_inversion_strategy(
         with ema_context():
             yield
 
-    @torch.no_grad()
     def on_after_optimize(lr: float):
+        if use_ema:
+            ema_embeddings.step(text_encoder.text_model.embeddings.temp_token_embedding.parameters())
+
+    @torch.no_grad()
+    def on_after_epoch(lr: float):
         if use_emb_decay:
             text_encoder.text_model.embeddings.normalize(
                 emb_decay_target,
                 min(1.0, max(0.0, emb_decay_factor * ((lr - emb_decay_start) / (learning_rate - emb_decay_start))))
             )
 
-        if use_ema:
-            ema_embeddings.step(text_encoder.text_model.embeddings.temp_token_embedding.parameters())
-
     def on_log():
         if use_ema:
             return {"ema_decay": ema_embeddings.decay}
@@ -157,7 +158,13 @@ def textual_inversion_strategy(
         on_train=on_train,
         on_eval=on_eval,
         on_after_optimize=on_after_optimize,
+        on_after_epoch=on_after_epoch,
         on_log=on_log,
         on_checkpoint=on_checkpoint,
         on_sample=on_sample,
     )
+
+
+textual_inversion_strategy = TrainingStrategy(
+    callbacks=textual_inversion_strategy_callbacks,
+)
diff --git a/training/util.py b/training/util.py
index 557b196..237626f 100644
--- a/training/util.py
+++ b/training/util.py
@@ -1,18 +1,11 @@
 from pathlib import Path
 import json
 import copy
-from typing import Iterable, Union
+from typing import Iterable, Any
 from contextlib import contextmanager
 
 import torch
 
-from transformers import CLIPTextModel
-from diffusers import AutoencoderKL, DDPMScheduler, UNet2DConditionModel, DPMSolverMultistepScheduler
-
-from pipelines.stable_diffusion.vlpn_stable_diffusion import VlpnStableDiffusion
-from models.clip.tokenizer import MultiCLIPTokenizer
-from models.clip.embeddings import ManagedCLIPTextEmbeddings, patch_managed_embeddings
-
 
 def save_args(basepath: Path, args, extra={}):
     info = {"args": vars(args)}
@@ -22,6 +15,8 @@ def save_args(basepath: Path, args, extra={}):
 
 
 class AverageMeter:
+    avg: Any
+
     def __init__(self, name=None):
         self.name = name
         self.reset()