From 4d19f3b7d461fe0f63e7ccff936909b0ce0c6126 Mon Sep 17 00:00:00 2001 From: Melan Date: Fri, 14 Oct 2022 22:45:26 +0200 Subject: Raise an assertion error if no training images have been found. --- modules/textual_inversion/dataset.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'modules/textual_inversion') diff --git a/modules/textual_inversion/dataset.py b/modules/textual_inversion/dataset.py index 67e90afe..12e2f43b 100644 --- a/modules/textual_inversion/dataset.py +++ b/modules/textual_inversion/dataset.py @@ -81,7 +81,8 @@ class PersonalizedBase(Dataset): entry.cond = cond_model([entry.cond_text]).to(devices.cpu) self.dataset.append(entry) - + + assert len(self.dataset) > 1, "No images have been found in the dataset." self.length = len(self.dataset) * repeats self.initial_indexes = np.arange(self.length) % len(self.dataset) -- cgit v1.2.1 From c7a86f7fe9c0b8967a87e8d709f507d2f44400d8 Mon Sep 17 00:00:00 2001 From: AUTOMATIC <16777216c@gmail.com> Date: Sat, 15 Oct 2022 09:24:59 +0300 Subject: add option to use batch size for training --- modules/textual_inversion/dataset.py | 31 ++++++++++++++++---------- modules/textual_inversion/textual_inversion.py | 17 +++++++------- 2 files changed, 27 insertions(+), 21 deletions(-) (limited to 'modules/textual_inversion') diff --git a/modules/textual_inversion/dataset.py b/modules/textual_inversion/dataset.py index 67e90afe..bd99c0cb 100644 --- a/modules/textual_inversion/dataset.py +++ b/modules/textual_inversion/dataset.py @@ -24,11 +24,12 @@ class DatasetEntry: class PersonalizedBase(Dataset): - def __init__(self, data_root, width, height, repeats, flip_p=0.5, placeholder_token="*", model=None, device=None, template_file=None, include_cond=False): - re_word = re.compile(shared.opts.dataset_filename_word_regex) if len(shared.opts.dataset_filename_word_regex)>0 else None + def __init__(self, data_root, width, height, repeats, flip_p=0.5, placeholder_token="*", model=None, device=None, template_file=None, include_cond=False, batch_size=1): + re_word = re.compile(shared.opts.dataset_filename_word_regex) if len(shared.opts.dataset_filename_word_regex) > 0 else None self.placeholder_token = placeholder_token + self.batch_size = batch_size self.width = width self.height = height self.flip = transforms.RandomHorizontalFlip(p=flip_p) @@ -78,13 +79,13 @@ class PersonalizedBase(Dataset): if include_cond: entry.cond_text = self.create_text(filename_text) - entry.cond = cond_model([entry.cond_text]).to(devices.cpu) + entry.cond = cond_model([entry.cond_text]).to(devices.cpu).squeeze(0) self.dataset.append(entry) - self.length = len(self.dataset) * repeats + self.length = len(self.dataset) * repeats // batch_size - self.initial_indexes = np.arange(self.length) % len(self.dataset) + self.initial_indexes = np.arange(len(self.dataset)) self.indexes = None self.shuffle() @@ -101,13 +102,19 @@ class PersonalizedBase(Dataset): return self.length def __getitem__(self, i): - if i % len(self.dataset) == 0: - self.shuffle() + res = [] - index = self.indexes[i % len(self.indexes)] - entry = self.dataset[index] + for j in range(self.batch_size): + position = i * self.batch_size + j + if position % len(self.indexes) == 0: + self.shuffle() - if entry.cond is None: - entry.cond_text = self.create_text(entry.filename_text) + index = self.indexes[position % len(self.indexes)] + entry = self.dataset[index] - return entry + if entry.cond is None: + entry.cond_text = self.create_text(entry.filename_text) + + res.append(entry) + + return res diff --git a/modules/textual_inversion/textual_inversion.py b/modules/textual_inversion/textual_inversion.py index da0d77a0..e754747e 100644 --- a/modules/textual_inversion/textual_inversion.py +++ b/modules/textual_inversion/textual_inversion.py @@ -199,7 +199,7 @@ def write_loss(log_directory, filename, step, epoch_len, values): }) -def train_embedding(embedding_name, learn_rate, data_root, log_directory, training_width, training_height, steps, create_image_every, save_embedding_every, template_file, save_image_with_stored_embedding, preview_from_txt2img, preview_prompt, preview_negative_prompt, preview_steps, preview_sampler_index, preview_cfg_scale, preview_seed, preview_width, preview_height): +def train_embedding(embedding_name, learn_rate, batch_size, data_root, log_directory, training_width, training_height, steps, create_image_every, save_embedding_every, template_file, save_image_with_stored_embedding, preview_from_txt2img, preview_prompt, preview_negative_prompt, preview_steps, preview_sampler_index, preview_cfg_scale, preview_seed, preview_width, preview_height): assert embedding_name, 'embedding not selected' shared.state.textinfo = "Initializing textual inversion training..." @@ -231,7 +231,7 @@ def train_embedding(embedding_name, learn_rate, data_root, log_directory, traini shared.state.textinfo = f"Preparing dataset from {html.escape(data_root)}..." with torch.autocast("cuda"): - ds = modules.textual_inversion.dataset.PersonalizedBase(data_root=data_root, width=training_width, height=training_height, repeats=shared.opts.training_image_repeats_per_epoch, placeholder_token=embedding_name, model=shared.sd_model, device=devices.device, template_file=template_file) + ds = modules.textual_inversion.dataset.PersonalizedBase(data_root=data_root, width=training_width, height=training_height, repeats=shared.opts.training_image_repeats_per_epoch, placeholder_token=embedding_name, model=shared.sd_model, device=devices.device, template_file=template_file, batch_size=batch_size) hijack = sd_hijack.model_hijack @@ -251,7 +251,7 @@ def train_embedding(embedding_name, learn_rate, data_root, log_directory, traini optimizer = torch.optim.AdamW([embedding.vec], lr=scheduler.learn_rate) pbar = tqdm.tqdm(enumerate(ds), total=steps-ititial_step) - for i, entry in pbar: + for i, entries in pbar: embedding.step = i + ititial_step scheduler.apply(optimizer, embedding.step) @@ -262,10 +262,9 @@ def train_embedding(embedding_name, learn_rate, data_root, log_directory, traini break with torch.autocast("cuda"): - c = cond_model([entry.cond_text]) - - x = entry.latent.to(devices.device) - loss = shared.sd_model(x.unsqueeze(0), c)[0] + c = cond_model([entry.cond_text for entry in entries]) + x = torch.stack([entry.latent for entry in entries]).to(devices.device) + loss = shared.sd_model(x, c)[0] del x losses[embedding.step % losses.shape[0]] = loss.item() @@ -307,7 +306,7 @@ def train_embedding(embedding_name, learn_rate, data_root, log_directory, traini p.width = preview_width p.height = preview_height else: - p.prompt = entry.cond_text + p.prompt = entries[0].cond_text p.steps = 20 p.width = training_width p.height = training_height @@ -348,7 +347,7 @@ def train_embedding(embedding_name, learn_rate, data_root, log_directory, traini

Loss: {losses.mean():.7f}
Step: {embedding.step}
-Last prompt: {html.escape(entry.cond_text)}
+Last prompt: {html.escape(entries[0].cond_text)}
Last saved embedding: {html.escape(last_saved_file)}
Last saved image: {html.escape(last_saved_image)}

-- cgit v1.2.1 From 9e846083b702a498fdb60accd72f075fa26701d9 Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Fri, 14 Oct 2022 14:50:25 +0100 Subject: add vector size to embed text --- modules/textual_inversion/textual_inversion.py | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) (limited to 'modules/textual_inversion') diff --git a/modules/textual_inversion/textual_inversion.py b/modules/textual_inversion/textual_inversion.py index e754747e..6f549d62 100644 --- a/modules/textual_inversion/textual_inversion.py +++ b/modules/textual_inversion/textual_inversion.py @@ -327,10 +327,16 @@ def train_embedding(embedding_name, learn_rate, batch_size, data_root, log_direc info.add_text("sd-ti-embedding", embedding_to_b64(data)) title = "<{}>".format(data.get('name', '???')) + + try: + vectorSize = list(data['string_to_param'].values())[0].shape[0] + except Exception as e: + vectorSize = '?' + checkpoint = sd_models.select_checkpoint() footer_left = checkpoint.model_name footer_mid = '[{}]'.format(checkpoint.hash) - footer_right = '{}'.format(embedding.step) + footer_right = 'v{} {}s'.format(vectorSize, embedding.step) captioned_image = caption_image_overlay(image, title, footer_left, footer_mid, footer_right) captioned_image = insert_image_data_embed(captioned_image, data) -- cgit v1.2.1 From 939f16529a72fe48c2ce3ef31bdaba785925a33c Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Fri, 14 Oct 2022 14:55:05 +0100 Subject: only save 1 image per embedding --- modules/textual_inversion/textual_inversion.py | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'modules/textual_inversion') diff --git a/modules/textual_inversion/textual_inversion.py b/modules/textual_inversion/textual_inversion.py index 6f549d62..1d697c90 100644 --- a/modules/textual_inversion/textual_inversion.py +++ b/modules/textual_inversion/textual_inversion.py @@ -242,6 +242,7 @@ def train_embedding(embedding_name, learn_rate, batch_size, data_root, log_direc last_saved_file = "" last_saved_image = "" + embedding_yet_to_be_embedded = False ititial_step = embedding.step or 0 if ititial_step > steps: @@ -281,6 +282,7 @@ def train_embedding(embedding_name, learn_rate, batch_size, data_root, log_direc if embedding.step > 0 and embedding_dir is not None and embedding.step % save_embedding_every == 0: last_saved_file = os.path.join(embedding_dir, f'{embedding_name}-{embedding.step}.pt') embedding.save(last_saved_file) + embedding_yet_to_be_embedded = True write_loss(log_directory, "textual_inversion_loss.csv", embedding.step, len(ds), { "loss": f"{losses.mean():.7f}", @@ -318,7 +320,7 @@ def train_embedding(embedding_name, learn_rate, batch_size, data_root, log_direc shared.state.current_image = image - if save_image_with_stored_embedding and os.path.exists(last_saved_file): + if save_image_with_stored_embedding and os.path.exists(last_saved_file) and embedding_yet_to_be_embedded: last_saved_image_chunks = os.path.join(images_embeds_dir, f'{embedding_name}-{embedding.step}.png') @@ -342,6 +344,7 @@ def train_embedding(embedding_name, learn_rate, batch_size, data_root, log_direc captioned_image = insert_image_data_embed(captioned_image, data) captioned_image.save(last_saved_image_chunks, "PNG", pnginfo=info) + embedding_yet_to_be_embedded = False image.save(last_saved_image) -- cgit v1.2.1 From 9a1dcd78edbf9caf68b9e6286d7b5ca81500e243 Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Fri, 14 Oct 2022 18:14:02 +0100 Subject: add webp for embed load --- modules/textual_inversion/textual_inversion.py | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'modules/textual_inversion') diff --git a/modules/textual_inversion/textual_inversion.py b/modules/textual_inversion/textual_inversion.py index 1d697c90..c07bffc3 100644 --- a/modules/textual_inversion/textual_inversion.py +++ b/modules/textual_inversion/textual_inversion.py @@ -96,6 +96,10 @@ class EmbeddingDatabase: else: data = extract_image_data_embed(embed_image) name = data.get('name', name) + elif filename.upper().endswith('.WEBP'): + embed_image = Image.open(path) + data = extract_image_data_embed(embed_image) + name = data.get('name', name) else: data = torch.load(path, map_location="cpu") -- cgit v1.2.1 From ddf6899df0cf87d4da77cb2ce223061f4a5edf18 Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Fri, 14 Oct 2022 18:23:20 +0100 Subject: generalise to popular lossless formats --- modules/textual_inversion/textual_inversion.py | 8 ++------ 1 file changed, 2 insertions(+), 6 deletions(-) (limited to 'modules/textual_inversion') diff --git a/modules/textual_inversion/textual_inversion.py b/modules/textual_inversion/textual_inversion.py index c07bffc3..b99df3b1 100644 --- a/modules/textual_inversion/textual_inversion.py +++ b/modules/textual_inversion/textual_inversion.py @@ -88,18 +88,14 @@ class EmbeddingDatabase: data = [] - if filename.upper().endswith('.PNG'): + if os.path.splitext(filename.upper())[-1] in ['.PNG', '.WEBP', '.JXL', '.AVIF']: embed_image = Image.open(path) - if 'sd-ti-embedding' in embed_image.text: + if hasattr(embed_image, 'text') and 'sd-ti-embedding' in embed_image.text: data = embedding_from_b64(embed_image.text['sd-ti-embedding']) name = data.get('name', name) else: data = extract_image_data_embed(embed_image) name = data.get('name', name) - elif filename.upper().endswith('.WEBP'): - embed_image = Image.open(path) - data = extract_image_data_embed(embed_image) - name = data.get('name', name) else: data = torch.load(path, map_location="cpu") -- cgit v1.2.1 From b6e3b96dab94a00f51725f9cc977eebc6b4072ab Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Sat, 15 Oct 2022 15:17:21 +0100 Subject: Change vector size footer label --- modules/textual_inversion/textual_inversion.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'modules/textual_inversion') diff --git a/modules/textual_inversion/textual_inversion.py b/modules/textual_inversion/textual_inversion.py index b99df3b1..2ed345b1 100644 --- a/modules/textual_inversion/textual_inversion.py +++ b/modules/textual_inversion/textual_inversion.py @@ -338,7 +338,7 @@ def train_embedding(embedding_name, learn_rate, batch_size, data_root, log_direc checkpoint = sd_models.select_checkpoint() footer_left = checkpoint.model_name footer_mid = '[{}]'.format(checkpoint.hash) - footer_right = 'v{} {}s'.format(vectorSize, embedding.step) + footer_right = '{}v {}s'.format(vectorSize, embedding.step) captioned_image = caption_image_overlay(image, title, footer_left, footer_mid, footer_right) captioned_image = insert_image_data_embed(captioned_image, data) -- cgit v1.2.1