From 019a3a88f07766f2d32c32fbe8e41625f28ecb5e Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Wed, 19 Oct 2022 17:15:47 +0100 Subject: [PATCH 01/14] Update ui.py --- modules/ui.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/ui.py b/modules/ui.py index d2e24880..1573ef82 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -1247,7 +1247,7 @@ def create_ui(wrap_gradio_gpu_call): run_preprocess = gr.Button(value="Preprocess", variant='primary') with gr.Tab(label="Train"): - gr.HTML(value="

Train an embedding; must specify a directory with a set of 1:1 ratio images

") + gr.HTML(value="

Train an embedding or Hypernetwork; you must specify a directory with a set of 1:1 ratio images
Initial learning rates: 0.005 for an Embedding, 0.00001 for Hypernetwork wiki

") with gr.Row(): train_embedding_name = gr.Dropdown(label='Embedding', elem_id="train_embedding", choices=sorted(sd_hijack.model_hijack.embedding_db.word_embeddings.keys())) create_refresh_button(train_embedding_name, sd_hijack.model_hijack.embedding_db.load_textual_inversion_embeddings, lambda: {"choices": sorted(sd_hijack.model_hijack.embedding_db.word_embeddings.keys())}, "refresh_train_embedding_name") From eb7ba4b713ac2fb960ecf6365b1de0c89451e583 Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Wed, 19 Oct 2022 19:50:46 +0100 Subject: [PATCH 02/14] update training header text --- modules/ui.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/ui.py b/modules/ui.py index 1573ef82..93c0767c 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -1247,7 +1247,7 @@ def create_ui(wrap_gradio_gpu_call): run_preprocess = gr.Button(value="Preprocess", variant='primary') with gr.Tab(label="Train"): - gr.HTML(value="

Train an embedding or Hypernetwork; you must specify a directory with a set of 1:1 ratio images
Initial learning rates: 0.005 for an Embedding, 0.00001 for Hypernetwork wiki

") + gr.HTML(value="

Train an embedding or Hypernetwork; you must specify a directory with a set of 1:1 ratio images
Initial learning rates: 0.005 for an Embedding, 0.00001 for Hypernetwork [wiki]

") with gr.Row(): train_embedding_name = gr.Dropdown(label='Embedding', elem_id="train_embedding", choices=sorted(sd_hijack.model_hijack.embedding_db.word_embeddings.keys())) create_refresh_button(train_embedding_name, sd_hijack.model_hijack.embedding_db.load_textual_inversion_embeddings, lambda: {"choices": sorted(sd_hijack.model_hijack.embedding_db.word_embeddings.keys())}, "refresh_train_embedding_name") From 4d663055ded968831ec97f047dfa8e94036cf1c1 Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Wed, 19 Oct 2022 20:33:18 +0100 Subject: [PATCH 03/14] update ui with extra training options --- modules/ui.py | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/modules/ui.py b/modules/ui.py index 93c0767c..cdb9d335 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -1206,6 +1206,7 @@ def create_ui(wrap_gradio_gpu_call): new_embedding_name = gr.Textbox(label="Name") initialization_text = gr.Textbox(label="Initialization text", value="*") nvpt = gr.Slider(label="Number of vectors per token", minimum=1, maximum=75, step=1, value=1) + overwrite_old_embedding = gr.Checkbox(value=False, label="Overwrite Old Embedding") with gr.Row(): with gr.Column(scale=3): @@ -1219,6 +1220,7 @@ def create_ui(wrap_gradio_gpu_call): new_hypernetwork_sizes = gr.CheckboxGroup(label="Modules", value=["768", "320", "640", "1280"], choices=["768", "320", "640", "1280"]) new_hypernetwork_layer_structure = gr.Textbox("1, 2, 1", label="Enter hypernetwork layer structure", placeholder="1st and last digit must be 1. ex:'1, 2, 1'") new_hypernetwork_add_layer_norm = gr.Checkbox(label="Add layer normalization") + overwrite_old_hypernetwork = gr.Checkbox(value=False, label="Overwrite Old Hypernetwork") with gr.Row(): with gr.Column(scale=3): @@ -1247,14 +1249,17 @@ def create_ui(wrap_gradio_gpu_call): run_preprocess = gr.Button(value="Preprocess", variant='primary') with gr.Tab(label="Train"): - gr.HTML(value="

Train an embedding or Hypernetwork; you must specify a directory with a set of 1:1 ratio images
Initial learning rates: 0.005 for an Embedding, 0.00001 for Hypernetwork [wiki]

") + gr.HTML(value="

Train an embedding or Hypernetwork; you must specify a directory with a set of 1:1 ratio images [wiki]

") with gr.Row(): train_embedding_name = gr.Dropdown(label='Embedding', elem_id="train_embedding", choices=sorted(sd_hijack.model_hijack.embedding_db.word_embeddings.keys())) create_refresh_button(train_embedding_name, sd_hijack.model_hijack.embedding_db.load_textual_inversion_embeddings, lambda: {"choices": sorted(sd_hijack.model_hijack.embedding_db.word_embeddings.keys())}, "refresh_train_embedding_name") with gr.Row(): train_hypernetwork_name = gr.Dropdown(label='Hypernetwork', elem_id="train_hypernetwork", choices=[x for x in shared.hypernetworks.keys()]) create_refresh_button(train_hypernetwork_name, shared.reload_hypernetworks, lambda: {"choices": sorted([x for x in shared.hypernetworks.keys()])}, "refresh_train_hypernetwork_name") - learn_rate = gr.Textbox(label='Learning rate', placeholder="Learning rate", value="0.005") + with gr.Row(): + embedding_learn_rate = gr.Textbox(label='Embedding Learning rate', placeholder="Embedding Learning rate", value="0.005") + hypernetwork_learn_rate = gr.Textbox(label='Hypernetwork Learning rate', placeholder="Hypernetwork Learning rate", value="0.00001") + batch_size = gr.Number(label='Batch size', value=1, precision=0) dataset_directory = gr.Textbox(label='Dataset directory', placeholder="Path to directory with input images") log_directory = gr.Textbox(label='Log directory', placeholder="Path to directory where to write outputs", value="textual_inversion") @@ -1288,6 +1293,7 @@ def create_ui(wrap_gradio_gpu_call): new_embedding_name, initialization_text, nvpt, + overwrite_old_embedding, ], outputs=[ train_embedding_name, @@ -1303,6 +1309,7 @@ def create_ui(wrap_gradio_gpu_call): new_hypernetwork_sizes, new_hypernetwork_layer_structure, new_hypernetwork_add_layer_norm, + overwrite_old_hypernetwork, ], outputs=[ train_hypernetwork_name, From d6ea5841374a28f3f6deb73abc251c8f0bcb240f Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Thu, 20 Oct 2022 00:07:57 +0100 Subject: [PATCH 04/14] change html output --- modules/hypernetworks/hypernetwork.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/hypernetworks/hypernetwork.py b/modules/hypernetworks/hypernetwork.py index 7d519cd9..73c1cb80 100644 --- a/modules/hypernetworks/hypernetwork.py +++ b/modules/hypernetworks/hypernetwork.py @@ -380,7 +380,7 @@ def train_hypernetwork(hypernetwork_name, learn_rate, batch_size, data_root, log Loss: {mean_loss:.7f}
Step: {hypernetwork.step}
Last prompt: {html.escape(entries[0].cond_text)}
-Last saved embedding: {html.escape(last_saved_file)}
+Last saved hypernetwork: {html.escape(last_saved_file)}
Last saved image: {html.escape(last_saved_image)}

""" From 166be3919b817cee5e702fd01c34afe9081b952c Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Thu, 20 Oct 2022 00:09:40 +0100 Subject: [PATCH 05/14] allow overwrite old hn --- modules/hypernetworks/ui.py | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/modules/hypernetworks/ui.py b/modules/hypernetworks/ui.py index 08f75f15..f45345ea 100644 --- a/modules/hypernetworks/ui.py +++ b/modules/hypernetworks/ui.py @@ -10,9 +10,10 @@ from modules import sd_hijack, shared, devices from modules.hypernetworks import hypernetwork -def create_hypernetwork(name, enable_sizes, layer_structure=None, add_layer_norm=False): +def create_hypernetwork(name, enable_sizes, overwrite_old, layer_structure=None, add_layer_norm=False): fn = os.path.join(shared.cmd_opts.hypernetwork_dir, f"{name}.pt") - assert not os.path.exists(fn), f"file {fn} already exists" + if not overwrite_old: + assert not os.path.exists(fn), f"file {fn} already exists" if type(layer_structure) == str: layer_structure = tuple(map(int, re.sub(r'\D', '', layer_structure))) From 0087079c2d487b67b06ffc30f36ce486a74e6318 Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Thu, 20 Oct 2022 00:10:59 +0100 Subject: [PATCH 06/14] allow overwrite old embedding --- modules/textual_inversion/textual_inversion.py | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/modules/textual_inversion/textual_inversion.py b/modules/textual_inversion/textual_inversion.py index 3be69562..5776778b 100644 --- a/modules/textual_inversion/textual_inversion.py +++ b/modules/textual_inversion/textual_inversion.py @@ -153,7 +153,7 @@ class EmbeddingDatabase: return None, None -def create_embedding(name, num_vectors_per_token, init_text='*'): +def create_embedding(name, num_vectors_per_token, overwrite_old, init_text='*'): cond_model = shared.sd_model.cond_stage_model embedding_layer = cond_model.wrapped.transformer.text_model.embeddings @@ -165,7 +165,8 @@ def create_embedding(name, num_vectors_per_token, init_text='*'): vec[i] = embedded[i * int(embedded.shape[0]) // num_vectors_per_token] fn = os.path.join(shared.cmd_opts.embeddings_dir, f"{name}.pt") - assert not os.path.exists(fn), f"file {fn} already exists" + if not overwrite_old: + assert not os.path.exists(fn), f"file {fn} already exists" embedding = Embedding(vec, name) embedding.step = 0 From 632e8d660293081cadb145d8062e5aff0a4a8f0d Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Thu, 20 Oct 2022 00:19:40 +0100 Subject: [PATCH 07/14] split learn rates --- modules/ui.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/ui.py b/modules/ui.py index cdb9d335..d07184ee 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -1342,7 +1342,7 @@ def create_ui(wrap_gradio_gpu_call): _js="start_training_textual_inversion", inputs=[ train_embedding_name, - learn_rate, + embedding_learn_rate, batch_size, dataset_directory, log_directory, @@ -1367,7 +1367,7 @@ def create_ui(wrap_gradio_gpu_call): _js="start_training_textual_inversion", inputs=[ train_hypernetwork_name, - learn_rate, + hypernetwork_learn_rate, batch_size, dataset_directory, log_directory, From c3835ec85cbb44fa3c46fa871c622b6fee235c89 Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Thu, 20 Oct 2022 00:24:24 +0100 Subject: [PATCH 08/14] pass overwrite old flag --- modules/textual_inversion/ui.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/modules/textual_inversion/ui.py b/modules/textual_inversion/ui.py index 36881e7a..e712284d 100644 --- a/modules/textual_inversion/ui.py +++ b/modules/textual_inversion/ui.py @@ -7,8 +7,8 @@ import modules.textual_inversion.preprocess from modules import sd_hijack, shared -def create_embedding(name, initialization_text, nvpt): - filename = modules.textual_inversion.textual_inversion.create_embedding(name, nvpt, init_text=initialization_text) +def create_embedding(name, initialization_text, nvpt, overwrite_old): + filename = modules.textual_inversion.textual_inversion.create_embedding(name, nvpt, overwrite_old, init_text=initialization_text) sd_hijack.model_hijack.embedding_db.load_textual_inversion_embeddings() From 4d6b9f76a55fd0ac0f72634071032dd9c6efb409 Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Thu, 20 Oct 2022 00:27:16 +0100 Subject: [PATCH 09/14] reorder create_hypernetwork params --- modules/ui.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/ui.py b/modules/ui.py index d07184ee..322c082b 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -1307,9 +1307,9 @@ def create_ui(wrap_gradio_gpu_call): inputs=[ new_hypernetwork_name, new_hypernetwork_sizes, + overwrite_old_hypernetwork, new_hypernetwork_layer_structure, new_hypernetwork_add_layer_norm, - overwrite_old_hypernetwork, ], outputs=[ train_hypernetwork_name, From fbcce66601994f6ed370db36d9c238840fed6bd2 Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Thu, 20 Oct 2022 00:46:54 +0100 Subject: [PATCH 10/14] add existing caption file handling --- modules/textual_inversion/preprocess.py | 32 ++++++++++++++++++------- 1 file changed, 24 insertions(+), 8 deletions(-) diff --git a/modules/textual_inversion/preprocess.py b/modules/textual_inversion/preprocess.py index 886cf0c3..5c43fe13 100644 --- a/modules/textual_inversion/preprocess.py +++ b/modules/textual_inversion/preprocess.py @@ -48,7 +48,7 @@ def preprocess_work(process_src, process_dst, process_width, process_height, pro shared.state.textinfo = "Preprocessing..." shared.state.job_count = len(files) - def save_pic_with_caption(image, index): + def save_pic_with_caption(image, index, existing_caption=None): caption = "" if process_caption: @@ -66,17 +66,26 @@ def preprocess_work(process_src, process_dst, process_width, process_height, pro basename = f"{index:05}-{subindex[0]}-{filename_part}" image.save(os.path.join(dst, f"{basename}.png")) + if preprocess_txt_action == 'prepend' and existing_caption: + caption = existing_caption + ' ' + caption + elif preprocess_txt_action == 'append' and existing_caption: + caption = caption + ' ' + existing_caption + elif preprocess_txt_action == 'copy' and existing_caption: + caption = existing_caption + + caption = caption.strip() + if len(caption) > 0: with open(os.path.join(dst, f"{basename}.txt"), "w", encoding="utf8") as file: file.write(caption) subindex[0] += 1 - def save_pic(image, index): + def save_pic(image, index, existing_caption=None): save_pic_with_caption(image, index) if process_flip: - save_pic_with_caption(ImageOps.mirror(image), index) + save_pic_with_caption(ImageOps.mirror(image), index, existing_caption=existing_caption) for index, imagefile in enumerate(tqdm.tqdm(files)): subindex = [0] @@ -86,6 +95,13 @@ def preprocess_work(process_src, process_dst, process_width, process_height, pro except Exception: continue + existing_caption = None + + try: + existing_caption = open(os.path.splitext(filename)[0] + '.txt', 'r').read() + except Exception as e: + print(e) + if shared.state.interrupted: break @@ -97,20 +113,20 @@ def preprocess_work(process_src, process_dst, process_width, process_height, pro img = img.resize((width, height * img.height // img.width)) top = img.crop((0, 0, width, height)) - save_pic(top, index) + save_pic(top, index, existing_caption=existing_caption) bot = img.crop((0, img.height - height, width, img.height)) - save_pic(bot, index) + save_pic(bot, index, existing_caption=existing_caption) elif process_split and is_wide: img = img.resize((width * img.width // img.height, height)) left = img.crop((0, 0, width, height)) - save_pic(left, index) + save_pic(left, index, existing_caption=existing_caption) right = img.crop((img.width - width, 0, img.width, height)) - save_pic(right, index) + save_pic(right, index, existing_caption=existing_caption) else: img = images.resize_image(1, img, width, height) - save_pic(img, index) + save_pic(img, index, existing_caption=existing_caption) shared.state.nextjob() From ab353b141df8eee042b0964bcb645015dabf3459 Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Thu, 20 Oct 2022 00:48:07 +0100 Subject: [PATCH 11/14] link existing txt option --- modules/ui.py | 2 ++ 1 file changed, 2 insertions(+) diff --git a/modules/ui.py b/modules/ui.py index 322c082b..7f52ac0c 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -1234,6 +1234,7 @@ def create_ui(wrap_gradio_gpu_call): process_dst = gr.Textbox(label='Destination directory') process_width = gr.Slider(minimum=64, maximum=2048, step=64, label="Width", value=512) process_height = gr.Slider(minimum=64, maximum=2048, step=64, label="Height", value=512) + preprocess_txt_action = gr.Dropdown(label='Existing Caption txt Action', choices=['ignore', 'copy', 'prepend', 'append']) with gr.Row(): process_flip = gr.Checkbox(label='Create flipped copies') @@ -1326,6 +1327,7 @@ def create_ui(wrap_gradio_gpu_call): process_dst, process_width, process_height, + preprocess_txt_action, process_flip, process_split, process_caption, From 9b65c4ecf4f8eb6187ee721918adebe68e9bc631 Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Thu, 20 Oct 2022 00:49:23 +0100 Subject: [PATCH 12/14] pass preprocess_txt_action param --- modules/textual_inversion/preprocess.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/modules/textual_inversion/preprocess.py b/modules/textual_inversion/preprocess.py index 5c43fe13..3713bc89 100644 --- a/modules/textual_inversion/preprocess.py +++ b/modules/textual_inversion/preprocess.py @@ -11,7 +11,7 @@ if cmd_opts.deepdanbooru: import modules.deepbooru as deepbooru -def preprocess(process_src, process_dst, process_width, process_height, process_flip, process_split, process_caption, process_caption_deepbooru=False): +def preprocess(process_src, process_dst, process_width, process_height, preprocess_txt_action, process_flip, process_split, process_caption, process_caption_deepbooru=False): try: if process_caption: shared.interrogator.load() @@ -21,7 +21,7 @@ def preprocess(process_src, process_dst, process_width, process_height, process_ db_opts[deepbooru.OPT_INCLUDE_RANKS] = False deepbooru.create_deepbooru_process(opts.interrogate_deepbooru_score_threshold, db_opts) - preprocess_work(process_src, process_dst, process_width, process_height, process_flip, process_split, process_caption, process_caption_deepbooru) + preprocess_work(process_src, process_dst, process_width, process_height, preprocess_txt_action, process_flip, process_split, process_caption, process_caption_deepbooru) finally: @@ -33,7 +33,7 @@ def preprocess(process_src, process_dst, process_width, process_height, process_ -def preprocess_work(process_src, process_dst, process_width, process_height, process_flip, process_split, process_caption, process_caption_deepbooru=False): +def preprocess_work(process_src, process_dst, process_width, process_height, preprocess_txt_action, process_flip, process_split, process_caption, process_caption_deepbooru=False): width = process_width height = process_height src = os.path.abspath(process_src) From 55d8c6cce6d3aef848b9f194adad2ce53064d8b7 Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Thu, 20 Oct 2022 00:53:29 +0100 Subject: [PATCH 13/14] default to ignore existing captions --- modules/ui.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/ui.py b/modules/ui.py index 7f52ac0c..bd5f1b05 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -1234,7 +1234,7 @@ def create_ui(wrap_gradio_gpu_call): process_dst = gr.Textbox(label='Destination directory') process_width = gr.Slider(minimum=64, maximum=2048, step=64, label="Width", value=512) process_height = gr.Slider(minimum=64, maximum=2048, step=64, label="Height", value=512) - preprocess_txt_action = gr.Dropdown(label='Existing Caption txt Action', choices=['ignore', 'copy', 'prepend', 'append']) + preprocess_txt_action = gr.Dropdown(label='Existing Caption txt Action', value="ignore", choices=["ignore", "copy", "prepend", "append"]) with gr.Row(): process_flip = gr.Checkbox(label='Create flipped copies') From 858462f719c22ca9f24b94a41699653c34b5f4fb Mon Sep 17 00:00:00 2001 From: DepFA <35278260+dfaker@users.noreply.github.com> Date: Thu, 20 Oct 2022 02:57:18 +0100 Subject: [PATCH 14/14] do caption copy for both flips --- modules/textual_inversion/preprocess.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/modules/textual_inversion/preprocess.py b/modules/textual_inversion/preprocess.py index 3713bc89..6bba3852 100644 --- a/modules/textual_inversion/preprocess.py +++ b/modules/textual_inversion/preprocess.py @@ -82,7 +82,7 @@ def preprocess_work(process_src, process_dst, process_width, process_height, pre subindex[0] += 1 def save_pic(image, index, existing_caption=None): - save_pic_with_caption(image, index) + save_pic_with_caption(image, index, existing_caption=existing_caption) if process_flip: save_pic_with_caption(ImageOps.mirror(image), index, existing_caption=existing_caption)