From da4ab2707b4cb0611cf181ba248a271d1937433e Mon Sep 17 00:00:00 2001 From: C43H66N12O12S2 <36072735+C43H66N12O12S2@users.noreply.github.com> Date: Fri, 7 Oct 2022 05:23:06 +0300 Subject: Update shared.py --- modules/shared.py | 1 + 1 file changed, 1 insertion(+) (limited to 'modules/shared.py') diff --git a/modules/shared.py b/modules/shared.py index 25bb6e6c..8cc3b2fe 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -43,6 +43,7 @@ parser.add_argument("--realesrgan-models-path", type=str, help="Path to director parser.add_argument("--scunet-models-path", type=str, help="Path to directory with ScuNET model file(s).", default=os.path.join(models_path, 'ScuNET')) parser.add_argument("--swinir-models-path", type=str, help="Path to directory with SwinIR model file(s).", default=os.path.join(models_path, 'SwinIR')) parser.add_argument("--ldsr-models-path", type=str, help="Path to directory with LDSR model file(s).", default=os.path.join(models_path, 'LDSR')) +parser.add_argument("--disable-opt-xformers-attention", action='store_true', help="force-disables xformers attention optimization") parser.add_argument("--opt-split-attention", action='store_true', help="force-enables cross-attention layer optimization. By default, it's on for torch.cuda and off for other torch devices.") parser.add_argument("--disable-opt-split-attention", action='store_true', help="force-disables cross-attention layer optimization") parser.add_argument("--opt-split-attention-v1", action='store_true', help="enable older version of split attention optimization that does not consume all the VRAM it can find") -- cgit v1.2.3 From 706d5944a075a6523ea7f00165d630efc085ca22 Mon Sep 17 00:00:00 2001 From: AUTOMATIC <16777216c@gmail.com> Date: Sat, 8 Oct 2022 13:38:57 +0300 Subject: let user choose his own prompt token count limit --- modules/processing.py | 6 ++++++ modules/sd_hijack.py | 13 +++++++------ modules/shared.py | 5 +++-- 3 files changed, 16 insertions(+), 8 deletions(-) (limited to 'modules/shared.py') diff --git a/modules/processing.py b/modules/processing.py index f773a30e..d814d5ac 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -123,6 +123,7 @@ class Processed: self.index_of_first_image = index_of_first_image self.styles = p.styles self.job_timestamp = state.job_timestamp + self.max_prompt_tokens = opts.max_prompt_tokens self.eta = p.eta self.ddim_discretize = p.ddim_discretize @@ -141,6 +142,7 @@ class Processed: self.all_subseeds = all_subseeds or [self.subseed] self.infotexts = infotexts or [info] + def js(self): obj = { "prompt": self.prompt, @@ -169,6 +171,7 @@ class Processed: "infotexts": self.infotexts, "styles": self.styles, "job_timestamp": self.job_timestamp, + "max_prompt_tokens": self.max_prompt_tokens, } return json.dumps(obj) @@ -266,6 +269,8 @@ def fix_seed(p): def create_infotext(p, all_prompts, all_seeds, all_subseeds, comments, iteration=0, position_in_batch=0): index = position_in_batch + iteration * p.batch_size + max_tokens = getattr(p, 'max_prompt_tokens', opts.max_prompt_tokens) + generation_params = { "Steps": p.steps, "Sampler": sd_samplers.samplers[p.sampler_index].name, @@ -281,6 +286,7 @@ def create_infotext(p, all_prompts, all_seeds, all_subseeds, comments, iteration "Seed resize from": (None if p.seed_resize_from_w == 0 or p.seed_resize_from_h == 0 else f"{p.seed_resize_from_w}x{p.seed_resize_from_h}"), "Denoising strength": getattr(p, 'denoising_strength', None), "Eta": (None if p.sampler is None or p.sampler.eta == p.sampler.default_eta else p.sampler.eta), + "Max tokens": (None if max_tokens == shared.vanilla_max_prompt_tokens else max_tokens) } generation_params.update(p.extra_generation_params) diff --git a/modules/sd_hijack.py b/modules/sd_hijack.py index d68f89cc..340329c0 100644 --- a/modules/sd_hijack.py +++ b/modules/sd_hijack.py @@ -18,7 +18,6 @@ attention_CrossAttention_forward = ldm.modules.attention.CrossAttention.forward diffusionmodules_model_nonlinearity = ldm.modules.diffusionmodules.model.nonlinearity diffusionmodules_model_AttnBlock_forward = ldm.modules.diffusionmodules.model.AttnBlock.forward - def apply_optimizations(): undo_optimizations() @@ -83,7 +82,7 @@ class StableDiffusionModelHijack: layer.padding_mode = 'circular' if enable else 'zeros' def tokenize(self, text): - max_length = self.clip.max_length - 2 + max_length = opts.max_prompt_tokens - 2 _, remade_batch_tokens, _, _, _, token_count = self.clip.process_text([text]) return remade_batch_tokens[0], token_count, max_length @@ -94,7 +93,6 @@ class FrozenCLIPEmbedderWithCustomWords(torch.nn.Module): self.wrapped = wrapped self.hijack: StableDiffusionModelHijack = hijack self.tokenizer = wrapped.tokenizer - self.max_length = wrapped.max_length self.token_mults = {} tokens_with_parens = [(k, v) for k, v in self.tokenizer.get_vocab().items() if '(' in k or ')' in k or '[' in k or ']' in k] @@ -116,7 +114,7 @@ class FrozenCLIPEmbedderWithCustomWords(torch.nn.Module): def tokenize_line(self, line, used_custom_terms, hijack_comments): id_start = self.wrapped.tokenizer.bos_token_id id_end = self.wrapped.tokenizer.eos_token_id - maxlen = self.wrapped.max_length + maxlen = opts.max_prompt_tokens if opts.enable_emphasis: parsed = prompt_parser.parse_prompt_attention(line) @@ -191,7 +189,7 @@ class FrozenCLIPEmbedderWithCustomWords(torch.nn.Module): def process_text_old(self, text): id_start = self.wrapped.tokenizer.bos_token_id id_end = self.wrapped.tokenizer.eos_token_id - maxlen = self.wrapped.max_length + maxlen = self.wrapped.max_length # you get to stay at 77 used_custom_terms = [] remade_batch_tokens = [] overflowing_words = [] @@ -268,8 +266,11 @@ class FrozenCLIPEmbedderWithCustomWords(torch.nn.Module): if len(used_custom_terms) > 0: self.hijack.comments.append("Used embeddings: " + ", ".join([f'{word} [{checksum}]' for word, checksum in used_custom_terms])) + position_ids_array = [min(x, 75) for x in range(len(remade_batch_tokens[0])-1)] + [76] + position_ids = torch.asarray(position_ids_array, device=devices.device).expand((1, -1)) + tokens = torch.asarray(remade_batch_tokens).to(device) - outputs = self.wrapped.transformer(input_ids=tokens) + outputs = self.wrapped.transformer(input_ids=tokens, position_ids=position_ids) z = outputs.last_hidden_state # restoring original mean is likely not correct, but it seems to work well to prevent artifacts that happen otherwise diff --git a/modules/shared.py b/modules/shared.py index 879d8424..864e772c 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -118,8 +118,8 @@ prompt_styles = modules.styles.StyleDatabase(styles_filename) interrogator = modules.interrogate.InterrogateModels("interrogate") face_restorers = [] -# This was moved to webui.py with the other model "setup" calls. -# modules.sd_models.list_models() + +vanilla_max_prompt_tokens = 77 def realesrgan_models_names(): @@ -221,6 +221,7 @@ options_templates.update(options_section(('sd', "Stable Diffusion"), { "use_old_emphasis_implementation": OptionInfo(False, "Use old emphasis implementation. Can be useful to reproduce old seeds."), "enable_batch_seeds": OptionInfo(True, "Make K-diffusion samplers produce same images in a batch as when making a single image"), "filter_nsfw": OptionInfo(False, "Filter NSFW content"), + "max_prompt_tokens": OptionInfo(vanilla_max_prompt_tokens, f"Max prompt token count. Two tokens are reserved for for start and end. Default is {vanilla_max_prompt_tokens}. Setting this to a different value will result in different pictures for same seed.", gr.Number, {"precision": 0}), "random_artist_categories": OptionInfo([], "Allowed categories for random artists selection when using the Roll button", gr.CheckboxGroup, {"choices": artist_db.categories()}), })) -- cgit v1.2.3 From 786d9f63aaa4515df82eb2cf357ea92f3dae1e29 Mon Sep 17 00:00:00 2001 From: Trung Ngo Date: Tue, 4 Oct 2022 22:56:30 -0500 Subject: Add button to skip the current iteration --- javascript/hints.js | 1 + javascript/progressbar.js | 20 ++++++++++++++------ modules/img2img.py | 4 ++++ modules/processing.py | 4 ++++ modules/shared.py | 5 +++++ modules/ui.py | 8 ++++++++ style.css | 14 ++++++++++++-- webui.py | 1 + 8 files changed, 49 insertions(+), 8 deletions(-) (limited to 'modules/shared.py') diff --git a/javascript/hints.js b/javascript/hints.js index 8adcd983..8e352e94 100644 --- a/javascript/hints.js +++ b/javascript/hints.js @@ -35,6 +35,7 @@ titles = { "Denoising strength": "Determines how little respect the algorithm should have for image's content. At 0, nothing will change, and at 1 you'll get an unrelated image. With values below 1.0, processing will take less steps than the Sampling Steps slider specifies.", "Denoising strength change factor": "In loopback mode, on each loop the denoising strength is multiplied by this value. <1 means decreasing variety so your sequence will converge on a fixed picture. >1 means increasing variety so your sequence will become more and more chaotic.", + "Skip": "Stop processing current image and continue processing.", "Interrupt": "Stop processing images and return any results accumulated so far.", "Save": "Write image to a directory (default - log/images) and generation parameters into csv file.", diff --git a/javascript/progressbar.js b/javascript/progressbar.js index f9e9290e..4395a215 100644 --- a/javascript/progressbar.js +++ b/javascript/progressbar.js @@ -1,8 +1,9 @@ // code related to showing and updating progressbar shown as the image is being made global_progressbars = {} -function check_progressbar(id_part, id_progressbar, id_progressbar_span, id_interrupt, id_preview, id_gallery){ +function check_progressbar(id_part, id_progressbar, id_progressbar_span, id_skip, id_interrupt, id_preview, id_gallery){ var progressbar = gradioApp().getElementById(id_progressbar) + var skip = id_skip ? gradioApp().getElementById(id_skip) : null var interrupt = gradioApp().getElementById(id_interrupt) if(opts.show_progress_in_title && progressbar && progressbar.offsetParent){ @@ -32,30 +33,37 @@ function check_progressbar(id_part, id_progressbar, id_progressbar_span, id_inte var progressDiv = gradioApp().querySelectorAll('#' + id_progressbar_span).length > 0; if(!progressDiv){ + if (skip) { + skip.style.display = "none" + } interrupt.style.display = "none" } } - window.setTimeout(function(){ requestMoreProgress(id_part, id_progressbar_span, id_interrupt) }, 500) + window.setTimeout(function() { requestMoreProgress(id_part, id_progressbar_span, id_skip, id_interrupt) }, 500) }); mutationObserver.observe( progressbar, { childList:true, subtree:true }) } } onUiUpdate(function(){ - check_progressbar('txt2img', 'txt2img_progressbar', 'txt2img_progress_span', 'txt2img_interrupt', 'txt2img_preview', 'txt2img_gallery') - check_progressbar('img2img', 'img2img_progressbar', 'img2img_progress_span', 'img2img_interrupt', 'img2img_preview', 'img2img_gallery') - check_progressbar('ti', 'ti_progressbar', 'ti_progress_span', 'ti_interrupt', 'ti_preview', 'ti_gallery') + check_progressbar('txt2img', 'txt2img_progressbar', 'txt2img_progress_span', 'txt2img_skip', 'txt2img_interrupt', 'txt2img_preview', 'txt2img_gallery') + check_progressbar('img2img', 'img2img_progressbar', 'img2img_progress_span', 'img2img_skip', 'img2img_interrupt', 'img2img_preview', 'img2img_gallery') + check_progressbar('ti', 'ti_progressbar', 'ti_progress_span', '', 'ti_interrupt', 'ti_preview', 'ti_gallery') }) -function requestMoreProgress(id_part, id_progressbar_span, id_interrupt){ +function requestMoreProgress(id_part, id_progressbar_span, id_skip, id_interrupt){ btn = gradioApp().getElementById(id_part+"_check_progress"); if(btn==null) return; btn.click(); var progressDiv = gradioApp().querySelectorAll('#' + id_progressbar_span).length > 0; + var skip = id_skip ? gradioApp().getElementById(id_skip) : null var interrupt = gradioApp().getElementById(id_interrupt) if(progressDiv && interrupt){ + if (skip) { + skip.style.display = "block" + } interrupt.style.display = "block" } } diff --git a/modules/img2img.py b/modules/img2img.py index da212d72..e60b7e0f 100644 --- a/modules/img2img.py +++ b/modules/img2img.py @@ -32,6 +32,10 @@ def process_batch(p, input_dir, output_dir, args): for i, image in enumerate(images): state.job = f"{i+1} out of {len(images)}" + if state.skipped: + state.skipped = False + state.interrupted = False + continue if state.interrupted: break diff --git a/modules/processing.py b/modules/processing.py index d814d5ac..6805039c 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -355,6 +355,10 @@ def process_images(p: StableDiffusionProcessing) -> Processed: state.job_count = p.n_iter for n in range(p.n_iter): + if state.skipped: + state.skipped = False + state.interrupted = False + if state.interrupted: break diff --git a/modules/shared.py b/modules/shared.py index 864e772c..7f802bd9 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -84,6 +84,7 @@ def selected_hypernetwork(): class State: + skipped = False interrupted = False job = "" job_no = 0 @@ -96,6 +97,10 @@ class State: current_image_sampling_step = 0 textinfo = None + def skip(self): + self.skipped = True + self.interrupted = True + def interrupt(self): self.interrupted = True diff --git a/modules/ui.py b/modules/ui.py index 4f18126f..e3e62fdd 100644 --- a/modules/ui.py +++ b/modules/ui.py @@ -191,6 +191,7 @@ def wrap_gradio_call(func, extra_outputs=None): # last item is always HTML res[-1] += f"

Time taken: {elapsed_text}

{vram_html}
" + shared.state.skipped = False shared.state.interrupted = False shared.state.job_count = 0 @@ -411,9 +412,16 @@ def create_toprow(is_img2img): with gr.Column(scale=1): with gr.Row(): + skip = gr.Button('Skip', elem_id=f"{id_part}_skip") interrupt = gr.Button('Interrupt', elem_id=f"{id_part}_interrupt") submit = gr.Button('Generate', elem_id=f"{id_part}_generate", variant='primary') + skip.click( + fn=lambda: shared.state.skip(), + inputs=[], + outputs=[], + ) + interrupt.click( fn=lambda: shared.state.interrupt(), inputs=[], diff --git a/style.css b/style.css index 50c5e557..6904fc50 100644 --- a/style.css +++ b/style.css @@ -393,10 +393,20 @@ input[type="range"]{ #txt2img_interrupt, #img2img_interrupt{ position: absolute; - width: 100%; + width: 50%; height: 72px; background: #b4c0cc; - border-radius: 8px; + border-radius: 0px; + display: none; +} + +#txt2img_skip, #img2img_skip{ + position: absolute; + width: 50%; + right: 0px; + height: 72px; + background: #b4c0cc; + border-radius: 0px; display: none; } diff --git a/webui.py b/webui.py index 480360fe..3b4cf5e9 100644 --- a/webui.py +++ b/webui.py @@ -58,6 +58,7 @@ def wrap_gradio_gpu_call(func, extra_outputs=None): shared.state.current_latent = None shared.state.current_image = None shared.state.current_image_sampling_step = 0 + shared.state.skipped = False shared.state.interrupted = False shared.state.textinfo = None -- cgit v1.2.3 From 00117a07efbbe8482add12262a179326541467de Mon Sep 17 00:00:00 2001 From: Trung Ngo Date: Sat, 8 Oct 2022 05:33:21 -0500 Subject: check specifically for skipped --- modules/img2img.py | 2 -- modules/processing.py | 3 +-- modules/sd_samplers.py | 4 ++-- modules/shared.py | 1 - 4 files changed, 3 insertions(+), 7 deletions(-) (limited to 'modules/shared.py') diff --git a/modules/img2img.py b/modules/img2img.py index e60b7e0f..24126774 100644 --- a/modules/img2img.py +++ b/modules/img2img.py @@ -34,8 +34,6 @@ def process_batch(p, input_dir, output_dir, args): state.job = f"{i+1} out of {len(images)}" if state.skipped: state.skipped = False - state.interrupted = False - continue if state.interrupted: break diff --git a/modules/processing.py b/modules/processing.py index 6805039c..3657fe69 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -357,7 +357,6 @@ def process_images(p: StableDiffusionProcessing) -> Processed: for n in range(p.n_iter): if state.skipped: state.skipped = False - state.interrupted = False if state.interrupted: break @@ -385,7 +384,7 @@ def process_images(p: StableDiffusionProcessing) -> Processed: with devices.autocast(): samples_ddim = p.sample(conditioning=c, unconditional_conditioning=uc, seeds=seeds, subseeds=subseeds, subseed_strength=p.subseed_strength) - if state.interrupted: + if state.interrupted or state.skipped: # if we are interruped, sample returns just noise # use the image collected previously in sampler loop diff --git a/modules/sd_samplers.py b/modules/sd_samplers.py index df17e93c..13a8b322 100644 --- a/modules/sd_samplers.py +++ b/modules/sd_samplers.py @@ -106,7 +106,7 @@ def extended_tdqm(sequence, *args, desc=None, **kwargs): seq = sequence if cmd_opts.disable_console_progressbars else tqdm.tqdm(sequence, *args, desc=state.job, file=shared.progress_print_out, **kwargs) for x in seq: - if state.interrupted: + if state.interrupted or state.skipped: break yield x @@ -254,7 +254,7 @@ def extended_trange(sampler, count, *args, **kwargs): seq = range(count) if cmd_opts.disable_console_progressbars else tqdm.trange(count, *args, desc=state.job, file=shared.progress_print_out, **kwargs) for x in seq: - if state.interrupted: + if state.interrupted or state.skipped: break if sampler.stop_at is not None and x > sampler.stop_at: diff --git a/modules/shared.py b/modules/shared.py index 7f802bd9..ca462628 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -99,7 +99,6 @@ class State: def skip(self): self.skipped = True - self.interrupted = True def interrupt(self): self.interrupted = True -- cgit v1.2.3 From 4999eb2ef9b30e8c42ca7e4a94d4bbffe4d1f015 Mon Sep 17 00:00:00 2001 From: AUTOMATIC <16777216c@gmail.com> Date: Sat, 8 Oct 2022 14:25:47 +0300 Subject: do not let user choose his own prompt token count limit --- README.md | 1 + modules/processing.py | 5 ----- modules/sd_hijack.py | 25 ++++++++++++------------- modules/shared.py | 3 --- 4 files changed, 13 insertions(+), 21 deletions(-) (limited to 'modules/shared.py') diff --git a/README.md b/README.md index d6e1d50b..ef9b5e31 100644 --- a/README.md +++ b/README.md @@ -65,6 +65,7 @@ Check the [custom scripts](https://github.com/AUTOMATIC1111/stable-diffusion-web - [Composable-Diffusion](https://energy-based-model.github.io/Compositional-Visual-Generation-with-Composable-Diffusion-Models/), a way to use multiple prompts at once - separate prompts using uppercase `AND` - also supports weights for prompts: `a cat :1.2 AND a dog AND a penguin :2.2` +- No token limit for prompts (original stable diffusion lets you use up to 75 tokens) ## Installation and Running Make sure the required [dependencies](https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Dependencies) are met and follow the instructions available for both [NVidia](https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Install-and-Run-on-NVidia-GPUs) (recommended) and [AMD](https://github.com/AUTOMATIC1111/stable-diffusion-webui/wiki/Install-and-Run-on-AMD-GPUs) GPUs. diff --git a/modules/processing.py b/modules/processing.py index 3657fe69..d5162ddc 100644 --- a/modules/processing.py +++ b/modules/processing.py @@ -123,7 +123,6 @@ class Processed: self.index_of_first_image = index_of_first_image self.styles = p.styles self.job_timestamp = state.job_timestamp - self.max_prompt_tokens = opts.max_prompt_tokens self.eta = p.eta self.ddim_discretize = p.ddim_discretize @@ -171,7 +170,6 @@ class Processed: "infotexts": self.infotexts, "styles": self.styles, "job_timestamp": self.job_timestamp, - "max_prompt_tokens": self.max_prompt_tokens, } return json.dumps(obj) @@ -269,8 +267,6 @@ def fix_seed(p): def create_infotext(p, all_prompts, all_seeds, all_subseeds, comments, iteration=0, position_in_batch=0): index = position_in_batch + iteration * p.batch_size - max_tokens = getattr(p, 'max_prompt_tokens', opts.max_prompt_tokens) - generation_params = { "Steps": p.steps, "Sampler": sd_samplers.samplers[p.sampler_index].name, @@ -286,7 +282,6 @@ def create_infotext(p, all_prompts, all_seeds, all_subseeds, comments, iteration "Seed resize from": (None if p.seed_resize_from_w == 0 or p.seed_resize_from_h == 0 else f"{p.seed_resize_from_w}x{p.seed_resize_from_h}"), "Denoising strength": getattr(p, 'denoising_strength', None), "Eta": (None if p.sampler is None or p.sampler.eta == p.sampler.default_eta else p.sampler.eta), - "Max tokens": (None if max_tokens == shared.vanilla_max_prompt_tokens else max_tokens) } generation_params.update(p.extra_generation_params) diff --git a/modules/sd_hijack.py b/modules/sd_hijack.py index 340329c0..2c1332c9 100644 --- a/modules/sd_hijack.py +++ b/modules/sd_hijack.py @@ -36,6 +36,13 @@ def undo_optimizations(): ldm.modules.diffusionmodules.model.AttnBlock.forward = diffusionmodules_model_AttnBlock_forward +def get_target_prompt_token_count(token_count): + if token_count < 75: + return 75 + + return math.ceil(token_count / 10) * 10 + + class StableDiffusionModelHijack: fixes = None comments = [] @@ -84,7 +91,7 @@ class StableDiffusionModelHijack: def tokenize(self, text): max_length = opts.max_prompt_tokens - 2 _, remade_batch_tokens, _, _, _, token_count = self.clip.process_text([text]) - return remade_batch_tokens[0], token_count, max_length + return remade_batch_tokens[0], token_count, get_target_prompt_token_count(token_count) class FrozenCLIPEmbedderWithCustomWords(torch.nn.Module): @@ -114,7 +121,6 @@ class FrozenCLIPEmbedderWithCustomWords(torch.nn.Module): def tokenize_line(self, line, used_custom_terms, hijack_comments): id_start = self.wrapped.tokenizer.bos_token_id id_end = self.wrapped.tokenizer.eos_token_id - maxlen = opts.max_prompt_tokens if opts.enable_emphasis: parsed = prompt_parser.parse_prompt_attention(line) @@ -146,19 +152,12 @@ class FrozenCLIPEmbedderWithCustomWords(torch.nn.Module): used_custom_terms.append((embedding.name, embedding.checksum())) i += embedding_length_in_tokens - if len(remade_tokens) > maxlen - 2: - vocab = {v: k for k, v in self.wrapped.tokenizer.get_vocab().items()} - ovf = remade_tokens[maxlen - 2:] - overflowing_words = [vocab.get(int(x), "") for x in ovf] - overflowing_text = self.wrapped.tokenizer.convert_tokens_to_string(''.join(overflowing_words)) - hijack_comments.append(f"Warning: too many input tokens; some ({len(overflowing_words)}) have been truncated:\n{overflowing_text}\n") - token_count = len(remade_tokens) - remade_tokens = remade_tokens + [id_end] * (maxlen - 2 - len(remade_tokens)) - remade_tokens = [id_start] + remade_tokens[0:maxlen - 2] + [id_end] + prompt_target_length = get_target_prompt_token_count(token_count) + tokens_to_add = prompt_target_length - len(remade_tokens) + 1 - multipliers = multipliers + [1.0] * (maxlen - 2 - len(multipliers)) - multipliers = [1.0] + multipliers[0:maxlen - 2] + [1.0] + remade_tokens = [id_start] + remade_tokens + [id_end] * tokens_to_add + multipliers = [1.0] + multipliers + [1.0] * tokens_to_add return remade_tokens, fixes, multipliers, token_count diff --git a/modules/shared.py b/modules/shared.py index ca462628..475d7e52 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -123,8 +123,6 @@ interrogator = modules.interrogate.InterrogateModels("interrogate") face_restorers = [] -vanilla_max_prompt_tokens = 77 - def realesrgan_models_names(): import modules.realesrgan_model @@ -225,7 +223,6 @@ options_templates.update(options_section(('sd', "Stable Diffusion"), { "use_old_emphasis_implementation": OptionInfo(False, "Use old emphasis implementation. Can be useful to reproduce old seeds."), "enable_batch_seeds": OptionInfo(True, "Make K-diffusion samplers produce same images in a batch as when making a single image"), "filter_nsfw": OptionInfo(False, "Filter NSFW content"), - "max_prompt_tokens": OptionInfo(vanilla_max_prompt_tokens, f"Max prompt token count. Two tokens are reserved for for start and end. Default is {vanilla_max_prompt_tokens}. Setting this to a different value will result in different pictures for same seed.", gr.Number, {"precision": 0}), "random_artist_categories": OptionInfo([], "Allowed categories for random artists selection when using the Roll button", gr.CheckboxGroup, {"choices": artist_db.categories()}), })) -- cgit v1.2.3 From ddfa9a97865c732193023a71521c5b7b53d8571b Mon Sep 17 00:00:00 2001 From: C43H66N12O12S2 <36072735+C43H66N12O12S2@users.noreply.github.com> Date: Sat, 8 Oct 2022 16:20:41 +0300 Subject: add xformers_available shared variable --- modules/shared.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'modules/shared.py') diff --git a/modules/shared.py b/modules/shared.py index 8cc3b2fe..6ed4b802 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -74,7 +74,7 @@ device = devices.device batch_cond_uncond = cmd_opts.always_batch_cond_uncond or not (cmd_opts.lowvram or cmd_opts.medvram) parallel_processing_allowed = not cmd_opts.lowvram and not cmd_opts.medvram - +xformers_available = False config_filename = cmd_opts.ui_settings_file -- cgit v1.2.3 From dc1117233ef8f9b25ff1ac40b158f20b70ba2fcb Mon Sep 17 00:00:00 2001 From: AUTOMATIC <16777216c@gmail.com> Date: Sat, 8 Oct 2022 17:02:18 +0300 Subject: simplify xfrmers options: --xformers to enable and that's it --- launch.py | 2 +- modules/sd_hijack.py | 2 +- modules/sd_hijack_optimizations.py | 20 +++++++++++++------- modules/shared.py | 2 +- 4 files changed, 16 insertions(+), 10 deletions(-) (limited to 'modules/shared.py') diff --git a/launch.py b/launch.py index a592e1ba..61f62096 100644 --- a/launch.py +++ b/launch.py @@ -32,7 +32,7 @@ def extract_arg(args, name): args, skip_torch_cuda_test = extract_arg(args, '--skip-torch-cuda-test') -args, xformers = extract_arg(args, '--xformers') +xformers = '--xformers' in args def repo_dir(name): diff --git a/modules/sd_hijack.py b/modules/sd_hijack.py index 5d93f7f6..91e98c16 100644 --- a/modules/sd_hijack.py +++ b/modules/sd_hijack.py @@ -22,7 +22,7 @@ def apply_optimizations(): undo_optimizations() ldm.modules.diffusionmodules.model.nonlinearity = silu - if not cmd_opts.disable_opt_xformers_attention and not (cmd_opts.opt_split_attention or torch.version.hip) and shared.xformers_available: + if cmd_opts.xformers and shared.xformers_available and not torch.version.hip: ldm.modules.attention.CrossAttention.forward = sd_hijack_optimizations.xformers_attention_forward ldm.modules.diffusionmodules.model.AttnBlock.forward = sd_hijack_optimizations.xformers_attnblock_forward elif cmd_opts.opt_split_attention_v1: diff --git a/modules/sd_hijack_optimizations.py b/modules/sd_hijack_optimizations.py index 05023b6f..d23d733b 100644 --- a/modules/sd_hijack_optimizations.py +++ b/modules/sd_hijack_optimizations.py @@ -1,4 +1,7 @@ import math +import sys +import traceback + import torch from torch import einsum @@ -7,13 +10,16 @@ from einops import rearrange from modules import shared -try: - import xformers.ops - import functorch - xformers._is_functorch_available = True - shared.xformers_available = True -except Exception: - print('Cannot find xformers, defaulting to split attention. Try adding --xformers commandline argument to your webui-user file if you wish to install it.') +if shared.cmd_opts.xformers: + try: + import xformers.ops + import functorch + xformers._is_functorch_available = True + shared.xformers_available = True + except Exception: + print("Cannot import xformers", file=sys.stderr) + print(traceback.format_exc(), file=sys.stderr) + # see https://github.com/basujindal/stable-diffusion/pull/117 for discussion def split_cross_attention_forward_v1(self, x, context=None, mask=None): diff --git a/modules/shared.py b/modules/shared.py index d68df751..02cb2722 100644 --- a/modules/shared.py +++ b/modules/shared.py @@ -43,7 +43,7 @@ parser.add_argument("--realesrgan-models-path", type=str, help="Path to director parser.add_argument("--scunet-models-path", type=str, help="Path to directory with ScuNET model file(s).", default=os.path.join(models_path, 'ScuNET')) parser.add_argument("--swinir-models-path", type=str, help="Path to directory with SwinIR model file(s).", default=os.path.join(models_path, 'SwinIR')) parser.add_argument("--ldsr-models-path", type=str, help="Path to directory with LDSR model file(s).", default=os.path.join(models_path, 'LDSR')) -parser.add_argument("--disable-opt-xformers-attention", action='store_true', help="force-disables xformers attention optimization") +parser.add_argument("--xformers", action='store_true', help="enable xformers for cross attention layers") parser.add_argument("--opt-split-attention", action='store_true', help="force-enables cross-attention layer optimization. By default, it's on for torch.cuda and off for other torch devices.") parser.add_argument("--disable-opt-split-attention", action='store_true', help="force-disables cross-attention layer optimization") parser.add_argument("--opt-split-attention-v1", action='store_true', help="enable older version of split attention optimization that does not consume all the VRAM it can find") -- cgit v1.2.3