diff --git a/launch.py b/launch.py index 68e08114d..94bba5ca0 100644 --- a/launch.py +++ b/launch.py @@ -238,12 +238,14 @@ def prepare_environment(): k_diffusion_repo = os.environ.get('K_DIFFUSION_REPO', 'https://github.com/crowsonkb/k-diffusion.git') codeformer_repo = os.environ.get('CODEFORMER_REPO', 'https://github.com/sczhou/CodeFormer.git') blip_repo = os.environ.get('BLIP_REPO', 'https://github.com/salesforce/BLIP.git') + tomesd_repo = os.environ.get('TOMESD_REPO', 'https://github.com/dbolya/tomesd.git') stable_diffusion_commit_hash = os.environ.get('STABLE_DIFFUSION_COMMIT_HASH', "cf1d67a6fd5ea1aa600c4df58e5b47da45f6bdbf") taming_transformers_commit_hash = os.environ.get('TAMING_TRANSFORMERS_COMMIT_HASH', "24268930bf1dce879235a7fddd0b2355b84d7ea6") k_diffusion_commit_hash = os.environ.get('K_DIFFUSION_COMMIT_HASH', "5b3af030dd83e0297272d861c19477735d0317ec") codeformer_commit_hash = os.environ.get('CODEFORMER_COMMIT_HASH', "c5b4593074ba6214284d6acd5f1719b6c5d739af") blip_commit_hash = os.environ.get('BLIP_COMMIT_HASH', "48211a1594f1321b00f14c9f7a5b4813144b2fb9") + tomesd_commit_hash = os.environ.get('TOMESD_COMMIT_HASH', "4f936c257e10848e0399fc6d0484a1761812092a") if not args.skip_python_version_check: check_python_version() @@ -280,6 +282,10 @@ def prepare_environment(): elif platform.system() == "Linux": run_pip(f"install {xformers_package}", "xformers") + if (not is_installed("tomesd") or args.reinstall_tomesd) and args.token_merging: + git_clone(tomesd_repo, repo_dir('tomesd'), "tomesd", tomesd_commit_hash) + run_pip(f"install {repo_dir('tomesd')}") + if not is_installed("pyngrok") and args.ngrok: run_pip("install pyngrok", "ngrok") diff --git a/modules/cmd_args.py b/modules/cmd_args.py index 81c0b82a3..4314f97b1 100644 --- a/modules/cmd_args.py +++ b/modules/cmd_args.py @@ -101,3 +101,7 @@ parser.add_argument("--no-gradio-queue", action='store_true', help="Disables gra parser.add_argument("--skip-version-check", action='store_true', help="Do not check versions of torch and xformers") parser.add_argument("--no-hashing", action='store_true', help="disable sha256 hashing of checkpoints to help loading performance", default=False) parser.add_argument("--no-download-sd-model", action='store_true', help="don't download SD1.5 model even if no model is found in --ckpt-dir", default=False) + +# token merging / tomesd +parser.add_argument("--token-merging", action='store_true', help="Provides generation speedup by merging redundant tokens. (compatible with --xformers)", default=False) +parser.add_argument("--token-merging-ratio", type=float, help="Adjusts ratio of merged to untouched tokens. Range: (0.0-1.0], Defaults to 0.5", default=0.5) diff --git a/modules/sd_models.py b/modules/sd_models.py index 6ea874dfc..0b74aa0f0 100644 --- a/modules/sd_models.py +++ b/modules/sd_models.py @@ -9,6 +9,7 @@ from omegaconf import OmegaConf from os import mkdir from urllib import request import ldm.modules.midas as midas +import tomesd from ldm.util import instantiate_from_config @@ -430,6 +431,13 @@ def load_model(checkpoint_info=None, already_loaded_state_dict=None, time_taken_ try: with sd_disable_initialization.DisableInitialization(disable_clip=clip_is_included_into_sd): sd_model = instantiate_from_config(sd_config.model) + + if shared.cmd_opts.token_merging: + ratio = shared.cmd_opts.token_merging_ratio + + tomesd.apply_patch(sd_model, ratio=ratio) + print(f"Model accelerated using {(ratio * 100)}% token merging via tomesd.") + timer.record("token merging") except Exception as e: pass