from collections import namedtuple from typing import List ModelInfo = namedtuple("ModelInfo", ["simple_name", "link", "description"]) model_info = {} def register_model_info( full_names: List[str], simple_name: str, link: str, description: str ): info = ModelInfo(simple_name, link, description) for full_name in full_names: model_info[full_name] = info def get_model_info(name: str) -> ModelInfo: if name in model_info: return model_info[name] else: # To fix this, please use `register_model_info` to register your model return ModelInfo( name, "", "Register the description at fastchat/model/model_registry.py" ) def get_model_description_md(model_list): model_description_md = """ | | | | | ---- | ---- | ---- | """ ct = 0 visited = set() for i, name in enumerate(model_list): minfo = get_model_info(name) if minfo.simple_name in visited: continue visited.add(minfo.simple_name) one_model_md = f"[{minfo.simple_name}]({minfo.link}): {minfo.description}" if ct % 3 == 0: model_description_md += "|" model_description_md += f" {one_model_md} |" if ct % 3 == 2: model_description_md += "\n" ct += 1 return model_description_md # regist image generation models register_model_info( ["imagenhub_LCM_generation", "fal_LCM_text2image"], "LCM", "https://huggingface.co/SimianLuo/LCM_Dreamshaper_v7", "Latent Consistency Models.", ) register_model_info( ["fal_LCM(v1.5/XL)_text2image"], "LCM(v1.5/XL)", "https://fal.ai/models/fast-lcm-diffusion-turbo", "Latent Consistency Models (v1.5/XL)", ) register_model_info( ["imagenhub_PlayGroundV2_generation", 'playground_PlayGroundV2_generation'], "Playground v2", "https://huggingface.co/playgroundai/playground-v2-1024px-aesthetic", "Playground v2 – 1024px Aesthetic Model", ) register_model_info( ["imagenhub_PlayGroundV2.5_generation", 'playground_PlayGroundV2.5_generation'], "Playground v2.5", "https://huggingface.co/playgroundai/playground-v2.5-1024px-aesthetic", "Playground v2.5 is the state-of-the-art open-source model in aesthetic quality", ) register_model_info( ["imagenhub_OpenJourney_generation"], "Openjourney", "https://huggingface.co/prompthero/openjourney", "Openjourney is an open source Stable Diffusion fine tuned model on Midjourney images, by PromptHero.", ) register_model_info( ["imagenhub_SDXLTurbo_generation", "fal_SDXLTurbo_text2image"], "SDXLTurbo", "https://huggingface.co/stabilityai/sdxl-turbo", "SDXL-Turbo is a fast generative text-to-image model.", ) register_model_info( ["imagenhub_SDXL_generation", "fal_SDXL_text2image"], "SDXL", "https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0", "SDXL is a Latent Diffusion Model that uses two fixed, pretrained text encoders.", ) register_model_info( ["imagenhub_SD3_generation"], "SD3", "https://huggingface.co/blog/sd3", "SD3 is a novel Multimodal Diffusion Transformer (MMDiT) model.", ) register_model_info( ["imagenhub_PixArtAlpha_generation"], "PixArtAlpha", "https://huggingface.co/PixArt-alpha/PixArt-XL-2-1024-MS", "Pixart-α consists of pure transformer blocks for latent diffusion.", ) register_model_info( ["imagenhub_PixArtSigma_generation", "fal_PixArtSigma_text2image"], "PixArtSigma", "https://github.com/PixArt-alpha/PixArt-sigma", "Improved version of Pixart-α.", ) register_model_info( ["imagenhub_SDXLLightning_generation", "fal_SDXLLightning_text2image"], "SDXL-Lightning", "https://huggingface.co/ByteDance/SDXL-Lightning", "SDXL-Lightning is a lightning-fast text-to-image generation model.", ) register_model_info( ["imagenhub_StableCascade_generation", "fal_StableCascade_text2image"], "StableCascade", "https://huggingface.co/stabilityai/stable-cascade", "StableCascade is built upon the Würstchen architecture and working at a much smaller latent space.", ) register_model_info( ["imagenhub_HunyuanDiT_generation"], "HunyuanDiT", "https://github.com/Tencent/HunyuanDiT", "HunyuanDiT is a Powerful Multi-Resolution Diffusion Transformer with Fine-Grained Chinese Understanding", ) register_model_info( ["imagenhub_Kolors_generation"], "Kolors", "https://huggingface.co/Kwai-Kolors/Kolors", "Kolors is a large-scale text-to-image generation model based on latent diffusion", ) register_model_info( ["fal_AuraFlow_text2image"], "AuraFlow", "https://huggingface.co/fal/AuraFlow", "Opensourced flow-based text-to-image generation model.", ) register_model_info( ["fal_FluxTimestep_text2image"], "FLUX.1-schnell", "https://huggingface.co/docs/diffusers/main/en/api/pipelines/flux", "Flux is a series of text-to-image generation models based on diffusion transformers. Timestep-distilled version.", ) register_model_info( ["fal_FluxGuidance_text2image"], "FLUX.1-dev", "https://huggingface.co/docs/diffusers/main/en/api/pipelines/flux", "Flux is a series of text-to-image generation models based on diffusion transformers. Guidance-distilled version.", ) # regist image edition models register_model_info( ["imagenhub_CycleDiffusion_edition"], "CycleDiffusion", "https://github.com/ChenWu98/cycle-diffusion?tab=readme-ov-file", "A latent space for stochastic diffusion models.", ) register_model_info( ["imagenhub_Pix2PixZero_edition"], "Pix2PixZero", "https://pix2pixzero.github.io/", "A zero-shot Image-to-Image translation model.", ) register_model_info( ["imagenhub_Prompt2prompt_edition"], "Prompt2prompt", "https://prompt-to-prompt.github.io/", "Image Editing with Cross-Attention Control.", ) register_model_info( ["imagenhub_InstructPix2Pix_edition"], "InstructPix2Pix", "https://www.timothybrooks.com/instruct-pix2pix", "An instruction-based image editing model.", ) register_model_info( ["imagenhub_MagicBrush_edition"], "MagicBrush", "https://osu-nlp-group.github.io/MagicBrush/", "Manually Annotated Dataset for Instruction-Guided Image Editing.", ) register_model_info( ["imagenhub_PNP_edition"], "PNP", "https://github.com/MichalGeyer/plug-and-play", "Plug-and-Play Diffusion Features for Text-Driven Image-to-Image Translation.", ) register_model_info( ["imagenhub_InfEdit_edition"], "InfEdit", "https://sled-group.github.io/InfEdit/", "Inversion-Free Image Editing with Natural Language.", ) register_model_info( ["imagenhub_CosXLEdit_edition"], "CosXLEdit", "https://huggingface.co/stabilityai/cosxl", "An instruction-based image editing model from SDXL.", ) register_model_info( ["imagenhub_UltraEdit_edition"], "UltraEdit", "https://ultra-editing.github.io/", "Instruction-based Fine-Grained Image Editing at Scale.", ) register_model_info( ["fal_stable-cascade_text2image"], "StableCascade", "https://fal.ai/models/stable-cascade/api", "StableCascade is a generative model that can generate high-quality images from text prompts.", ) register_model_info( ["fal_AnimateDiff_text2video"], "AnimateDiff", "https://fal.ai/models/fast-animatediff-t2v", "AnimateDiff is a text-driven models that produce diverse and personalized animated images.", ) register_model_info( ["fal_StableVideoDiffusion_text2video"], "StableVideoDiffusion", "https://fal.ai/models/fal-ai/fast-svd/text-to-video/api", "Stable Video Diffusion empowers individuals to transform text and image inputs into vivid scenes.", ) register_model_info( ["fal_AnimateDiffTurbo_text2video"], "AnimateDiff Turbo", "https://fal.ai/models/fast-animatediff-t2v-turbo", "AnimateDiff Turbo is a lightning version of AnimateDiff.", ) register_model_info( ["videogenhub_LaVie_generation"], "LaVie", "https://github.com/Vchitect/LaVie", "LaVie is a video generation model with cascaded latent diffusion models.", ) register_model_info( ["videogenhub_VideoCrafter2_generation"], "VideoCrafter2", "https://ailab-cvc.github.io/videocrafter2/", "VideoCrafter2 is a T2V model that disentangling motion from appearance.", ) register_model_info( ["videogenhub_ModelScope_generation"], "ModelScope", "https://arxiv.org/abs/2308.06571", "ModelScope is a a T2V synthesis model that evolves from a T2I synthesis model.", ) register_model_info( ["videogenhub_OpenSora_generation"], "OpenSora", "https://github.com/hpcaitech/Open-Sora", "A community-driven opensource implementation of Sora.", ) register_model_info( ["videogenhub_OpenSora12_generation"], "OpenSora v1.2", "https://github.com/hpcaitech/Open-Sora", "A community-driven opensource implementation of Sora. v1.2", ) register_model_info( ["videogenhub_CogVideoX_generation"], "CogVideoX", "https://github.com/THUDM/CogVideo", "Text-to-Video Diffusion Models with An Expert Transformer.", ) register_model_info( ["videogenhub_T2VTurbo_generation"], "T2V-Turbo", "https://github.com/Ji4chenLi/t2v-turbo", "Video Consistency Model with Mixed Reward Feedback.", )