Naclbit
Rank:Average Model Cost: $0.0000
Number of Runs: 23,209
Models by this creator
trinart_stable_diffusion_v2
trinart_stable_diffusion_v2
The trinart_stable_diffusion_v2 model is a text-to-image model that generates anime/manga-style illustrations based on provided text prompts. It has been trained on approximately 40,000 high-resolution manga/anime-style images for 8 epochs. The model aims to retain the original aesthetic of the source images while improving the quality of the generated outputs. This version of the model uses dropouts, an additional 10,000 images for training, and a new tagging strategy. It has been trained for a longer period of time compared to the previous version. The model offers three different checkpoints for inference, and it can be used with the latent-diffusion's stock ddim img2img script. The model is created by Sta, AI Novelist Dev and the Stable Diffusion team.
$-/run
23.0K
Huggingface
gpt-j-japanese-6.8b
gpt-j-japanese-6.8b
This pre-trained model is work in progress! Model weight download will be available in the future. A 6.8 billion parameter pre-trained model for Japanese language, based on EleutherAI's Mesh Transformer JAX, that has a similar model structure to their GPT-J-6B pre-trained model. EleutherAIによるMesh Transformer JAXをコードベースとした、GPT-J-6Bに似たストラクチャと約68.7億パラメータを持つ日本語pre-trainedモデルです。 We used T5Tokenizer and SentencePiece instead of GPT-2/3 tokenizer. Normalization done by SentencePiece is must for Japanese tokenizing as there are so much many more variations for common symbols than Western languages. Tokenizer has a vocabulary of 52,500 tokens and trained on Japanese Wikipedia dump as of 01 Aug 2021. The model fits within 16GB VRAM GPUs like P100 for inference up to 1688 context length. Full 2048 context length output requires 20GB VRAM or more (e.g. GTX3090/A5000). The model was trained with TPUv3-128 generously provided by Google TRC for about 4 weeks. We are currently formatting additional datasets and preparing for more training time. Specifications Instructions We recommend to use finetuneanon's forked transformer codebase for inferencing as split checkpoint loads up a lot faster than monolithic checkpoint supported by HuggingFace Transformers repository. The tokenizer still uses 50256 as the <|endoftext|> substitute. Therefore 50256 should be excluded when inferencing. Datasets Lack of quality Japanese corpus was one of the major challenges when we trained the model. We aimed to compile well-formatted corpuses outside of Common Crawl. The dataset is normalized and sanitized against leading and trailing spaces, excessive CR/LF repetitions. The whole dataset is about 400GB (as of October 2021) and 106B tokens (compared to 825GB/300B tokens for The Pile). ** Common Crawl Jan-Dec 2018 72GB CC100-Japanese (https://metatext.io/datasets/cc100-japanese) November 2018 106GB OSCAR-Japanese (https://oscar-corpus.com) 75GB Converted 860GB Google C4 Multilingual Japanese (re-formatted) ** Books 140GB Web Fictions, non-fictions and blogs corpus 5GB Books and Aozora Bunko corpus (weighted 2x) ** News 1GB Scientific news, medical news and web news corpus ** Wikipedia Aug 2021 3GB Assorted and Deduplicated Japanese Wikipedia (weighted 2x) Aug 2021 Wikibooks, Wikinews, Wikiquote, Wikisource, Wiktionary, Wikiversity and Wikivoyage ** Other Corpuses 2018 OpenSubtitles (https://opus.nlpl.eu/OpenSubtitles-v2018.php) 80-90's BBS Logs Assorted Blogs Crawl QED-ja TED 2020-ja
$-/run
88
Huggingface
trinart_characters_19.2m_stable_diffusion_v1
trinart_characters_19.2m_stable_diffusion_v1
Note A newer version of this model has been released: https://huggingface.co/naclbit/trinart_derrida_characters_v2_stable_diffusion Stable Diffusion TrinArt Characters model v1 trinart_characters_19.2m_stable_diffusion_v1 is a stable diffusion v1-based model trained by roughly 19.2M anime/manga style images (pre-rolled augmented images included) plus final finetuning by about 50,000 images. This model seeks for a sweet spot between artistic style versatility and anatomical quality within the given model spec of SDv1. This is the same version 1 model that was released in AI Novelist/TrinArt service from early September through Oct 14. We are currently experimenting with the new Derrida model on TrinArt service for further improvement and anatomical stabilization. In the mean time, please enjoy this real-service-tested Characters v1! 8xNVIDIA A100 40GB Note: There was a wrong checkpoint uploaded before 5 Nov 2022. The file has been replaced with the latest checkpoint. We also provide a separate checkpoint for the custom KL autoencoder. As suggested by the Latent Diffusion paper, we found that training the autoencoder and the latent diffusion model separately improves the result. Since the official stable diffusion script does not support loading the other VAE, in order to run it in your script, you'll need to override state_dict for first_stage_model. The popular WebUI has the script to load separate first_stage_model parameters. The dataset is filtered to exclude NSFW or unsafe contents. After our extensive experimentation and testing with 10M+ user generated images, we decided that this model is safe enough and less likely to spit out questionable (nudity/overly sexual/realistic gore) content than the stock SD v1.4 model or other anime/manga models. However, if the user tortures this model enough until it talks, it may be still possible to force this model to generate obnoxious materials. We do not consider this model to be 100% risk-free. *This statement does not necessarily restrict third-party from training a derivative of this model that includes NSFW. Below images are directly generated by the native TrinArt service with its idiosyncratic upscaler, parser and processes. Your mileage may vary. (assorted random examples) wide shot, high quality, htgngg animal arm rest brown hair merry chair cup dress flower from above jacket on shoulders long hair sitting solo sugar bowl fantasy adventurer's inn table teacup teapot landscape miniature (2022 Artstyle preset) highres wide shot bangs bare shoulders water bird cage terrarium detached sleeves frilled frilled legwear frills hair ornament hair ribbon hood long hair medium breasts ribbon thighhighs (2019 Artstyle preset) 1girl standing holding sword hizzrd arm up bangs bare shoulders boots bow breasts bright pupils choker detached sleeves diamond (shape) floating floating hair footwear bow from side full body gloves leg up long hair looking at viewer open mouth outstretched arm solo streaked hair swept bangs two tone hair very long hair::4 angry::1 (2022 Artstyle preset) 1boy male focus standing hizzrd holding sword arm up bow bright pupils cape coat diamond (shape) floating floating hair fold-over boots footwear bow from side full body gloves leg up long sleeves looking at viewer open mouth outstretched arm open coat open clothes solo swept two tone hair thigh boots::4 angry::1.25 (2022 Artstyle preset) cathedral 1girl schoolgirl momoko school uniform cats particles beautiful shooting stars detailed cathedral jacket open mouth glasses cats (2022 Artstyle preset) highres 2girls yuri wide shot bangs bare shoulders water bird cage terrarium detached sleeves frilled frilled legwear frills hair ornament hair ribbon hood long hair medium breasts ribbon thighhighs (More Details preset) wide shot, best quality lapis erebcir highres 1boy bangs black gloves brown hair closed mouth gloves hair between eyes looking at viewer male focus flowers green eyes (More Details preset) TrinArt 2022 Artstyle preset negative prompts: retro style, 1980s, 1990s, 2000s, 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 TrinArt More Details preset negative prompts: flat color, flat shading We recommend to add known sets of negative prompts in order to stabilize the anatomy such as: bad hands, fewer digits, etc. Sta, AI Novelist Dev (https://ai-novel.com/) @ Bit192, Inc. Twitter https://twitter.com/naclbbr (Japanese) https://twitter.com/naclbbre (English) Stable Diffusion - Rombach, Robin and Blattmann, Andreas and Lorenz, Dominik and Esser, Patrick and Ommer, Bjorn CreativeML OpenRAIL-M
$-/run
61
Huggingface
trin_tokenizer_v3
$-/run
54
Huggingface
danbooru2021_256px_guided_diffusion
danbooru2021_256px_guided_diffusion
Platform did not provide a description for this model.
$-/run
0
Huggingface
trinart_derrida_characters_v2_stable_diffusion
trinart_derrida_characters_v2_stable_diffusion
Stable Diffusion TrinArt Derrida model (Characters v2) Derrida (formerly TrinArt Characters v2) is a stable diffusion v1-based model that was further improved on the previous characters v1 model. While this is still a versatility and compositional variation anime/manga model like other TrinArt models, when compared to the v1 model, Derrida was focused on more anatomical stability and slightly less on variation due to further multi-epoch training and finetuning. The pre-rolled augmentation phase by generating slight variations w/ img2img is still applied right before the finetuning phase. This is the same model that was released in AI Novelist/TrinArt service from mid-Oct through early November. 8xNVIDIA A100 40GB Note: The autoencoder uploaded here is the same checkpoint as v1. We also provide a separate checkpoint for the custom KL autoencoder. As suggested by the Latent Diffusion paper, we found that training the autoencoder and the latent diffusion model separately improves the result. Since the official stable diffusion script does not support loading the other VAE, in order to run it in your script, you'll need to override state_dict for first_stage_model. The popular WebUI has the script to load separate first_stage_model parameters. The dataset has been filtered to avoid extremely NSFW materials, but slightly less strict than v1. As with any other image generation model, we don't recommend deploying this model publicly without safety considerations and measures. Depends on prompting, one may still be able to extract highly questionable images from this model. This statement does not necessarily restrict third-party from training a derivative or mix of this model that includes NSFW. Below images are directly generated by the native TrinArt service with its idiosyncratic upscaler, parser and processes. Your mileage may vary. TrinArt 2022 Artstyle preset negative prompts: retro style, 1980s, 1990s, 2000s, 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 TrinArt More Details preset negative prompts: flat color, flat shading We recommend to add known sets of negative prompts in order to stabilize the anatomy such as: bad hands, fewer digits, etc. Sta, AI Novelist Dev (https://ai-novel.com/) @ Bit192, Inc. Twitter https://twitter.com/naclbbr (Japanese) https://twitter.com/naclbbre (English) Stable Diffusion - Rombach, Robin and Blattmann, Andreas and Lorenz, Dominik and Esser, Patrick and Ommer, Bjorn CreativeML OpenRAIL-M
$-/run
0
Huggingface
trinart_stable_diffusion
$-/run
0
Huggingface