全站搜索

AI 模型 2119

• moeFussion↗

  LoliRimuru

  The goal of this project is to improve generation of moe characters since most of other models are not good at it. \ \ __Some of the models can be used online on these plarforms:__ \ __Aipictors (Japanese) - https://www.aipictors.com__ It is recommende to use standard resolution such as 832x1144 \ Positive prompt example: masterpiece, best quality, 1girl, solo, full body, night, purple kimono, japanese clothes, barefoot, obi, bare legs, braided ponytail, feet, mole, toes, collarbone, toenails, electricity, looking at viewer, night sky, holding weapon, holding, holding polearm, floating, sakura, petals \ Negative prompt example: lowres, worst quality, low quality, old, early, bad anatomy, bad hands, 4koma, comic, greyscale, censored, jpeg artifacts, monochrome, simple background, \ All examples were generated using custom workflow in ComfyUI and weren't edited using inpainting. You can load the workflow by either importing the example images or importing the workflow directly v3.0 \ Anima v1 based \ Better anatomy, smaller model, better prompt adherance \ Use @artist for artists, case sensitive \ You can use both tags and natural language prompts and you will likely achieve better results combining both \ It's better to use longer prompts for better results, shorter (sub 10 tags) may result in lower quality \ Use er_sde sampler and simple or maybe beta scheduler, cfg around 4 maybe even closer to 3 v0.4.3 \ Fixed color issue \ General improvements \ \ v0.5.3 \ Integrated VAE\ File size reduced \ CLIP force reset fix \ \ v0.6.3 \ Style improvements \ Added PastelMix and Counterfeit style \ \ v0.7.x \ Style impovements \ Composition improvements \ \ v0.8.x \ Major improvement on higher resolutions \ Style improvements \ Flexibility and responsivity \ Added support for Night Sky YOZORA model \ \ v0.9.x \ Different approach at merging, you might find v0.8.x versions better \ Changes at supported models \ \ v2.1.X EXPERIMENTAL RELEASE \ Stable Diffusion 2.1-768 based \ Default negative prompt: (low quality, worst quality:1.4), (bad anatomy), extra finger, fewer digits, jpeg artifacts \ For positive prompt it's good to include tags: anime, (masterpiece, best quality) alternatively you may achieve positive response with: (exceptional, best aesthetic, new, newest, best quality, masterpiece, extremely detailed, anime, waifu:1.2) \ Though it's Loli Diffusion model it's

• bit-50↗

  Google

  The BiT model was proposed in Big Transfer (BiT): General Visual Representation Learning by Alexander Kolesnikov, Lucas Beyer, Xiaohua Zhai, Joan Puigcerver, Jessica Yung, Sylvain Gelly, Neil Houlsby. BiT is a simple recipe for scaling up pre-training of ResNet-like architectures (specifically, ResNetv2). The method results in significant improvements for transfer learning. Disclaimer: The team releasing ResNet did not write a model card for this model so this model card has been written by the Hugging Face team. *Transfer of pre-trained representations improves sample efficiency and simplifies hyperparameter tuning when training deep neural networks for vision. We revisit the paradigm of pre-training on large supervised datasets and fine-tuning the model on a target task. We scale up pre-training, and propose a simple recipe that we call Big Transfer (BiT). By combining a few carefully selected components, and transferring using a simple heuristic, we achieve strong performance on over 20 datasets. BiT performs well across a surprisingly wide range of data regimes -- from 1 example per class to 1M total examples. BiT achieves 87.5% top-1 accuracy on ILSVRC-2012, 99.4% on CIFAR-10, and 76.3% on the 19 task Visual Task Adaptation Benchmark (VTAB). On small datasets, BiT attains 76.8% on ILSVRC-2012 with 10 examples per class, and 97.0% on CIFAR-10 with 10 examples per class. We conduct detailed analysis of the main components that lead to high transfer performance.*

• Qwen3.6-27B-AEON-Ultimate-Uncensored-Multimodal-NVFP4-MTP↗

  AEON-7

  > [!IMPORTANT] > **Superseded by Qwen3.8 — use this instead:** `AEON-7/Qwen3.8-27B-AEON-ULTIMATE-UNCENSORED-NVFP4-MIXED` > > **Operator recipes (Spark Dynamic DFlash lattice, TP=2, RTX):** `AEON-7/Qwen3.8-27B-AEON-ULTIMATE-UNCENSORED` > > This Qwen3.6-27B card stays online for history and existing downloads. For new work, move to the Qwen3.8 Ultimate Uncensored NVFP4-MIXED release. That line is far superior in capability and reaches substantially more of what the architecture can do, with a more refined mixed quantization methodology (NVFP4 + FP8 lattice) than these older Qwen3.6 exports. > > Serve with the unified container: `ghcr.io/aeon-7/aeon-vllm-ultimate:latest` (Spark / GB10) or `ghcr.io/aeon-7/aeon-vllm-ultimate-rtx:latest` (RTX discrete). > **Deployment, operations & benchmarks → github.com/AEON-7/Qwen3.6-27B-AEON-Ultimate-Uncensored-DFlash** > > The GitHub repo is the source of truth for the production deployment guide, hardware-tuned docker-compose configs, full configuration reference, measured benchmarks, and `AGENTS.md` — an operator's manual that pre-empts common stale-documentation traps. > **🙏 Reference recipe credit:** The modelopt + MTP graft pipeline used to build this variant is based on **sakamakismile**'s validated Qwen3.6-27B-NVFP4-MTP series (22K+ downloads). They worked out the modelopt config, the per-projection quantization choices, and the MTP-head graft technique on the un-abliterated base; we adapted the same recipe to AEON-Ultimate's abliterated weights. The reference benchmark numbers cited below are theirs. Full credit for the recipe → sakamakismile.

• HunyuanVideo-I2V-gguf↗

  city96

  NOTE: Hunyuan recently replaced the original model in the repo above with a **new and updated version** - the files in this repo have been reconverted and replaced. If you want to test against the original models, you can find them on this branch As this is a quantized model not a finetune, all the same restrictions/original license terms still apply. Place model files in `ComfyUI/models/unet` - see the GitHub readme for further install instructions.

• Qwen3.6-35B-A3B-Claude-4.6-Opus-Reasoning-Distilled↗

  hesamation

• parakeet-redux↗

  moondream

  A 1.58-bit version of parakeet-tdt-0.6b-v3. Same architecture, same tokenizer, but every encoder weight is -1, 0 or +1. It fits in 178 MB, runs at 113× real time on eight x86 CPU cores, 2.5× the fastest other Parakeet runtime we measured, and stays within 0.3 WER of the original on English while beating it on the 25-language FLEURS set and on long-form audio. Its sibling Parakeet Ultra is the full-precision version of the same architecture, trained further, for GPUs: better than the original on every benchmark. Read the release post for the story behind both models. Run it with Photon, which reads the packed weights directly: AVX-512 VNNI on x86, NEON on ARM, Metal on Apple GPUs. Every speed number on this page is Photon.

• U-2-Netp↗

  BritishWerewolf

  U-2-Netp is a lightweight version of the U2Net model designed for efficient and effective image segmentation tasks, especially for generating masks. It retains the core architectural design of U2Net while being optimized for faster inference times and reduced memory usage. The U-2-Netp model is based on a simplified version of the original U2Net architecture, designed to be more lightweight while still achieving high performance in segmentation tasks. The model consists of several stages with down-sampling and up-sampling paths, using Residual U-blocks (RSU) for enhanced feature representation. To use the model for inference, you can follow the example provided above. The `AutoProcessor` and `AutoModel` classes from the `transformers` library make it easy to load the model and processor.

• sam-audio-base↗

  facebook

• ECE-TW3-JRGL-V1↗

  paloalma

  >_**ECE**, a multi-program, multi-campus, and multi-sector engineering school specializing in digital engineering, > trains engineers and technology experts for the 21st century, capable of meeting the challenges of the dual digital and sustainable development revolutions. >French Engineering School ECE_ --- @misc{paech2023eqbench, title={EQ-Bench: An Emotional Intelligence Benchmark for Large Language Models}, author={Samuel J. Paech}, year={2023}, eprint={2312.06281}, archivePrefix={arXiv}, primaryClass={cs.CL} } ---

• limite-1b-violetto↗

  paradigma-inc

  Limite 1B - Violetto is Paradigma’s first model, designed for high-throughput solutions of difficult mathematical problems. Limite is a 1-billion parameter dense autoregressive transformer, trained from scratch on a mixture of less than 300B highly curated tokens, capable of handling sequences up to 131k tokens long. Apart from data curation, we achieve such high sample efficiency by equipping Limite with an architecture strongly inspired by recent advancements made by the community on pre-training speedrun competitions. The base model itself scores well on math benchmarks with few-shot prompting.