全站搜索

AI 模型 2119

• Kohaku-XL-Zeta↗

  KBlueLeaf

  .image-viewer {position: relative;width: 100%;margin: 0 auto;display: flex;flex-flow: wrap;align-items: center;justify-content: center;} .image-viewer input[type="radio"] {display: none;} .image-viewer label {border-radius: 10%;padding: 20px;background-color: #B398F5;background-size: cover;background-position: center;cursor: pointer;color: black;margin: 8px;} .image-viewer label:hover {background-color: #4C88F5;padding: 24px;margin: 4px;} .image-viewer input[type="radio"]:checked + label {background-color: #6296F5;padding: 28px;margin: 0px;} .image-container {position: relative;width: 100%;height: 50vh;margin: 1rem 1rem 0 0;} .inner-container {position: absolute;width:100%;height: 100%;display: flex;align-items: center;justify-content: center;} .inner-container img {border-radius: 10px;max-height: 100%;max-width: 100%;height: 0;width: 0;opacity: 0;transition: opacity 0.5s ease, height 0.25s ease, width 0.25s ease;} As same as Kohaku XL Epsilon or Delta, but you can replace "general tags" with "natural language caption". You can also put both together. For better ability on some certain concepts, I use full danbooru dataset instead of filterd one. Than use crawled Pixiv dataset (from 3~5 tag with popularity sort) as addon dataset. Since Pixiv's search system only allow 5000 page per tag so there is not much meaningful image, and some of them are duplicated with danbooru set(but since I want to reinforce these concept I directly ignore the duplication)

• Indigo_Furry_Mix_XL_CkNoob_EPS_11_checkpoint↗

  Quiho

  These are some XL models based on NoobAI/pony XL with a little modifications, they are mainly for anime/realistic bara kemono content, and should be ok for all contents. I chose pony purely because of its NSFW ability. (Now I'm thinking about switching to NoobXL because Pony is a pain in the ass to tweak) NoobAI based model out now. If you like my work, consider buy me a coffee please <3 (completely voluntary) \u5982\u540cLicense\u5185\u63d0\u53ca\u7684\uff0c\u672c\u6a21\u578b\u53ca\u5176\u751f\u6210\u7684\u56fe\u50cf\u7981\u6b62\u5546\u4e1a\u5316\u4f7f\u7528\uff01 For CkNoob_EPS_\u216a_11: This is a new EPS model that based on Chenkin Noob v0.1. Based on my testing, compared to EPS10, this model features better background coherence, higher contrast, slightly more detail, and a bit more responsiveness to artist styles. Just note that the composition of the generated images with the same prompt can sometimes differ noticeably from EPS10. And the overall feel and behavior of this model can sometimes differ from EPS10. Just as the EPS10, when using this model, try using a CFG scale between 3\u20137 (recommended 5). It is recommended to keep using Euler a sampler for best overall results, but switching to DPM++ 2M may produce more detailed (and chaotic) outputs. For NoobAI_EPS_\u2169_10: This is a new EPS model that merges the the core features (hopefully) of EPS3 and EPS7. It aims to increase overall versatility while sacrificing just bit of fine detail. Backgrounds are much more stable compared to EPS3, though occasionally not as consistent as EPS7. EPS10 fixed the issue of overly saturated or blown-out colors at high CFG values. Artist tags still work well as before, and stability when using the DPM++ 2M sampler has been improved. When using this model, try using a CFG scale between 3\u20137 (recommended 5). It is recommended to keep using Euler a sampler for best overall results, but switching to DPM++ 2M may produce more detailed (chaotic) outputs.

• stable-diffusion-2-1-base↗

  sd2-community

  This model card focuses on the model associated with the Stable Diffusion v2-1-base model. This `stable-diffusion-2-1-base` model fine-tunes stable-diffusion-2-base (`512-base-ema.ckpt`) with 220k extra steps taken, with `punsafe=0.98` on the same dataset. - Use it with the `stablediffusion` repository: download the `v2-1_512-ema-pruned.ckpt` here. - Use it with 🧨 `diffusers`

• cryptobert↗

  ElKulako

  For academic reference, cite the following paper: https://ieeexplore.ieee.org/document/10223689 CryptoBERT is a pre-trained NLP model to analyse the language and sentiments of cryptocurrency-related social media posts and messages. It was built by further training the vinai's bertweet-base language model on the cryptocurrency domain, using a corpus of over 3.2M unique cryptocurrency-related social media posts. (A research paper with more details will follow soon.) The model was trained on the following labels: "Bearish" : 0, "Neutral": 1, "Bullish": 2 CryptoBERT's sentiment classification head was fine-tuned on a balanced dataset of 2M labelled StockTwits posts, sampled from ElKulako/stocktwits-crypto.

• bert-large-uncased-whole-word-masking-finetuned-squad↗

  google-bert

  Pretrained model on English language using a masked language modeling (MLM) objective. It was introduced in this paper and first released in this repository. This model is uncased: it does not make a difference between english and English. Differently to other BERT models, this model was trained with a new technique: Whole Word Masking. In this case, all of the tokens corresponding to a word are masked at once. The overall masking rate remains the same. The training is identical -- each masked WordPiece token is predicted independently.

• gemma-4-26B-A4B-it-litert-lm↗

  litert-community

  This model card provides the Gemma 4 26B (A4B) mixture-of-experts model in LiteRT-LM format that is ready for deployment on web and desktop. Please check back here regularly for updates on wider platform support and further functionality improvements. The current LiteRT-LM version supports text; audio, image, and multitoken prediction support will be available in a future update. Gemma is a family of lightweight, state-of-the-art open models from Google, built from the same research and technology used to create the Gemini models. This particular Gemma 4 model is a medium size so it is ideal for desktop use cases. By running this model on device, users can have private access to Generative AI technology without even requiring an internet connection. These models are provided in the `.litertlm` format for use with the LiteRT-LM framework. LiteRT-LM is a specialized orchestration layer built directly on top of LiteRT, Google’s high-performance multi-platform runtime trusted by millions of Android and edge developers. LiteRT provides the foundational hardware acceleration via XNNPack for CPU and ML Drift for GPU. LiteRT-LM adds the specialized GenAI libraries and APIs, such as KV-cache management, prompt templating, and function calling. This integrated stack is the same technology powering the Google AI Edge Gallery showcase app.

• deberta-v3-xsmall-zeroshot-v1.1-all-33↗

  MoritzLaurer

  This model was fine-tuned using the same pipeline as described in the model card for MoritzLaurer/deberta-v3-large-zeroshot-v1.1-all-33 and in this paper. The foundation model is microsoft/deberta-v3-xsmall. The model only has 22 million backbone parameters and 128 million vocabulary parameters. The backbone parameters are the main parameters active during inference, providing a significant speedup over larger models. The model is 142 MB small. This model was trained to provide a small and highly efficient zeroshot option, especially for edge devices or in-browser use-cases with transformers.js.

• gemma-4-E4B-it-qat-q4_0-unquantized↗

  Google

  > [!Note] > This model card is for the new versions of the Gemma 4 family optimized with Quantization-Aware Training (QAT), which allows preserving similar quality to bfloat16 while dramatically reducing the memory requirements to load the model. > Four versions of the QAT checkpoints are available: > * **Unquantized QAT checkpoints** (Q4_0): Half-precision weights extracted from the QAT pipeline, ideal for custom downstream compilation and research. Available for Gemma 4 E2B, E4B, 12B, 26B A4B, and 31B, and their drafter models. > * **GGUF** (Q4_0): Ready-to-deploy formats for broad ecosystem compatibility. Available for Gemma 4 E2B, E4B, 12B, 26B A4B, and 31B. > * **Mobile-optimized** (wNa8o8): A custom schema engineered explicitly for mobile hardware efficiency. It features targeted 2-bit decoding layers, optimized KV caches, and static activations to maximize VRAM savings. Available for Gemma 4 E2B and E4B. > * **Compressed Tensors** (w4a16): QAT checkpoints serialized in the compressed-tensors format for native, optimized inference with vLLM. Available for Gemma 4 E2B, E4B, 12B, and 31B. > * **Assistant Compatibility**: When using multi-token prediction (speculative decoding) with an assistant model alongside a QAT target model, the assistant model must also be a QAT checkpoint with the same precision to ensure compatibility. Gemma is a family of open models built by Google DeepMind. Gemma 4 models are multimodal, handling text and image input (with audio supported on E2B, E4B, and 12B) and generating text output. This release includes open-weights models in both pre-trained and instruction-tuned variants. Gemma 4 features a context window of up to 256K tokens and maintains multilingual support in over 140 languages. Featuring both Dense and Mixture-of-Experts (MoE) architectures, Gemma 4 is well-suited for tasks like text generation, coding, and reasoning. The models are available in five distinct sizes: **E2B**, **E4B**, **12B**, **26B A4B**, and **31B**. Their diverse sizes make them deployable in environments ranging from high-end phones to laptops and servers, democratizing access to state-of-the-art AI.

• hviske-v5.3↗

  syvai

  Danish ASR. Fine-tuned from syvai/hviske-v5.1 on the CoRal v3 train splits with **layer-wise learning-rate decay** (encoder LR = 0.75 × decoder LR) for 5 epochs. A 2B-parameter Conformer encoder-decoder ASR model, optimized for Danish read-aloud and conversational speech. Evaluated on the *complete* test splits (17,560 samples). Two normalization conventions:

• blossom-v5.1-9b↗

  Azure99

  Blossom is a conversational large language model, fine-tuned on the Blossom Orca/Wizard/Chat/Math mixed dataset based on the Yi-1.5-9B pre-trained model. Blossom possesses robust general capabilities and context comprehension. Additionally, the high-quality Chinese and English datasets used for training have been made open source. Training was conducted in two stages. The first stage used 40K Wizard, 40K Orca, 10K Math single-turn instruction datasets, training for 1 epoch; the second stage used 10K Blossom chat multi-turn dialogue dataset, and 10% randomly sampled data from the first stage, training for 3 epochs.