全站搜索

AI 模型 2119

• RE-USE-GGUF↗

  audio-cpp

  GGUF weights for native inference with audio.cpp. RE-USE restores degraded speech while preserving its input sample rate. It uses a convolutional encoder/decoder and 30 bidirectional time-frequency Mamba blocks. The checkpoint and reference implementation are pinned to nvidia/RE-USE, revision 022e920d727347a64d6c21fbf0628f2a5f37ad78. Each GGUF includes the model weights, configuration, and audio.cpp model spec. No separate encoder or tokenizer is required. Packages are converted directly from the original safetensors, not from another GGUF. Only F32 is provided. F16 and Q8 were rejected because their small disk savings did not improve speed or peak VRAM in testing, while introducing waveform drift and additional packaging complexity.

• byt5-leetspeak-decoder↗

  ilyyeees

  **The definitive byte-level translator for leetspeak, internet slang, and visual character obfuscation.** Built on `google/byt5-base`, **V3** represents a major architectural shift from previous versions. It utilizes **Curriculum Learning** and **Adversarial Filtering** to solve the complex context ambiguity between leetspeak numbers (e.g., "2" meaning "to") and actual quantities (e.g., "2 cats"). V3 is the first model in this series to perfectly distinguish between numbers used as letters and numbers used as quantities within the same sentence. * **Input:** `1t5 2 l8 4 2 people` * **V2 Output:** *It's to late for to people.* (Fail) * **V3 Output:** *It is too late for 2 people.* (Pass)

• TIPO-200M-ft2↗

  KBlueLeaf

  In this project, we introduce "TIPO" (**T**ext to **I**mage with text presampling for **P**rompt **O**ptimization), an innovative framework designed to significantly enhance the quality and usability of Text-to-Image (T2I) generative models. TIPO utilizes the Large Language Models (LLMs) to perform "Text Presampling" within the inference pipeline of text-to-image generative modeling. By refining and extending user input prompts, TIPO enables generative models to produce superior results with minimal user effort, making T2I systems more accessible and effective for a wider range of users. Use updated version of DTG extension (renamed to z-tipo-extension), current version of z-tipo-extension support stable-diffusion-webui, stable-diffusion-webui-forge and ComfyUI. SD-Next haven't been tested. https://github.com/KohakuBlueleaf/z-tipo-extension This model is LLaMA arch with 200M parameters, the training data is combined version of Danbooru2023, Coyo-HD-11M. The total token seen is around 50B tokens. For more information please refer to the tech report and following table.

• CreativeWriter-Llama3.2-3B-GGUF↗

  theprint

  This Llama 3.2 model was fine tuned on two combined data sets, one full of writing coach conversations, the other containing actual creative writing samples. Both data sets used were synthetic. You can find a more in-depth description of this model and its intended use in this blog post.

• chronos-t5-large↗

  amazon

  🚀 **Update Feb 14, 2025**: Chronos-Bolt & original Chronos models are now available on Amazon SageMaker JumpStart! Check out the tutorial notebook to learn how to deploy Chronos endpoints for production use in a few lines of code. 🚀 **Update Nov 27, 2024**: We have released Chronos-Bolt⚡️ models that are more accurate (5% lower error), up to 250 times faster and 20 times more memory-efficient than the original Chronos models of the same size. Check out the new models here. Chronos is a family of **pretrained time series forecasting models** based on language model architectures. A time series is transformed into a sequence of tokens via scaling and quantization, and a language model is trained on these tokens using the cross-entropy loss. Once trained, probabilistic forecasts are obtained by sampling multiple future trajectories given the historical context. Chronos models have been trained on a large corpus of publicly available time series data, as well as synthetic data generated using Gaussian processes.

• NV-Reason-CT↗

  NVIDIA

  NV-Reason-CT is a 3D vision-language model (VLM) for CT image analysis. It combines a native 3D vision encoder (3D ViT) with a language model and is designed for radiology report generation, general question answering, and multi-step reasoning across chest and abdominal CT volumes. Computed tomography encodes clinically important anatomy across hundreds of slices, yet most vision–language systems either operate on 2D images or compress volumetric features before language decoding. The 3D vision encoder converts a 384×384×384-mm input volume into a 24×24×24 grid of 13,824 visual tokens. All tokens and their corresponding 3D positions are passed to the language model without spatial downsampling, while 3D MRoPE preserves their spatial relationships within the LLM. Input volumes are automatically cropped to the chest or abdomen and resampled to 2-mm isotropic resolution before processing. The model was trained end-to-end using Supervised Fine-Tuning (SFT) and Group Relative Policy Optimization (GRPO) on a curated dataset of approximately 550,000 structured QA examples from 70,111 unique CT volume inputs. The training corpus integrates standardized reports, abnormality-focused QA, multi-turn interactions, and radiologist-authored reasoning collected through recorded and transcribed expert CT interpretations. These expert annotations also guide the generation of report-grounded synthetic reasoning data for SFT, while GRPO uses verifiable rewards over chest and abdominal abnormality sets.

• MOSS-Image-Text-Encoder-Heretic-DHQ-GGUF↗

  pottokao

  One file does two jobs: it **rewrites** a short idea into a full MOSS-Image caption, then **encodes** that caption for the DiT. Weights are loaded once. A Q4-class file fits an 8 GB card. The quantization is **DHQ** (DiT-aware quantization): every tensor's format is chosen from measured damage on two rulers at once — the error after MOSS-Image's own text stage (what the DiT sees) and the KL of the rewrite — instead of llama.cpp's fixed per-type rules. At the same file size, every DHQ build beats the standard llama.cpp quant on both rulers. Use the **Dual-Head Dragon** ComfyUI node: **https://github.com/pottokao-dotcom/ComfyUI-DualHead-Dragon** (MOSS-Image support: `products/moss.py`).

• EdgeTAM-hf↗

  yonigozlan

  EdgeTAM is an on-device executable variant of the SAM 2 for promptable segmentation and tracking in videos. It runs 22× faster than SAM 2 and achieves 16 FPS on iPhone 15 Pro Max without quantization. EdgeTAM can be used for automatic mask generation to segment all objects in an image using the `mask-generation` pipeline: EdgeTAM's key strength is its ability to track objects across video frames. Here's how to use it for video segmentation:

• NVIDIA-Nemotron-3.5-Bolt-74B-A9B-Abliterated-24GB-GGUF↗

  BoldingBuilds

  An uncensored edit of **NVIDIA Nemotron-3.5-Bolt-74B-A9B**, a hybrid Mamba/MoE model, produced by editing NVIDIA's **shipped 24GB 2-bit GGUF** (nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB-GGUF) **directly** — no BF16 requantize at any point. That matters for this model specifically. Bolt is a LatentMoE: every routed expert sums through one dense `ffn_latent_up` per layer, and the experts themselves are ~17GB of IQ2_XXS weights produced by NVIDIA's quantization-aware distillation. A conventional abliterate-then-requantize pipeline would throw that QAD work away. Instead the projection is applied in place and codes are re-picked under the **original block scales**, so at λ=0 the output file is byte-identical to the input by construction — which is the self-test the build gates on (verified: all edited tensors byte-identical at λ=0). The published file is the same length as NVIDIA's, 22,018,326,528 bytes. Bolt is multimodal, so the projector ships alongside. The edit touched only text-model tensors, so the projector is byte-for-byte NVIDIA's file; download both or vision silently does nothing.

• Hentai-NSFW-LoRA-v3↗

  perpetual3x

  Let's be honest: I know what many of you were hoping to see when you flipped through manga—and it was not the table of contents. Qwen Image 2512 can now turn that long-standing wish into pixels. Apparently dreams do come true; they just need a LoRA, a sufficiently shameless prompt, and 96GB of VRAM. Some datasets choose one style and stay there. This one changed artists, decades, proportions, line weight, and rendering habits before breakfast. Training it felt less like teaching a single illustrator and more like trying to keep an entire convention moving in roughly the same direction. > **Important Anatomy Limitation:** The training dataset contains a very wide > range of visual styles and anatomical conventions. Because the style changes > so substantially from image to image, this LoRA does **not** guarantee > anatomical accuracy. It may reproduce exaggerated proportions, inconsistent > structures, or anatomy that follows a particular drawing style rather than > physical reality. Consider this a documented limitation, not a surprise > anatomy expansion pack.