Saved toolsSubmit a tool
FOXAI SEARCH

Search FoxAI

AI Models 2119

  • MOSS-Image_4-step_adapter_lora↗

    darask0

    **English summary.** A 4-step, CFG-free adapter for OpenMOSS-Team/MOSS-Image-SFT (50 steps × 2 forwards with CFG 3). It continues lodestones/MOSS-Image-SFT-8-steps-dmd (data-space DMD, arXiv 2610.12102) restricted to 4 iterations, trained with DMAD (arXiv 2610.02188): a LoRA critic on the frozen teacher separates teacher samples from student samples, and the student minimises the negative critic logit. DiT time per 1024² image on one H100: 16.2 s (teacher) → 0.81 s. Compared with the lodestones checkpoint run for 4 steps, images have more detail and livelier composition, with some regressions (see Results). Same file format and sampler as the lodestones checkpoint; `inference.py` runs it with the release pipeline. Training code is in `code/`. MOSS-Image-SFT(50 ステップ・CFG 3 で 1 枚 100 回の DiT 計算)を、**4 回の計算・CFG なし**で画像を出せるようにする高速化アダプタです。 lodestones の 8 ステップ版(data-space DMD、update 2600)を出発点に、4 ステップに絞って DMAD で追加学習しました。 カスタムノード `comfyui/rapid-moss/` を同梱しています。ComfyUI 本体も MOSS-Image に対応していますが、専用サンプラーと time path の置き換えはなく、標準の `LoraLoaderModelOnly` は LoRA を bf16 の重みに足し込みます(当たる差分の相対誤差 81%)。標準ノードだけでも絵は出ますが、学習した生成器とは別物になります(下の比較)。このノードを使ってください。

  • sesame-csm-1b-GGUF↗

    ggml-org

  • Qwen3-4B-SafeRL↗

    阿里巴巴 · Qwen

    **Qwen3-4B-SafeRL** is a safety-aligned version of the Qwen3-4B model. It has been trained using Reinforcement Learning (RL) with a reward signal from **Qwen3Guard-Gen** to enhance its robustness against harmful or adversarial prompts. This process aims to ensure strong safety guarantees without leading to overly simplistic or evasive refusal behaviors, thereby maintaining a positive user experience. To prevent the model from defaulting to refusal across all prompts in an attempt to remain safe, we introduce a hybrid reward function that jointly optimizes three key objectives: Qwen3-4B-SafeRL is used in the same way as Qwen3-4B, preserving the ability of hybrid thinking modes. The code of Qwen3 has been in the latest Hugging Face `transformers` and we advise you to use the latest version of `transformers`.

  • GLM-4-32B-0414-GGUF↗

    unsloth

    The GLM family welcomes new members, the **GLM-4-32B-0414** series models, featuring 32 billion parameters. Its performance is comparable to OpenAI’s GPT series and DeepSeek’s V3/R1 series. It also supports very user-friendly local deployment features. GLM-4-32B-Base-0414 was pre-trained on 15T of high-quality data, including substantial reasoning-type synthetic data. This lays the foundation for subsequent reinforcement learning extensions. In the post-training stage, we employed human preference alignment for dialogue scenarios. Additionally, using techniques like rejection sampling and reinforcement learning, we enhanced the model’s performance in instruction following, engineering code, and function calling, thus strengthening the atomic capabilities required for agent tasks. GLM-4-32B-0414 achieves good results in engineering code, Artifact generation, function calling, search-based Q&A, and report generation. In particular, on several benchmarks, such as code generation or specific Q&A tasks, GLM-4-32B-Base-0414 achieves comparable performance with those larger models like GPT-4o and DeepSeek-V3-0324 (671B). **GLM-Z1-32B-0414** is a reasoning model with deep thinking capabilities. This was developed based on GLM-4-32B-0414 through cold start, extended reinforcement learning, and further training on tasks including mathematics, code, and logic. Compared to the base model, GLM-Z1-32B-0414 significantly improves mathematical abilities and the capability to solve complex tasks. During training, we also introduced general reinforcement learning based on pairwise ranking feedback, which enhances the model's general capabilities. **GLM-Z1-Rumination-32B-0414** is a deep reasoning model with rumination capabilities (against OpenAI's Deep Research). Unlike typical deep thinking models, the rumination model is capable of deeper and longer thinking to solve more open-ended and complex problems (e.g., writing a comparative analysis of AI development in two cities and their future development plans). Z1-Rumination is trained through scaling end-to-end reinforcement learning with responses graded by the ground truth answers or rubrics and can make use of search tools du

  • vakyansh-wav2vec2-hindi-him-4200↗

    Harveenchadha

    Fine-tuned on Multilingual Pretrained Model CLSRIL-23. The original fairseq checkpoint is present here. When using this model, make sure that your speech input is sampled at 16kHz. **Note: The result from this model is without a language model so you may witness a higher WER in some cases.** This model was trained on 4200 hours of Hindi Labelled Data. The labelled data is not present in public domain as of now.

  • flux-3-image-klein↗

    ml-resources

    > Ungated mirror of black-forest-labs/flux-3-image-klein. All credit to Black Forest Labs. This is not an official Black Forest Labs release. The model is licensed under the FLUX Kommunity License (non-commercial), included here as `LICENSE.md` with `NOTICE.md`; any rights to use it are granted directly by Black Forest Labs under that license. FLUX 3 Image Klein is a 5.6B open weights model for text-to-image generation and image editing. It takes a prompt and up to 10 reference images and returns one image. Text-to-image, single-reference edits and multi-reference compositions run through the same model, with Qwen3-VL-4B-Instruct as the text encoder. This repository holds the distilled checkpoint for generation and image editing. For fine-tuning, LoRA training and research, use FLUX 3 Image Klein Base.

  • sam2.1-hiera-tiny-ONNX↗

    onnx-community

  • Xiaomi-OCR-0-GGUF↗

    prithivMLmods

    > Xiaomi-OCR-0 is a compact 0.8B OCR-specialized vision-language model based on Qwen3.5-0.8B-Base, designed for unified document parsing and OCR-centric understanding. It is trained on an approximately 170M-sample OCR-focused corpus using Q-Mask text anchoring, continued pretraining, and mixed-task reinforcement learning, enabling document parsing, OCR-based visual question answering, and key information extraction. The model is designed to handle standard, scanned, warped, photographed, and skewed documents, while also supporting structured outputs such as Markdown, OTSL tables, LaTeX formulas, and JSON-based information extraction. **Xiaomi-OCR-0** on Hugging Face — https://huggingface.co/SeerRay-Lab/Xiaomi-OCR-0.

  • Carnice-Qwen3.6-MoE-35B-A3B-GGUF↗

    samuelcardillo

  • Taeh3-Comfy↗

    t8star

    Kijai原仓库也把2D权重命名为`taeh3.safetensors`，这里**只改文件名、不改张量**，避免覆盖时序版。 已有时序文件请先核对SHA256，不要拿2D版覆盖；这是上游两种不同结构，不可随意互换。 默认选择`taeh3.safetensors`、`phase=low`、每2步更新、最长边256。 要用2D版，在`checkpoint`主动选择`taeh3_2d_kijai.safetensors`。 原sigmas、采样器、步数、最终VAE和声音接线不变；`enabled=false`直接旁路。 预览只观察实际callback的x0开头片段；缓存命中没有新callback时不会制造实时预览。 面板“取消”仅针对绑定的当前请求，不清其他队列、不关闭ComfyUI服务。 预览近似图不代表最终清晰度、整片动作、音频或接缝；2D轮播速度不是源视频帧率。 Merge this repository's `models/` into `ComfyUI/models/`. The default file is the original temporal TAEH3 decoder. Kijai's 2D checkpoint is renamed **without tensor changes** to `taeh3_2d_kijai.safetensors`; explicitly choose it in the T8 preview node. Never overwrite the temporal checkpoint with the 2D model. Insert the MODEL observer before the existing sampler; final VAE/audio/sigmas remain unchanged. Approximate previews are not final videos. MIT applies to the temporal file; Apache-2.0 applies to the 2D file. See the license files.