Open-Weight Models

Model · Image classification

convnextv2-base-22k-384

AI at Meta

ConvNeXt V2 model pretrained using the FCMAE framework and fine-tuned on the ImageNet-22K dataset at resolution 384x384. It was introduced in the paper ConvNeXt V2: Co-designing and Scaling ConvNets with Masked Autoencoders by Woo et al. and first released in this repository. Disclaimer: The team releasing ConvNeXT V2 did not write a model card for this model so this model card has been written by the Hugging Face team. ConvNeXt V2 is a pure convolutional model (ConvNet) that introduces a fully convolutional masked autoencoder framework (FCMAE) and a new Global Response Normalization (GRN) layer to ConvNeXt. ConvNeXt V2 significantly improves the performance of pure ConvNets on various…

Model · Any to any

Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16

NVIDIA

NVIDIA Nemotron 3 Nano Omni is a multimodal large language model that unifies video, audio, image, and text understanding to support enterprise-grade Q&A, summarization, transcription, and document intelligence workflows. It extends the Nemotron Nano family with integrated video+speech comprehension, Graphical User Interface (GUI), Optical Character Recognition (OCR), and speech transcription capabilities, enabling end-to-end processing of rich enterprise content such as meeting recordings, M&E assets, training videos, and complex business documents. NVIDIA Nemotron 3 Nano Omni was developed by NVIDIA as part of the Nemotron model family. This model is available for commercial use. This…

Model · Text to video

FastVideo-FastH3-4-step-Preview-v1-VSA-DataFree

FastVideo

The recommended FastH3 Preview v1 checkpoint from FastVideo. It generates synchronized video and audio from text with four transformer forwards. This step-1300 model was trained with data-free DMD2 and VSA-H3 at 90% sparsity. Install uv, then use the CUDA 13 / Blackwell path below. It selects FastVideo's published CUDA kernel wheel instead of compiling the kernel locally. See the for other platforms. The tested defaults use four B200 GPUs and the trained four-forward schedule. On other multi-GPU CUDA systems, follow the installation guide and add --no-replicated-dit --vsa-kernel triton --no-fa4. The GPU count must divide H3's 56 attention heads. This preview supports text-to-audio-video…

Model · Image classification

vit_base_patch8_224.augreg2_in21k_ft_in1k

PyTorch Image Models

A Vision Transformer (ViT) image classification model. Trained on ImageNet-21k by paper authors and (re) fine-tuned on ImageNet-1k with additional augmentation and regularization by Ross Wightman. - How to train your ViT? Data, Augmentation, and Regularization in Vision Transformers: https://arxiv.org/abs/2106.10270 - An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale: https://arxiv.org/abs/2010.11929v2 Explore the dataset and runtime metrics of this model in timm model results.

Model · Feature extraction

jina-embeddings-v2-base-code

Jina AI

The easiest way to starting using jina-embeddings-v2-base-code is to use Jina AI's Embedding API. jina-embeddings-v2-base-code is an multilingual embedding model speaks English and 30 widely used programming languages. Same as other jina-embeddings-v2 series, it supports 8192 sequence length. jina-embeddings-v2-base-code is based on a Bert architecture (JinaBert) that supports the symmetric bidirectional variant of ALiBi to allow longer sequence length. The backbone jina-bert-v2-base-code is pretrained on the github-code dataset. The model is further trained on Jina AI's collection of more than 150 millions of coding question answer and docstring source code pairs. These pairs were obtained…

Model · Speech recognition

canary-180m-flash-gguf

Handy

GGUF conversions of nvidia/canary-180m-flash for use with transcribe.cpp. Ported from upstream commit pinned 2026-05-08. Validated against the NeMo reference at transcribe.cpp commit Offline multilingual speech-to-text and translation. A 182M-parameter multitask AED with a 17-layer FastConformer encoder and a 4-layer Transformer decoder. Supports automatic speech recognition in English, German, Spanish, and French, and bidirectional EN↔{DE, ES, FR} translation. Takes a 16 kHz mono WAV and produces a transcript. Not a streaming model; word/segment timestamps are upstream-experimental and not exposed in the v1 port. WER on the full LibriSpeech test-clean split (2,620 utterances), batch size…

Model · Fill mask

chinese-bert-wwm

Joint Laboratory of HIT and iFLYTEK Research (HFL)

For further accelerating Chinese natural language processing, we provide Chinese pre-trained BERT with Whole Word Masking. Yiming Cui, Wanxiang Che, Ting Liu, Bing Qin, Ziqing Yang, Shijin Wang, Guoping Hu This repository is developed based on：https://github.com/google-research/bert You may also interested in, If you find the technical report or resource is useful, please cite the following technical report in your paper.

Model · Any to any

gemma-4-31B-it-assistant

Google

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…

Model · Fill mask

esm2_t36_3B_UR50D

AI at Meta

ESM-2 is a state-of-the-art protein model trained on a masked language modelling objective. It is suitable for fine-tuning on a wide range of tasks that take protein sequences as input. For detailed information on the model architecture and training data, please refer to the accompanying paper. You may also be interested in some demo notebooks (PyTorch, TensorFlow) which demonstrate how to fine-tune ESM-2 models on your tasks of interest. Several ESM-2 checkpoints are available in the Hub with varying sizes. Larger sizes generally have somewhat better accuracy, but require much more memory and time to train

Model · Token classification

unbiased-toxic-roberta-onnx

Protect AI

This model is a conversion of unitary/unbiased-toxic-roberta to ONNX format using the Optimum library. Trained models & code to predict toxic comments on 3 Jigsaw challenges: Toxic comment classification, Unintended Bias in Toxic comments, Multilingual toxic comment classification. Built by Laura Hanu at Unitary. The huggingface models currently give different results to the detoxify library (see issue here). All challenges have a toxicity label. The toxicity labels represent the aggregate ratings of up to 10 annotators according the following schema: - Very Toxic (a very hateful, aggressive, or disrespectful comment that is very likely to make you leave a discussion or give up on sharing…

Model · Text to image

FLUX.1-dev-gguf

City

This is a direct GGUF conversion of black-forest-labs/FLUX.1-dev As this is a quantized model not a finetune, all the same restrictions/original license terms still apply. The model files can be used with the ComfyUI-GGUF custom node. Place model files in ComfyUI/models/unet - see the GitHub readme for further install instructions. Please refer to this chart for a basic overview of quantization types.

Model · Translation

opus-mt-es-en

Helsinki-NLP Research Group

hfname: spa-eng - sourcelanguages: spa - targetlanguages: eng - opusreadmeurl: https://github.com/Helsinki-NLP/Tatoeba-Challenge/tree/master/models/spa-eng/README.md - originalrepo: Tatoeba-Challenge - srcconstituents: {'spa'} - tgtconstituents: {'eng'} - srcmultilingual: False - tgtmultilingual: False - urlmodel: https://object.pouta.csc.fi/Tatoeba-MT-models/spa-eng/opus-2020-08-18.zip - urltestset: https://object.pouta.csc.fi/Tatoeba-MT-models/spa-eng/opus-2020-08-18.test.txt - srcalpha3: spa - tgtalpha3: eng - shortpair: es-en - chrF2score: 0.7390000000000001 - brevitypenalty: 0.9740000000000001 - reflen: 79376.0 - srcname: Spanish - tgtname: English - traindate: 2020-08-18 00:00:00…

Model · Token classification

OpenMed-NER-PharmaDetect-SuperClinical-434M

OpenMed

Specialized model for Chemical Entity Recognition - Chemical entities from the BC5CDR dataset This model is a state-of-the-art fine-tuned transformer engineered to deliver enterprise-grade accuracy for chemical entity recognition - chemical entities from the bc5cdr dataset. This specialized model excels at identifying and extracting biomedical entities from clinical texts, research papers, and healthcare documents, enabling applications such as drug interaction detection, medication extraction from patient records, adverse event monitoring, literature mining for drug discovery, and biomedical knowledge graph construction with production-ready reliability for clinical and research…

Model · Text classification

roberta_toxicity_classifier

S NLP

This model is trained for toxicity classification task. The dataset used for training is the merge of the English parts of the three datasets by Jigsaw (Jigsaw 2018, Jigsaw 2019, Jigsaw 2020), containing around 2 million examples. We split it into two parts and fine-tune a RoBERTa model (RoBERTa: A Robustly Optimized BERT Pretraining Approach) on it. The classifiers perform closely on the test set of the first Jigsaw competition, reaching the AUC-ROC of 0.98 and F1-score of 0.76. To acknowledge our work, please, use the corresponding citation: This model is licensed under the OpenRAIL++ License, which supports the development of various technologies—both industrial and academic—that serve…

This model is a fine-tuned version of the DistilBERT model to classify toxic comments. You can use the model with the following code. This model is intended to use for classify toxic online classifications. However, one limitation of the model is that it performs poorly for some comments that mention a specific identity subgroup, like Muslim. The following table shows a evaluation score for different identity group. You can learn the specific meaning of this metrics here. But basically, those metrics shows how well a model performs for a specific group. The larger the number, the better. The table above shows that the model performs poorly for the muslim and jewish group. In fact, you pass…

Model · Text to image

Pony_Diffusion_V6_XL

Narontaka

Pony Diffusion V6 is a versatile SDXL finetune capable of producing stunning SFW and NSFW visuals of various anthro, feral, or humanoids species and their interactions based on simple natural language prompts. CHECK "ABOUT THIS VERSION" ON THE RIGHT IF YOU ARE NOT ON "V6" FOR IMPORTANT INFORMATION. Please join our Discord Server to support development of new versions of this model and get access to free SD bot and check out more examples of this model capabilities on our prompt sharing website or follow the author on Twitter. Important information Make sure you load this model with clip skip 2 (or -2 in some software), otherwise you will be getting low quality blobs. This model supports a…

Model · Text to speech

Higgs-Audio-v3-Studio

DRBAPH

This repository hosts the downloadable runtime files for Higgs Audio v3 Studio, a Windows desktop app for local Higgs Audio v3 TTS, voice cloning, speech continuation, and multi-speaker generation. This repository is not the original upstream model release. It provides GGUF model builds, the Windows CUDA engine DLL package, checksums, and a manifest used by the desktop app downloader. Higgs Audio v3 Studio is a Rust/Tauri desktop application that runs a ported native C++/CUDA implementation of Higgs Audio v3 locally. - Engine dependency diagnostics for missing CUDA/MSVC runtime DLLs Use the desktop app from GitHub releases: https://github.com/Saganaki22/Higgs-Audio-v3-Studio/releases…

In today's digital world, user-generated content is a double-edged sword. While it fosters creativity and engagement, it also opens the door to inappropriate or illegal content being shared. Our NSFW Image Classifier is specifically designed to identify and filter out explicit images, including pornography, hentai, and sexually suggestive content, ensuring your platform remains safe, secure, and legally compliant. With more than 2M downloads, our NSFW Image Classifier has become the go-to solution for platforms looking to maintain a clean and safe environment for their users. Many developers and companies have already chosen our solution to protect their communities—will you be next? 1.…

Model · Zero-shot classification

mDeBERTa-v3-base-mnli-xnli

Moritz Borrett-Laurer (formerly Laurer)

This multilingual model can perform natural language inference (NLI) on 100 languages and is therefore also suitable for multilingual zero-shot classification. The underlying model was pre-trained by Microsoft on the CC100 multilingual dataset. It was then fine-tuned on the XNLI dataset, which contains hypothesis-premise pairs from 15 languages, as well as the English MNLI dataset. As of December 2021, mDeBERTa-base is the best performing multilingual base-sized transformer model, introduced by Microsoft in this paper. If you are looking for a smaller, faster (but less performant) model, you can try multilingual-MiniLMv2-L6-mnli-xnli. This model was trained on the XNLI development dataset…

Model · Sentence similarity

rubert-tiny2

David Dale

This is an updated version of cointegrated/rubert-tiny: a small Russian BERT-based encoder with high-quality sentence embeddings. This post in Russian gives more details. The differences from the previous version include: - sentence embeddings approximate LaBSE closer than before; - meaningful segment embeddings (tuned on the NLI task) - the model is focused only on Russian. The model should be used as is to produce sentence embeddings (e.g. for KNN classification of short texts) or fine-tuned for a downstream task. Sentence embeddings can be produced as follows: Alternatively, you can use the model with sentencetransformers: For those who want to run the inference with VLLM, there is a…

Model · Text classification

twitter-roberta-base-sentiment

Cardiff NLP

This is a roBERTa-base model trained on ~58M tweets and finetuned for sentiment analysis with the TweetEval benchmark. This model is suitable for English (for a similar multilingual model, see XLM-T). 0 -> Negative; 1 -> Neutral; 2 -> Positive See twitter-roberta-base-sentiment-latest and TweetNLP for more details. Please cite the reference paper if you use this model.

Model · Token classification

OpenMed-PII-SuperClinical-Small-44M-v1

OpenMed

is a transformer-based token classification model fine-tuned for Personally Identifiable Information (PII) detection in text. This model identifies and classifies 54 types of sensitive information including names, addresses, SSNs, medical record numbers, and more. Evaluated on a stratified 2,000-sample test set from NVIDIA Nemotron-PII: These entity types have lower performance and may benefit from additional post-processing: This model detects 54 PII entity types organized into categories: Important: This model is intended as an assistive tool, not a replacement for human review.

Model

IndicBART

AI4Bharat

IndicBART is a multilingual, sequence-to-sequence pre-trained model focusing on Indic languages and English. It currently supports 11 Indian languages and is based on the mBART architecture. You can use IndicBART model to build natural language generation applications for Indian languages by finetuning the model with supervised training data for tasks like machine translation, summarization, question generation, etc. Some salient features of the IndicBART are: You can read more about IndicBART in this paper. For detailed documentation, look here: https://github.com/AI4Bharat/indic-bart/ and https://indicnlp.ai4bharat.org/indic-bart/ We used the IndicCorp data spanning 12 languages with 452…

Model

mt5-small

Google

mT5 is pretrained on the mC4 corpus, covering 101 languages: Afrikaans, Albanian, Amharic, Arabic, Armenian, Azerbaijani, Basque, Belarusian, Bengali, Bulgarian, Burmese, Catalan, Cebuano, Chichewa, Chinese, Corsican, Czech, Danish, Dutch, English, Esperanto, Estonian, Filipino, Finnish, French, Galician, Georgian, German, Greek, Gujarati, Haitian Creole, Hausa, Hawaiian, Hebrew, Hindi, Hmong, Hungarian, Icelandic, Igbo, Indonesian, Irish, Italian, Japanese, Javanese, Kannada, Kazakh, Khmer, Korean, Kurdish, Kyrgyz, Lao, Latin, Latvian, Lithuanian, Luxembourgish, Macedonian, Malagasy, Malay, Malayalam, Maltese, Maori, Marathi, Mongolian, Nepali, Norwegian, Pashto, Persian, Polish…