source languages: de; target languages: fr; OPUS readme: de-fr; dataset: opus; model: transformer-align; pre-processing: normalization + SentencePiece.
SAVRN Model Hub
Open-Weight Models
An open-weight model is an AI model whose trained weights are published for anyone to download. The weights are what the model learned in training. With a copy of them you can run the model on hardware you control and train it further on your own data.
Open weights are not the same as open source. Many publishers release the weights without the training data or code, and the license sets what you may do with the model. This library puts each model's full card, architecture, files, license and published evaluations on one page.
2,760Models
859Datasets
254Papers
1,692Publishers
5,040Sourced relationships
Updated 2026-09-18 · How the library is built
2,760 models, sorted by most downloaded.
Haystack version 1.x distillation feature was used for training. deepset/bert-large-uncased-whole-word-masking-squad2 was used as the teacher model. Haystack is an AI orchestration framework to build customizable, production-ready LLM applications. You can use this model in Haystack to do extractive question answering on documents. To load and run the model with Haystack: For a complete example with an extractive question answering pipeline that scales over many documents, check out the corresponding Haystack tutorial. - Timo Möller: timo.moeller [at] deepset.ai deepset is the company behind the production-ready open-source AI framework Haystack. We also have a Discord community open to…
NVIDIA Isaac GR00T N1.6 is an open vision-language-action (VLA) model for generalized humanoid robot skills. This cross-embodiment model takes multimodal input, including language and images, to perform manipulation tasks in diverse environments. GR00T N1.6 is trained on a diverse mixture of robot data including bimanual, semi-humanoid and an expansive humanoid dataset, consisting of real captured data, synthetic data generated using the components of NVIDIA Isaac GR00T Blueprint. It is adaptable through post-training for specific embodiments, tasks and environments. The neural network architecture of GR00T N1.6 is a combination of vision-language foundation model and diffusion transformer…
TabFM is a zero-shot tabular foundation model from Google Research. It supports classification and regression on structured/tabular data with mixed numerical and categorical columns, requiring no fine-tuning or hyperparameter search - training examples are passed as context and predictions are made in a single forward pass. This repository contains the PyTorch weights. For the JAX/Flax weights see You can also load directly using the HuggingFace Hub API: Developed by the Google Research team. - Tabular data with numerical and/or categorical columns - Binary and multiclass classification (up to 10 classes) TabFM uses alternating row and column attention to capture both feature interactions…
GGUF packages of Soprano-1.1-80M for audio.cpp. Original model: ekwek1/soprano — an ultra-lightweight (~80M) English-only TTS model by @ekwek1. ` ash python3 tools/modelmanagerv2.py install soprano1180mq80 This conversion is provided as a community package for use with audio.cpp. All rights to the model and its weights belong to the original authors.
BioLinkBERT-large model pretrained on PubMed abstracts along with citation link information. It is introduced in the paper LinkBERT: Pretraining Language Models with Document Links (ACL 2022). The code and data are available in this repository. This model achieves state-of-the-art performance on several biomedical NLP benchmarks such as BLURB and MedQA-USMLE. LinkBERT is a transformer encoder (BERT-like) model pretrained on a large corpus of documents. It is an improvement of BERT that newly captures document links such as hyperlinks and citation links to include knowledge that spans across multiple documents. Specifically, it was pretrained by feeding linked documents into the same…
This model is a fine tuning for the deepfake audio classification task. It achieves the following results on its evalutation data: It achieves the following results on ASVspoof2019 evaluation subset
A HuggingFace-format conversion of Meta AI's V-JEPA 2.1 ViT-L/16 video encoder and predictor, operating at 384x384 resolution. The weights are Meta's, copied without modification. This repository provides the transformers-compatible packaging plus a documented numerical validation against the original implementation. An equivalent community port already exists (Dev-Jahn/vjepa2.1-vitl-fpc64-384). This repository adds an independently reproduced conversion together with the validation results below. Forward outputs match the existing port to all reported digits. The only structural change is that the fused QKV projection of each attention block is split into separate query / key / value…
Mask2Former model trained on ADE20k semantic segmentation (small-sized version, Swin backbone). It was introduced in the paper Masked-attention Mask Transformer for Universal Image Segmentation and first released in this repository. Disclaimer: The team releasing Mask2Former did not write a model card for this model so this model card has been written by the Hugging Face team. Mask2Former addresses instance, semantic and panoptic segmentation with the same paradigm: by predicting a set of masks and corresponding labels. Hence, all 3 tasks are treated as if they were instance segmentation. Mask2Former outperforms the previous SOTA, MaskFormer both in terms of performance an efficiency by (i)…
This is a direct GGUF conversion of Wan-AI/Wan2.1-T2V-14B All quants are created from the FP32 base file, though I only uploaded FP16 due to it exceeding the 50GB max file limit and gguf-split loading not currently being supported in ComfyUI-GGUF. 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. The VAE can be downloaded from this repository by Kijai Please refer to this chart for a basic overview of quantization types.
Model · Image and text to text
Qari-OCR-v0.3-VL-2B-Instruct
QARI-OCR v0.3 is a specialized vision-language model fine-tuned for Arabic Optical Character Recognition with a focus on structural document understanding. - Built on Qwen2-VL-2B-Instruct, this model excels at preserving document layouts, HTML tags, and formatting while transcribing Arabic text. - It is described in detail in the paper QARI-OCR: High-Fidelity Arabic Text Recognition through Multimodal Large Language Model Adaptation. While QARI v0.2 achieves better raw text accuracy (CER: 0.061), QARI v0.3 excels in: - HTML/Markdown structure preservation - Document layout understanding - Handwritten text recognition (initial capabilities) - 5x faster training than v0.2 You can load this…
GGUF conversions of nvidia/canary-1b-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. An 883M-parameter multitask AED with a 32-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 1…
GIT (short for GenerativeImage2Text) model, base-sized version. It was introduced in the paper GIT: A Generative Image-to-text Transformer for Vision and Language by Wang et al. and first released in this repository. Disclaimer: The team releasing GIT did not write a model card for this model so this model card has been written by the Hugging Face team. GIT is a Transformer decoder conditioned on both CLIP image tokens and text tokens. The model is trained using "teacher forcing" on a lot of (image, text) pairs. The goal for the model is simply to predict the next text token, giving the image tokens and previous text tokens. The model has full access to (i.e. a bidirectional attention mask…
News (2025.08) Sundial has been integrated into Apache IoTDB, a native time-series database. News (2025.06) Sundial has been accepted as ICML 2025 Oral (Top 1%). News (2025.05) Get 1st MASE on the GIFT-Eval Benchmark. News (2025.02) Get 1st MSE/MAE zero-shot performance on Time-Series-Library datasets. Sundial is a family of generative time series foundation models. This version is pre-trained on 1 trillion time points with 128M parameters. For more information, please refer to this paper. [[Slides]](https://cloud.tsinghua.edu.cn/f/8d526337afde465e87c9/) [[Poster]](https://cloud.tsinghua.edu.cn/f/cc2a156315e9453f99b3/) [[Intro (CN)]](https://mp.weixin.qq.com/s/y3sc2e2lmW1sqfnoK-ZdDA). The…
SmolVLA is a compact, efficient vision-language-action model that achieves competitive performance at reduced computational costs and can be deployed on consumer-grade hardware. This policy has been trained and pushed to the Hub using LeRobot. See the full documentation at LeRobot Docs. For a complete walkthrough, see the training guide. Below is the short version on how to train and run inference/eval: Writes checkpoints to outputs/train/ /checkpoints/. Prefix the dataset repo with eval\ and supply --policy.path pointing to a local or hub checkpoint.
In this repository, we present Wan2.1, a comprehensive and open suite of video foundation models that pushes the boundaries of video generation. Wan2.1 offers these key features: This repository hosts our T2V-1.3B model, a versatile solution for video generation that is compatible with nearly all consumer-grade GPUs. In this way, we hope that Wan2.1 can serve as an easy-to-use tool for more creative teams in video creation, providing a high-quality foundational model for academic teams with limited computing resources. This will facilitate both the rapid development of the video creation community and the swift advancement of video technology. Your browser does not support the video tag.…
English ｜ 中文 Hy-MT2 is a family of “fast-thinking” multilingual translation models designed for complex real-world scenarios. It includes three model sizes: 1.8B, 7B, and 30B-A3B (MoE), all of which support translation among 33 languages and effectively follow translation instructions in multiple languages. For on-device deployment, AngelSlim 1.25-bit extreme quantization reduces the storage requirement of the 1.8B model to only 440 MB and improves inference speed by 1.5x. Multi-dimensional evaluations show that Hy-MT2 delivers outstanding performance across general, real-world business, domain-specific, and instruction-following translation tasks. The 7B and 30B-A3B models outperform…
The developers of the Text-To-Text Transfer Transformer (T5) write: T5-11B is the checkpoint with 11 billion parameters. The developers write in a blog post that the model: See the blog post and research paper for further details. The model is pre-trained on the Colossal Clean Crawled Corpus (C4), which was developed and released in the context of the same research paper as T5. The model was pre-trained on a on a multi-task mixture of unsupervised (1.) and supervised tasks (2.). Thereby, the following datasets were being used for (1.) and (2.): 1. Datasets used for Unsupervised denoising objective: 2. Datasets used for Supervised text-to-text language modeling objective - CoLA Warstadt et…
Finetuned ai-forever/ruT5-base for text and dialogue summarization. All 'train' subsets was concatenated and shuffled with seed 1000 - 7. Train subset = 155678 rows. Evaluation on 10% of concatenated 'validation' subsets = 1458 rows. See WandB logs. See report at REPORT WIP.
Example workflow - based on the Comfyui example workflow This is a direct GGUF conversion of Wan-AI/Wan2.1-VACE-14B All quants are created from the FP32 base file, though I only uploaded the Q80 and less, if you want the F16 or BF16 one I would upload it per request. 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. The VAE can be downloaded from here Please refer to this chart for a basic overview of quantization types. For conversion I used the conversion scripts from city96
UperNet framework for semantic segmentation, leveraging a Swin Transformer backbone. UperNet was introduced in the paper Unified Perceptual Parsing for Scene Understanding by Xiao et al. Combining UperNet with a Swin Transformer backbone was introduced in the paper Swin Transformer: Hierarchical Vision Transformer using Shifted Windows. Disclaimer: The team releasing UperNet + Swin Transformer did not write a model card for this model so this model card has been written by the Hugging Face team. UperNet is a framework for semantic segmentation. It consists of several components, including a backbone, a Feature Pyramid Network (FPN) and a Pyramid Pooling Module (PPM). Any visual backbone can…
This is a video detailer model on top of LTXV13B098DEV trained on custom data. IC LoRA is a method that enables adding video context into the video generation process. This approach allows for video-to-video control on top of the text-to-video model, providing more precise control over the generated content by conditioning the model on reference video frames during inference. For licensing information, please refer to the LTXV Open Weights License. This model is designed to be used with the LTXV (Lightricks Text-to-Video) pipeline. In order to use the trained lora in comfy: 1. Copy the comfyui trained LoRA weights to the models/loras folder in your ComfyUI installation. 2. Use…
a sentence initial language token is required in the form of >>id<< (id = valid target language ID) - hfname: eng-ara - sourcelanguages: eng - targetlanguages: ara - opusreadmeurl: https://github.com/Helsinki-NLP/Tatoeba-Challenge/tree/master/models/eng-ara/README.md - originalrepo: Tatoeba-Challenge - srcconstituents: {'eng'} - tgtconstituents: {'apc', 'ara', 'arqLatn', 'arq', 'afb', 'araLatn', 'apcLatn', 'arz'} - srcmultilingual: False - tgtmultilingual: False - urlmodel: https://object.pouta.csc.fi/Tatoeba-MT-models/eng-ara/opus-2020-07-03.zip - urltestset: https://object.pouta.csc.fi/Tatoeba-MT-models/eng-ara/opus-2020-07-03.test.txt - srcalpha3: eng - tgtalpha3: ara - shortpair: en-ar…
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 Collections
Hand-picked starting points, each with the reason it exists.
Collection · 4 entries
Embedding models for retrieval
Sentence and document embedding models used to build retrieval systems. Dimension and sequence length matter more than size here, and both come from the publisher.
Collection · 4 entries
Models that fit on one accelerator
Models whose publisher-reported parameter count puts them within reach of a single accelerator at common precisions. Memory needed depends on precision and serving configuration, so treat the parameter count as the starting point, not the answer.
Collection · 6 entries
Open-weight text models worth knowing
Widely used open-weight language models, chosen because each one is a distinct family rather than a variant of the one above it. Selection, not a ranking.
Collection · 3 entries
Speech and audio models
Recognition and synthesis models, grouped so the two directions are easy to compare.
Open-Weight Models Explained
What is an open-weight model?
An AI model whose trained weights are published for anyone to download, so it can be run, tested and fine-tuned on hardware the user controls.
Is an open-weight model the same as open source?
Not always. Open weights means the trained model can be downloaded. Open source usually also means the training code and data are available and the license allows broad reuse. Many open-weight models release the weights only.
Can I use an open-weight model commercially?
It depends on the license. Apache 2.0 and MIT allow commercial use. Other licenses limit it, for example to non-commercial use or below a set number of users. Every model page here shows its license.
How much memory does an open-weight model need?
About two bytes per parameter at 16-bit precision, so a 7-billion-parameter model needs roughly 14 GB for its weights, plus memory for the context it processes. Each model page lists its parameter count and the size of its files.
Related SAVRN Research
The hub sits beside SAVRN's market data and infrastructure research: what models cost to run, and what it takes to run them.
SAVRN Index
What open models cost to run
The same open-weight model priced by every host that serves it, per million tokens.
Research Hub
Data center trackers and maps
Moratoriums, permits, power, water and capital behind the facilities that run these models.
Method
How the Model Hub is built
Sources, evidence labels, refresh behaviour, and the limits of every comparison here.