Open-Weight Models

Model · Question answering

bert-base-spanish-wwm-cased-finetuned-spa-squad2-es

Manuel Romero

This model is provided by BETO team and fine-tuned on SQuAD-es-v2.0 for Q&A downstream task. Language model ('dccuchile/bert-base-spanish-wwm-cased'): BETO is a BERT model trained on a big Spanish corpus. BETO is of size similar to a BERT-Base and was trained with the Whole Word Masking technique. Below you find Tensorflow and Pytorch checkpoints for the uncased and cased versions, as well as some results for Spanish benchmarks comparing BETO with Multilingual BERT as well as other (not BERT-based) models. The model was trained on a Tesla P100 GPU and 25GB of RAM with the following command: 1. Set the context and ask some questions: 2. Run predictions

Model · Text generation

SMOLM2Prover-GGUF

Convergent Intelligence

GGUF quantized version of the SMOLM2Prover model for use with llama.cpp and compatible runtimes. Create a Modelfile: 1. Download SMOLM2Prover-Q4KM.gguf 3. Load and chat! The Q4KM quantization uses: This model is part of the Convergent Intelligence LLC: Research Division portfolio. All models in this portfolio are developed under the Discrepancy Calculus (DISC) framework — a measure-theoretic approach to understanding and controlling the gap between what a model should produce and what it actually produces. DISC treats training singularities (loss plateaus, mode collapse, catastrophic forgetting) not as failures to be smoothed over, but as structural signals that reveal the geometry of the…

Model · Video classification

videomae-small-finetuned-kinetics

Multimedia Computing Group-Nanjing University

VideoMAE model pre-trained for 1600 epochs in a self-supervised way and fine-tuned in a supervised way on Kinetics-400. It was introduced in the paper VideoMAE: Masked Autoencoders are Data-Efficient Learners for Self-Supervised Video Pre-Training by Tong et al. and first released in this repository. Disclaimer: The team releasing VideoMAE did not write a model card for this model so this model card has been written by the Hugging Face team. VideoMAE is an extension of Masked Autoencoders (MAE) to video. The architecture of the model is very similar to that of a standard Vision Transformer (ViT), with a decoder on top for predicting pixel values for masked patches. Videos are presented to…

Model · Video classification

videoprism-lvt-large-f8r288

Google

Blog: https://research.google/blog/videoprism-a-foundational-visual-encoder-for-video-understanding/ VideoPrism is a foundational video encoder that enables state-of-the-art performance on a large variety of video understanding tasks. It takes video frames as input and outputs compact embeddings of the frames, which one can conveniently feed into classifiers, LLMs, retrieval models, etc. When tested on 33 public video understanding benchmarks over four task categories, a single frozen VideoPrism checkpoint outperforms previous best-performing foundation models on 31 of them, with no fine-tuning on target task datasets. We release the following model variants: VideoPrism-B/L are the…

Model · Question answering

Gecko-110m-en

LiteRT Community (FKA TFLite)

This model provides a few variants of the embedding model published in the Gecko paper that are ready for deployment on Android or iOS using LiteRT stack or google ai edge RAG SDK. Try out the gecko embedding model in the google ai edge RAG SDK. You can find the SDK on GitHub or follow our android guide to install directly from Maven. We have also published a Use the sentencepiece model as the tokenizer for the Gecko embedding model. Note that all benchmark stats are from a Samsung S23 Ultra. The inference is run on CPU is accelerated via the LiteRT XNNPACK delegate with 4 threads The inference on GPU is accelerated via LiteRT GPU delegate. Benchmark is done assuming XNNPACK cache is…

Model · Question answering

vit5-large

VietAI

State-of-the-art pretrained Transformer-based encoder-decoder model for Vietnamese. For more details, do check out our Github repo.

Model · Video classification

videoprism-lvt-base-f16r288

Google

Blog: https://research.google/blog/videoprism-a-foundational-visual-encoder-for-video-understanding/ VideoPrism is a foundational video encoder that enables state-of-the-art performance on a large variety of video understanding tasks. It takes video frames as input and outputs compact embeddings of the frames, which one can conveniently feed into classifiers, LLMs, retrieval models, etc. When tested on 33 public video understanding benchmarks over four task categories, a single frozen VideoPrism checkpoint outperforms previous best-performing foundation models on 31 of them, with no fine-tuning on target task datasets. We release the following model variants: VideoPrism-B/L are the…

Model · Question answering

dynamic_tinybert

Intel

Dynamic-TinyBERT has been fine-tuned for the NLP task of question answering, trained on the SQuAD 1.1 dataset. Guskin et al. (2021) note: Here is how to import this model in Python

Model · Zero-shot classification

nli-deberta-v3-small

Joshua

https://huggingface.co/cross-encoder/nli-deberta-v3-small with ONNX weights to be compatible with Transformers.js. If you haven't already, you can install the Transformers.js JavaScript library from NPM using: Note: Having a separate repo for ONNX weights is intended to be a temporary solution until WebML gains more traction. If you would like to make your models web-ready, we recommend converting to ONNX using Optimum and structuring your repo like this one (with ONNX weights located in a subfolder named onnx).

Model · Summarization

FRED-T5-Summarizer

Natural Language Processing in Russian

The model was trained by SberDevices. The model is trained on a mix of open summarisation data RussianNLP/Mixed-Summarization-Dataset for the Russian language and use prefix tokenen '\ '

static quants of https://huggingface.co/behbudiy/Llama-3.1-8B-Instruct-Uz For a convenient overview and download list, visit our model page for this model. weighted/imatrix quants seem not to be available (by me) at this time. If they do not show up a week or so after the static ones, I have probably not planned for them. Feel free to request them by opening a Community Discussion. If you are unsure how to use GGUF files, refer to one of TheBloke's READMEs for more details, including on how to concatenate multi-part files. (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants) Here is a handy graph by ikawrakow comparing some lower-quality…

This model is a fine-tuned version of distilbert-base-uncased on the naturalquestions dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 2e-05 - trainbatchsize: 64 - evalbatchsize: 64 - lrschedulertype: linear - numepochs: 40 - Transformers 4.17.0 - Pytorch 1.10.0 - Datasets 1.18.4 - Tokenizers 0.11.6

LinkBERT-large model pretrained on English Wikipedia articles along with hyperlink 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. 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 language model context, besides a single document. LinkBERT can be used as a drop-in replacement for BERT. It achieves…

Model · Text generation

Qwen3-1.7B-Distilled-30B-A3B-SFT-GGUF

Convergent Intelligence

GGUF quantizations of reaperdoesntknow/Qwen3-1.7B-Distilled-30B-A3B-SFT for local and edge deployment via llama.cpp and compatible runtimes. Stage 1 — DISC-Informed Knowledge Distillation: Qwen3-1.7B distilled from Qwen3-30B-A3B-Instruct on 6,122 STEM chain-of-thought samples using proof-weighted cross-entropy loss (2.5x → 1.5x decay on derivation tokens) and KL divergence at T=2.0. The distillation emphasized multi-step reasoning over final-answer pattern matching. Stage 2 — Legal SFT: Follow-up supervised fine-tuning on Alignment-Lab-AI/Lawyer-Instruct to add instruction-following capability and legal domain knowledge on top of the STEM reasoning backbone. The result is a 1.7B model that…

static quants of https://huggingface.co/behbudiy/Llama-3.1-8B-Instruct-Uz For a convenient overview and download list, visit our model page for this model. weighted/imatrix quants seem not to be available (by me) at this time. If they do not show up a week or so after the static ones, I have probably not planned for them. Feel free to request them by opening a Community Discussion. If you are unsure how to use GGUF files, refer to one of TheBloke's READMEs for more details, including on how to concatenate multi-part files. (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants) Here is a handy graph by ikawrakow comparing some lower-quality…

Source code available at https://github.com/phhusson/llm-rl/blob/main/grpo-tldr.py

Model · Text generation

Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT-GGUF

Convergent Intelligence

GGUF quantizations of reaperdoesntknow/Qwen3-0.6B-Distilled-30B-A3B-Thinking-SFT for local, mobile, and edge deployment via llama.cpp and compatible runtimes. A 30B Thinking teacher compressed 50x into a model that fits on a smartwatch. Stage 1 — Thinking Teacher Distillation: Qwen3-0.6B distilled from Qwen3-30B-A3B-Thinking on 6,122 STEM chain-of-thought samples. The Thinking variant teacher produces extended reasoning traces with higher-entropy distributions, transferring richer deliberation structure into the student. Proof-weighted cross-entropy (2.5x → 1.5x on derivation tokens) + KL divergence at T=2.0. Stage 2 — Legal SFT: Supervised fine-tuning on Alignment-Lab-AI/Lawyer-Instruct at…

weighted/imatrix quants of https://huggingface.co/nicoboss/Medra27B For a convenient overview and download list, visit our model page for this model. static quants are available at https://huggingface.co/mradermacher/Medra27B-GGUF This is a vision model - mmproj files (if any) will be in the static repository. If you are unsure how to use GGUF files, refer to one of TheBloke's READMEs for more details, including on how to concatenate multi-part files. (sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants) Here is a handy graph by ikawrakow comparing some lower-quality quant And here are Artefact2's thoughts on the matter…

Model

DualMind-TKD-Agentic-1.7B-GGUF

Convergent Intelligence

This model was finetuned and converted to GGUF format using Unsloth. - DualMind-TKD-Agentic-1.7B.Q6K.gguf - DualMind-TKD-Agentic-1.7B.Q80.gguf - DualMind-TKD-Agentic-1.7B.Q4KM.gguf This was trained 2x faster with Unsloth

Model · Question answering

gelectra-base-germanquad

Deepset

We trained a German question answering model with a gelectra-base model as its basis. - The dataset is GermanQuAD, a new, German language dataset, which we hand-annotated and published online. - The training dataset is one-way annotated and contains 11518 questions and 11518 answers, while the test dataset is three-way annotated so that there are 2204 questions and with 2204·3−76 = 6536answers, because we removed 76 wrong answers. See https://deepset.ai/germanquad for more details and dataset download in SQuAD format. 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…

Model · Question answering

gelectra-large-germanquad

Deepset

We trained a German question answering model with a gelectra-large model as its basis. - The dataset is GermanQuAD, a new, German language dataset, which we hand-annotated and published online. - The training dataset is one-way annotated and contains 11518 questions and 11518 answers, while the test dataset is three-way annotated so that there are 2204 questions and with 2204·3−76 = 6536 answers, because we removed 76 wrong answers. See https://deepset.ai/germanquad for more details and dataset download in SQuAD format. 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…

Model · Summarization

it5-base-news-summarization

Gabriele Sarti

This repository contains the checkpoint for the IT5 Base model fine-tuned on news summarization on the Fanpage and Il Post corpora as part of the experiments of the paper IT5: Large-scale Text-to-text Pretraining for Italian Language Understanding and Generation by Gabriele Sarti and Malvina Nissim. A comprehensive overview of other released materials is provided in the gsarti/it5 repository. Refer to the paper for additional details concerning the reported scores and the evaluation approach. Model checkpoints are available for usage in Tensorflow, Pytorch and JAX. They can be used directly with pipelines as: If you use this model in your research, please cite our work as

Model · Question answering

bert-multi-cased-finetuned-xquadv1

Manuel Romero

This model was created by Google and fine-tuned on XQuAD like data for multilingual (11 different languages) Q&A downstream task. Deepmind XQuAD As the dataset is based on SQuAD v1.1, there are no unanswerable questions in the data. We chose this setting so that models can focus on cross-lingual transfer. We show the average number of tokens per paragraph, question, and answer for each language in the table below. The statistics were obtained using Jieba for Chinese and the Moses tokenizer for the other languages. As XQuAD is just an evaluation dataset, I used Data augmentation techniques (scraping, neural machine translation, etc) to obtain more samples and split the dataset in order to…