Open-Weight Models

Model · Text generation

JiRackDeltaNet_27b

Center Business Solutions inc

Benefits high quality CPU inference TQ2 on Llama.cpp and Ollama via QAT - Robotcs, Routing, Coding, Multimedia, Advanced tool calling via JiRackDeltaNetTokenizer - JiRack DeltaNet understand video and images that best for Robotics also A fast and efficient 27B model optimized for CPU inference. Built on a Qwen3.8-style DeltaNet architecture (hybrid attention + SSM), with an updated tokenizer that includes Routing, Media, Vision, Sound, Tool call, and Robotics tags. Ready-to-run GGUF quantizations, and native Ollama support with reasoning disabled by default for fast, direct responses. - JiRack is a cloud-ready model that helps save money on cloud infrastructure. It can be used as an expert…

Model · Video classification

vjepa2-vitg-fpc64-384

AI at Meta

A frontier video understanding model developed by FAIR, Meta, which extends the pretraining objectives of VJEPA, resulting in state-of-the-art video understanding capabilities, leveraging data and model sizes at scale. The code is released in this repository. To run V-JEPA 2 model, ensure you have installed the latest transformers: V-JEPA 2 is intended to represent any video (and image) to perform video classification, retrieval, or as a video encoder for VLMs. To load a video, sample the number of frames according to the model. For this model, we use 64. To load an image, simply copy the image to the desired number of frames. For more code examples, please refer to the V-JEPA 2…

Model · Text generation

Qwen3-1.7B-Thinking-Distil

Convergent Intelligence

Extended Reasoning Distillation from Qwen3-30B-A3B-Thinking → 1.7B The most downloaded model in the Convergent Intelligence portfolio. Qwen3-1.7B-Thinking-Distil captures extended deliberation patterns from the Qwen3-30B-A3B Thinking teacher — the variant that generates long-form reasoning chains before committing to an answer — and compresses them into a 1.7B student via supervised fine-tuning on the longwriter-6k dataset. The Thinking teacher produces the richest signal of the three teacher variants in the DistilQwen family (Instruct, Thinking, Coder). Where Instruct distillation captures clean instruction-following and Coder captures hierarchical decomposition, Thinking distillation…

Model · Text generation

Qwen3.5-2B-CyberSec

Convergent Intelligence

An English Qwen3.5 2B checkpoint associated with the Trendyol Cybersecurity Instruction Tuning Dataset and exported in Transformers / Safetensors format. This release is intended for research and local experimentation. The repository does not currently publish benchmark or safety-evaluation results, so the model should not be treated as a validated cybersecurity authority. The configuration identifies a Qwen3.5 conditional-generation architecture with text and vision components. Use a recent Transformers release that supports this architecture. Dependency and device behavior can vary across Transformers versions. Pin a tested environment for reproducible use. - Research on small-model…

Model · Text generation

Qwen3-1.7B-Distilled-30B-A3B

Convergent Intelligence

A 1.7B-parameter causal language model distilled from Qwen3-30B-A3B on 6,122 STEM chain-of-thought samples using discrepancy-informed knowledge distillation. The training objective emphasizes proof structure, detects reasoning pivot tokens through token-level divergence dynamics, smooths high-entropy student singularities before distillation, and monitors structural drift through discrepancy energy. Standard knowledge distillation treats all tokens uniformly. Even proof-weighted approaches typically apply a static multiplier over the entire derivation span. That helps, but it still misses the internal structure of reasoning: some regions are smooth procedural continuation, while others are…

Model · Text generation

Qwen3-1.7B-Coder-Distilled-SFT

Convergent Intelligence

A 1.7B model built in two stages: knowledge distillation from a 30B Coder teacher to establish a structured reasoning backbone, then supervised fine-tuning on ~54,600 logical inference problems. The Coder teacher's decomposition patterns meet formal propositional logic. The hypothesis: a model that learned STEM derivation from a Coder teacher (Stage 1) already has latent structure for sequential logic, state tracking, and compositional reasoning. Logical inference SFT (Stage 2) activates that structure explicitly — the model doesn't learn logic from scratch, it surfaces what the Coder teacher already gave it. Qwen3-1.7B distilled from Qwen3-Coder-30B-A3B-Instruct — the coding-specialized…

Model · Summarization

distilbart-news-summarizer

Sachin Rao

Imagine you have a very long news article, and you want someone to read it and tell you the key points in just a few sentences. That's exactly what this model does! This model takes a long news article and turns it into a short, easy-to-read summary. - It writes back a 3-4 sentence summary that captures all the important information The special thing about this model is that it's: 1. Very accurate - It understands news writing style very well 2. Very fast - It works quickly even on regular computers (not just expensive AI servers) 3. Specialized in news - It was trained specifically on news articles, so it understands how journalists write 4. Good with financial news - It knows market…

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.

UlizaLlama is a 7B Parameters language model that builds upon the foundation of Jacaranda/kiswallama-pretrained. Jacaranda/kiswallama-pretrained is a large language model continually-pretrained with 321,530,045 swahili tokens and a customized tokenizer with a swahili vocabulary of 20,000 tokens to extend the capabilities of Meta/Llama2. It offers significant improvements in both encoding and decoding for Swahili text, surpassing the Swahili performance of Meta/Llama2. Moreover, Jacaranda/kiswallama-pretrained excels in providing accurate next-word completions in Swahili, a capability which Meta/Llama2 falls short of. UlizaLlama is optimized for downstream tasks, notably those demanding…

Model · Text generation

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

Convergent Intelligence

A 0.6B parameter model built in two stages: knowledge distillation from a 30B Thinking teacher to establish a structured reasoning backbone, then supervised fine-tuning on legal instruction data. 50x compression. Under 500MB quantized. Runs on a phone. The training order is the thesis: teach the model how to reason first (distillation from Thinking teacher), then teach it what to reason about (legal SFT). The Thinking teacher's extended deliberation traces transfer deeper reasoning structure than an Instruct teacher — critical when the student has only 0.6B parameters to work with. Qwen3-0.6B distilled from Qwen3-30B-A3B-Thinking-2507 — a Mixture-of-Experts model with 30B total parameters…

Model · Image and text to text

openvla-7b-finetuned-libero-goal

OpenVLA Collaboration

This model was produced by fine-tuning the OpenVLA 7B model via LoRA (r=32) on the LIBERO-Goal dataset from the LIBERO simulation benchmark. We made a few modifications to the training dataset to improve final performance (see the OpenVLA paper for details). Below are the hyperparameters we used for all LIBERO experiments: - No gradient accumulation (i.e. gradaccumulationsteps == 1) - shufflebuffersize == 100000 See the OpenVLA GitHub README for instructions on how to run and evaluate this model in the LIBERO simulator.

Model · Text generation

Qemma-sft

Convergent Intelligence

Qemma is a HuggingFace-native hybrid model that merges Gemma-3 (1B) and Qwen-3 (0.6B) at the weight level (no adapters). Design: Gemma MLP/body + Qwen attention/head, projected and aligned to Gemma’s hidden size. The model is then SFT-tuned for stepwise reasoning. Use: research, instruction following, code/help, analysis, further SFT/RLHF. Limits: may hallucinate; not for safety-critical, medical, legal, or financial decisions. Follow dataset/model licenses. ~512 warm-start steps (Alpaca-style data) 256 Additional pretraining steps on (O1-OPEN/OpenO1-SFT) 128 SFT steps with (Jackrong/gpt-oss-120b-reasoning-STEM-5K) 256 SFT steps with (O1-OPEN/OpenO1-SFT) This model is part of the Convergent…

Model · Zero-shot classification

ELISARCyberAIEdge7B-LoRA-GGUF

Dr Sabri ALLANI

Offline-ready, quantized LLaMA edge model for cybersecurity use cases ELISAR: An Adaptive Framework for Cybersecurity Risk Assessment Powered by GenAI - Sabri ALLANI, PhD – AI & Cybersecurity Expert - Karam BOU-CHAAYA, PhD – AI & Cybersecurity Expert - Helmi RAIS – Global Practice Lead, Expleo France May 31, 2025 This work will be published by Springer in the following book: ELISAR is a fine-tuned LoRA model based on Mistral-7B, designed for contextualized cybersecurity risk assessment using Retrieval-Augmented Generation and Agentic AI capabilities. The model targets real-world use cases including: ELISARCyberAIEdge7B-LoRA-GGUF is a LoRA-finetuned, GGUF-quantized version of the Mistral-7B…

Model · Text generation

Qwen3-0.6B-Distilled-30B-A3B

Convergent Intelligence

A 0.6B parameter model distilled from Qwen3-30B-A3B-Thinking on 6,122 STEM chain-of-thought samples. 50x parameter compression. The Thinking variant teacher produces richer extended reasoning traces than the Instruct variant, transferring deeper deliberation structure into the smallest possible student. The result: a model under 500MB quantized that produces structured STEM derivations because a 30B thinking model showed it how to reason. Two key differences from standard small-model distillation: 1. Thinking teacher, not Instruct teacher. The Qwen3-30B-A3B-Thinking variant generates extended internal reasoning before committing to an answer. Its softmax distributions are higher-entropy…

Model · Robotics

xvla-base

LeRobot

X-VLA is a Vision-Language-Action foundation model that uses soft prompts to handle cross-embodiment and cross-domain robot control within a unified Transformer architecture. Original paper: X-VLA: Soft-Prompted Transformer as Scalable Cross-Embodiment Vision-Language-Action Model For full installation details (including optional video dependencies such as ffmpeg for torchcodec), see the official documentation: https://huggingface.co/docs/lerobot/installation If you’re training / fine-tuning, you typically call forward(...) to get a loss and then: - -policy.chunksize=... - -policy.nactionsteps=... - -policy.maxactiontokens=... - -policy.gradientcheckpointing=true You can use the record…

Model · Text generation

Qemma-redux

Convergent Intelligence

Redux This Model underwent an additional merge between Qemma-sft and Qwen3-0.6B, in addition to adding Rope Scaling. Qemma is a HuggingFace-native hybrid model that merges Gemma-3 (1B) and Qwen-3 (0.6B) at the weight level (no adapters). Design: Gemma MLP/body + Qwen attention/head, projected and aligned to Gemma’s hidden size. The model is then SFT-tuned for stepwise reasoning. This variant uses Yarn based Rope Scaling with 1:1 Ratio from maxpositionembeddings Use: research, instruction following, code/help, analysis, further SFT/RLHF. Limits: may hallucinate; not for safety-critical, medical, legal, or financial decisions. Follow dataset/model licenses. ~512 warm-start steps (Alpaca-style…

Model · Robotics

Giga-World-Policy-0.5

GigaAI

The MoT design keeps a visual expert stream (reference + future latents) and an action expert stream (state + action), with multi-modal self-attention across both. Weights are sharded at ~10GB per file. This repo contains the transformer only; runtime also needs the Wan2.2 VAE / scheduler from the base Diffusers checkpoint. For usage, training, and inference details, see our open source code page.

DermLIP is a vision-language model for dermatology, trained on the Derm1M dataset—the largest dermatological image-text corpus to date. python import openclip from PIL import Image import torch model,, preprocess = openclip.createmodelandtransforms( 'hf-hub:redlessone/DermLIPViT-B-16' model.eval() tokenizer = openclip.gettokenizer('hf-hub:redlessone/DermLIPViT-B-16') image = preprocess(Image.open("yourskinimage.png")).unsqueeze(0) PADCLASSNAMES = [ "nevus", "basal cell carcinoma", "actinic keratosis", "seborrheic keratosis", "squamous cell carcinoma", "melanoma" template = lambda c: f'This is a skin image of {c}' text = tokenizer([template(c) for c in PADCLASSNAMES]) with torch.nograd()…

Model · Summarization

ptt5-base-summ

Recogna NLP

PTT5 Summ is a fine-tuned PTT5 model to perform Abstractive Summarization in Brazilian Portuguese texts. This model was fine-tuned on the datasets: RecognaSumm, WikiLingua, XL-Sum, TeMário.pdf) and CSTNews. For further information, please go to PTT5 Summ repository.

Model · Text generation

Qemma-Q14B

Convergent Intelligence

My mathematical formulation to utilize space projections to "measure" the Jump between points of discontinuity found in Non-Differentialable Functions. This Model underwent an additional merge between Qemma-redux and Qwen3-14B, in addition to adding Rope Scaling. Fusion Logic was updated to aid per layer fusion and post fusion embedding alignment. Qemma is a HuggingFace-native hybrid model that merges Gemma-3 (1B) and Qwen-3 (14B) at the weight level (no adapters). This variant uses Yarn based Rope Scaling with 1: Ratio from maxpositionembeddings = 524288 Gemma-3 backbone (26 layers, hidden 1152, MLP 6912) Qwen-style attention regrouped to Gemma’s 4×256 heads. (headdim=128, hidden=5120…

Model · Summarization

bert-small2bert-small-finetuned-cnn_daily_mail-summarization

Manuel Romero

This model is a warm-started BERT2BERT (small) model fine-tuned on the CNN/Dailymail summarization dataset. The model achieves a 17.37 ROUGE-2 score on CNN/Dailymail's test dataset. For more details on how the model was fine-tuned, please refer to this notebook.

Model · Zero-shot classification

deberta-v3-large-zeroshot-v1.1-all-33

Moritz Borrett-Laurer (formerly Laurer)

The model is designed for zero-shot classification with the Hugging Face pipeline. The model can do one universal classification task: determine whether a hypothesis is "true" or "not true" given a text (entailment vs. notentailment). This task format is based on the Natural Language Inference task (NLI). The task is so universal that any classification task can be reformulated into this task. A detailed description of how the model was trained and how it can be used is available in this paper. The model was trained on a mixture of 33 datasets and 387 classes that have been reformatted into this universal format. 1. Five NLI datasets with ~885k texts: "mnli", "anli", "fever", "wanli"…

Model · Image and text to text

RynnBrain-2B

DAMO Academy

[2026.02.02] Release RynnBrain family weights and inference code. - [2026.02.02] Add cookbooks for cognition, localization, reasoning, and planning. RynnBrain aims to serve as a physics-aware embodied brain: it observes egocentric scenes, grounds language to physical space and time, and supports downstream robotic systems with reliable localization and planning outputs. - Comprehensive egocentric understanding Strong spatial comprehension and egocentric cognition across embodied QA, counting, OCR, and fine-grained video understanding. - Diverse spatiotemporal localization Locates objects, target areas, and predicts trajectories across long episodic context, enabling global spatial…

Model · Text generation

Qemma-Q1.7B

Convergent Intelligence

My mathematical formulation to utilize space projections to "measure" the Jump between points of discontinuity found in Non-Differentialable Functions. This Model underwent an additional merge between Qemma-redux and Qwen3-1.7B, in addition to adding Rope Scaling. Fusion Logic was updated to aid per layer fusion and post fusion embedding alignment. Qemma is a HuggingFace-native hybrid model that merges Gemma-3 (1B) and Qwen-3 (1.7B) at the weight level (no adapters). This variant uses Yarn based Rope Scaling with 1: Ratio from maxpositionembeddings = 242144 Gemma-3 backbone (26 layers, hidden 1152, MLP 6912) Qwen-style attention regrouped to Gemma’s 4×256 heads. (headdim=128, hidden=2048…