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.
Updated 2026-10-18 · How the library is built
11,122 models, sorted by most downloaded.
Object-hallucination probe trained on a LLaVA-1.5-13B COCO corpus, with three independent training seeds (0, 42, 1337). It is not interchangeable with the 7B release: features, geometry and thresholds are model-specific. 10,000 COCO train2014 images split 80/20 by image: 8,000 train / 54,253 object spans and 2,000 holdout / 13,789 object spans. Captions are generated by 13B itself, so this corpus is not the 7B corpus. Thresholds are selected by best F1 on the COCO validation split and then frozen. Nothing is tuned on the test holdout, AMBER, or POPE. COCO test holdout — mention view, n=13,787, 16.19% positive AMBER, strict zero-shot — n=2,367 rows / 977 images, 5.37% positive, COCO…
This is the model card of a transformers model that has been pushed on the Hub. Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Use the code below to get started with the model. Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).
License: Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) (Strictly Non-Commercial / Private) Offline Deployment: Ready for 1-click GGUF quantization (< 1 GB RAM footprint under Q4KM). 1. Modern Criminal Jurisprudence: Bharatiya Nyaya Sanhita (BNS 2023), BNSS 2023, BSA 2023, electronic evidence certification (Sec 63 BSA), first-time undertrial bail (Sec 479 BNSS). 2. Financial & Tax Law: GST Input Tax Credit rules (Sec 16(2) CGST), RBI Master Directions (KYC, lending norms), SARFAESI possession procedures, IBC 2016 insolvency (Sec 7/9). 3. Multi-State Vernacular & Constitutional Rights: Landmark Supreme Court precedents (Puttaswamy privacy, Maneka Gandhi…
Three single-speaker Vietnamese narrator voices in Piper/VITS format (22,050 Hz, ~63 MB each), ready for sherpa-onnx: the.onnx files carry sherpa-onnx metadata and each folder ships a tokens.txt. Phonemization uses eSpeak NG (vi); espeak-ng-data is not included. None of the voices belongs to a real person: each was designed from a text description with VoxCPM2 Voice Design. 1. Training data. For each voice, roughly 5,000 Vietnamese sentences were synthesized with VoxCPM2 (Apache-2.0). About 30% of the sentences contain English names and terms, so the voices learn eSpeak's English phonemes. Every clip was transcribed back with Whisper, and clips with a high character error rate, an…
Sentiment (negative / neutral / positive) for Vietnamese student feedback about university courses. PhoBERT-base fine-tuned on UIT-VSFC with diacritic and teencode augmentation and, at every step, trained to give a real informally typed comment the label a frozen teacher gives its normalized form. Real typing changes 12% of its labels, against 18% for the same model without that training. Shipped as the ONNX FP32 graph the ViFeedback service runs, plus the PyTorch checkpoint it was exported from. Code, evaluation protocol and every number below: https://github.com/ThanhDatVN/ViFeedback-NLP-Service The model expects pyvi word-segmented, lowercase text (giảngviên nhiệttình), at most 96…
A GGUF-quantized Mixture-of-Experts (MoE) variant of the Vigyan-7B STEM model, designed for efficient on-device inference. - For research and non-commercial use only (CC-BY-NC-4.0) - Experimental MoE architecture — use production Vigyan-7B-STEM-Instruct for stable inference If you use this model, please cite the Vigyan AI project
Sovereign DARE-TIES Merged & DPO-Aligned Multi-Disciplinary SLM License: Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) (Strictly Private ) Tested against 100 completely unseen frontier Ph.D. and Olympiad examination problems: 1. DARE-TIES Weight Delta Fusion: Merged 4 specialized domain LoRA adapters (phd-pure-math, silicon-rtl-eda, astro-logic, biomed-chem) into base parameters without catastrophic interference. 2. Dynamic DPO Alignment: Aligned on shreyansh12183/vigyan-gen2-7b-alignment-dataset with strict negative prompt loss masking. 3. 100% Offline Edge Ready: Quantized to Q4KM (~4.3 GB) for air-gapped workstations and defense labs.
Sovereign Direct Preference Optimization (DPO) Aligned MoE Foundation License: Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) (Strictly Private ) Base Architecture: allenai/OLMoE-1B-7B-0924-Instruct (Sparse 64-Expert Top-8 Routing) 1. Contrasting Negative Prompt Gate Calibration: Router weights calibrated across 50,000 domain samples (vigyan-olmoe-master-curriculum-v1) to prevent expert cross-bleeding. 2. Selective Router Precision Guardrail: Router gating matrices (ffngateinp / gateproj) preserved at FP16/Q80 precision to eliminate token routing entropy drift. 3. High-Throughput Edge MoE: Activates only a fraction of parameters per token, delivering 7B-class…
A 12,537,920-parameter Japanese decoder-only language model for local contextual candidate reranking and short within-sentence continuation. This release contains the selected V2.1 extend5 best / step40000 FP32 inference bundle and its INT8 block32, FP32-compute Core ML variant. V2 experiments completed on 2026-10-08; the current quantization loss and observed input latency are accepted for this deployment. This is a custom PyTorch model, not a Transformers AutoModel integration or an instruction/chat model. SentencePiece, kana retrieval, candidate scoring and search run outside the Core ML graph. The original architecture, loader and scoring code are included under source/; no…
VinaySLM is a small language model (SLM) built from scratch in PyTorch to learn how decoder-only Transformers work end to end -- tokenization, embeddings, causal self-attention, training, and generation -- without importing a pretrained architecture. This is not a production-quality language model. - A tiny, hand-written, decoder-only (GPT-style) Transformer. - 466,176 parameters. - Trained on a ~600-token toy corpus of AI/programming sentences. - Packaged here as a custom Hugging Face model (trustremotecode=True) so it can be loaded and inspected through the transformers ecosystem, while the actual Transformer implementation (modelingvinayslm.py) is a thin wrapper around this project's…
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.