trained option-scoring head. It answers choice / noul / score questions about a document and up to four images in one forward pass, and speaks the Images need the vision extra (Pillow, torchvision): pip install "rsi-jev[vision] @ git+https://github.com/Shanghua-Gao/RSI-Jev". The vision encoder and image processor are read from the base model. This checkpoint (seed 17, one seed): Re-computed from these exact files after upload-time reload. Kernel stack fla-0.5.2/torch-2.7.1+cu128; full recipe in meta.json. loadrelease applies calibration.safetensors when the checkpoint carries one, so the probabilities are the calibrated ones this card reports. A loader that skips those files returns the raw…
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.
11,122Models
3,164Datasets
285Papers
6,792Publishers
11,852Sourced relationships
Updated 2026-10-18 · How the library is built
11,122 models, sorted by most downloaded.
Checkpoints from the real-world arm of the RSM-as-value-function study (rsmrise), trained on the rsmcookingv5 FR3 / Agilex tabletop cooking dataset: 248 teleop episodes, 181530 motion-filtered frames, 2 cameras (global, wrist), success rate 0.202 under the operator's arrival rule ("next.reward > 0 somewhere, i.e. an operator-marked arrival at a reward state [4] (state 4 = pan on grate, tomato in bowl)"). Shared for the robotics collaboration (@jmarangola). This repo is public: no token and no access request are needed to download anything here. It was created private and switched to public on 2026-09-22 so the binaries could be uploaded at all (the account had no private storage headroom…
All 445 policy adapters from the paper From Constitutions to Control: Interpretable Rewards for Aligning Language Models, one folder per fine-tuning arm, plus the harm-softened response generator for GPT-OSS. Layout: Every policy adapter is a standard PEFT LoRA directory (peft 0.19.1; r = 16, α = 32, dropout 0.05 on q/k/v/o), trained with the recipe in the paper's appendix; load with PeftModel.frompretrained(base, "iclr1367/rubric-rewards-adapters", subfolder="gptoss/free/help1lean+3"). Training sets, generations, judgments and every number behind the figures are in the companion dataset iclr1367/rubric-rewards; the code is at github.com/iclr1367-commits/rubric-rewards. Adapters inherit the…
static quants of https://huggingface.co/ritiksuman/rudra For a convenient overview and download list, visit our model page for this model. weighted/imatrix quants are available at https://huggingface.co/mradermacher/rudra-i1-GGUF 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: https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9 See…
weighted/imatrix quants of https://huggingface.co/ritiksuman/rudra For a convenient overview and download list, visit our model page for this model. static quants are available at https://huggingface.co/mradermacher/rudra-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 · Text generation
rus-100mb-after-wc-zipf-newlex-before-ckpt500-packed-bfdiso_seed10
This model is a fine-tuned version of francesca9805/ppt-wc-zipf-newlex-rus-before-100mb-packed-bfdisoseed10. It has been trained using TRL. This model was trained with SFT.
Model · Text generation
rus-100mb-after-wc-zipf-newlex-before-ckpt500-packed-bfdiso_seed455
This model is a fine-tuned version of francesca9805/ppt-wc-zipf-newlex-rus-before-100mb-packed-bfdisoseed455. It has been trained using TRL. This model was trained with SFT.
Model · Text generation
rus-cyrl-100mb-after-ppt-Dp-10mb-packed-bfdiso-ckpt500_seed3407
This model is a fine-tuned version of francesca9805/rus-cyrl-100mb-ppt-Dp-10mb-packed-bfdisoseed3407. It has been trained using TRL. This model was trained with SFT.
Model · Text generation
rus-cyrl-10mb-after-ppt-Dp-100mb-packed-bfdiso-ckpt500_seed10
This model is a fine-tuned version of francesca9805/rus-cyrl-10mb-ppt-Dp-100mb-packed-bfdisoseed10. It has been trained using TRL. This model was trained with SFT.
ONNX exports of the shared RVC feature/pitch networks: - hubertv2.onnx — ContentVec HuBERT feature extractor (layer 12, 768-D), exported from hubertbase.pt - rmvpemodel.onnx — RMVPE salience network, exported from rmvpe.pt - hubertbase.pt / rmvpe.pt — original checkpoints (half-precision copies) from via Jmica/audiobookmodels.
We introduce DeepSeek-V4.1-Flash, a multimodal Mixture-of-Experts (MoE) model with 552B backbone parameters and support for contexts of up to one million tokens. The model natively processes images and text, and generates text autoregressively. Architecture. DeepSeek-V4.1-Flash adopts a Causal Encoder-Decoder (CED) architecture: a 40-layer Transformer organized as a 20-layer causal encoder followed by a 20-layer decoder. With CED, the decoder's global KV cache is projected from the final encoder hidden states rather than derived from each decoder layer's own hidden states. This allows the model to activate only 8B parameters per token during prefill and 16B during decode, substantially…
Inference checkpoint from the S015 CTR50 without IdleMask arm. It contains the OpenPI parameter tree and normalization assets. Optimizer and training loader state are excluded. - params/: OpenPI model parameters - assets/: normalization assets - resolvedconfig.json: resolved inference configuration with machine-local paths removed - training-provenance.json: pinned training, dataset, normalization, and runtime identities
Inference checkpoint from the S015 CTR50 with IdleMask ablation. It contains the OpenPI parameter tree and normalization assets. Optimizer and training loader state are excluded. - params/: OpenPI model parameters - assets/: normalization assets - resolvedconfig.json: resolved inference configuration with machine-local paths removed - training-provenance.json: pinned training, dataset, normalization, and runtime identities
S017 Water Delivery / mixed / PI0.5 formal training Archived under user instruction on 2026-10-04. This preserves the actual checkpoint and normalization. It does not establish paper-result identity or audit approval. Source Run and immutable dataset/runtime/source identities are recorded in archive-provenance.json. Historical defects and experiment scope restrictions remain in force. Inference parameters and actual normalization/processor state only. Optimizer and RNG are not included.
s1 is an open, calibrated System One decision model. Given a state (text or JSON) and a typed question, it returns a probability for every option in one forward pass, with no text generated: It is google/gemma-4-26B-A4B-it (mixture of experts, 26B total and about 4B active parameters) fine-tuned with LoRA on about 100,000 typed decision rows, and merged. Code, data and Held-out test sets. The released snapshot was chosen on separate dev splits by a rule fixed before training. - Mean accuracy over JevBench's 11 accuracy slices; ECE with 10 bins. Raw ECE is 0.083 (Jev: 0.081). - Paired against the untuned model on the same JevBench cases: 103 answers fixed, 58 broken (McNemar p = 0.0005).…
A local, open, Jev-compatible "System One" decision model. s1-jev-cvltist is the name of this release model (project codename "system one", s1; the internal checkpoint id is s1-v6). Base: Qwen3.5-4B (Apache-2.0) · Adapter: LoRA r=16, α=32 · Context: 4096 Interface: the TypeSafe Jev wire spec (choice / score / noul) · Size: 124 MB adapter system one (s1) is a small decision model trained to answer Jev-style questions (choice, score, noul) over a free-form "state", returning calibrated probabilities rather than a bare answer. It is deliberately built on a general instruction base (Qwen3.5-4B) with a thin LoRA adapter, so it carries knowledge a purpose-built encoder can't, while the readout…
A local, open, Jev-compatible "System One" decision model. s1-jev-cvltist is the name of this release model (project codename "system one", s1; the internal checkpoint id is s1-v6). Base: Qwen3.5-4B (Apache-2.0) · Adapter: LoRA r=16, α=32 · Context: 4096 Interface: the TypeSafe Jev wire spec (choice / score / noul) · Size: 124 MB adapter system one (s1) is a small decision model trained to answer Jev-style questions (choice, score, noul) over a free-form "state", returning calibrated probabilities rather than a bare answer. It is deliberately built on a general instruction base (Qwen3.5-4B) with a thin LoRA adapter, so it carries knowledge a purpose-built encoder can't, while the readout…
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.