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 2026-10-23 · How the library is built
12,579 models, sorted by most downloaded.
NAJY 11단계 ACT 체크포인트 (1호기 DGX1 학습). 분석·채점용 업로드 — 실기 배포 후보 확정과는 별개. 받은 뒤: sha256sum model.safetensors 가 위와 같은지. 채점 도구는 같은 폴더의 multimanifest.json 에서 플래그를 읽는다 (--checkpoint; pretrainedmodel 구조 그대로이므로 stageconddiag.py 등에 바로 쓴다). 배경: trossen-ai-simulation docs/mobilebaseinvestigation.md §94.
NAJY 11단계 ACT 체크포인트 (1호기 DGX1 학습). 분석·채점용 업로드 — 실기 배포 후보 확정과는 별개. 받은 뒤: sha256sum model.safetensors 가 위와 같은지. 채점 도구는 같은 폴더의 multimanifest.json 에서 플래그를 읽는다 (--checkpoint; pretrainedmodel 구조 그대로이므로 stageconddiag.py 등에 바로 쓴다). 배경: trossen-ai-simulation docs/mobilebaseinvestigation.md §94.
NAJY 11단계 ACT 체크포인트 (1호기 DGX1 학습). 분석·채점용 업로드 — 실기 배포 후보 확정과는 별개. 받은 뒤: sha256sum model.safetensors 가 위와 같은지. 채점 도구는 같은 폴더의 multimanifest.json 에서 플래그를 읽는다 (--checkpoint; pretrainedmodel 구조 그대로이므로 stageconddiag.py 등에 바로 쓴다). 배경: trossen-ai-simulation docs/mobilebaseinvestigation.md §94.
NAJY 11단계 ACT 체크포인트 (1호기 DGX1 학습). 분석·채점용 업로드 — 실기 배포 후보 확정과는 별개. 받은 뒤: sha256sum model.safetensors 가 위와 같은지. 채점 도구는 같은 폴더의 multimanifest.json 에서 플래그를 읽는다 (--checkpoint; pretrainedmodel 구조 그대로이므로 stageconddiag.py 등에 바로 쓴다). 배경: trossen-ai-simulation docs/mobilebaseinvestigation.md §94.
NEO = Neural Expert Orchestrator. Release: v1 (September 29, 2026). Model card only; this repository does not yet contain model weights. NEO v1 is the HYPERneocloud serving name for DeepSeek-R1-Distill-Qwen-14B. The currently verified serving configuration uses upstream revision 1df8507178afcc1bef68cd8c393f61a886323761. It is not presented as newly trained or independently owned base weights. No NEO fine-tuned adapter is verified as loaded in this endpoint. - https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-14B - Exact revision was verified using the model-hub revision API. - Upstream card identifies the license as MIT. Preserve the original license and attribution in any…
游戏 UGC 评论「语义单元抽取」模型。基于 Qwen/Qwen3-4B，LoRA（r16 / alpha32 / dropout0.05）训练后合并进底座，fp16 权重，可直接用 transformers 加载，无 peft 依赖。 - 任务：从一条评论文本中抽取语义单元（mention / span / occurrence），只做抽取，不做分类 - 硬件：实测峰值显存 8.82 GB，T4/T10（16GB，Turing）可直接部署，attention 走 sdpa，无需 FlashAttention2 生成参数已固化在 generationconfig.json：贪心解码（dosample=False）+ maxnewtokens=2048，请勿改动，否则与评测结果不一致。 ok=False 表示输出 JSON 解析失败或存在非法 span（mention 不在正文 / occurrence 越界），此时 semanticunits 为已通过校验的子集；线上建议对 ok=False 记录走兜底（重试或降级）。 - JSON 成功率 98.99%，非法 span 0 - SDK 与评测输出平价验证：20/20 DEV 样本单元集合完全一致 1. dtype 必须 fp16：Turing 架构无 bf16 支持；Ampere+（A10/A100/4090）可改用 bf16 3. 时延：A100 mean 4.8s / p50 3.8s 每条；Turing 预计 8~15s，随单元数线性增长 4. 输入 text 建议 ≤ 1024 token（训练 maxseq…
A local research instrument for connecting separate, provenance-preserving neural graphs through an artificial trainable routing layer. The working package includes typed sparse graphs, CSV and human-matrix importers, recurrent graph encoders, sparse synthetic grafts, a delayed-memory benchmark with matched controls, validated inference checkpoints, a persistent live 3D navigation and continuous-learning workspace (python -m neurograft live), and a differentiable interface to frozen Qwen backbones. The pretrained model is an optional dependency and is not downloaded automatically. Current evidence: the circuit implementation runs and trains locally. The OpenWorm source has been acquired and…
NIDAR adds synthesized intensity to an existing point cloud using RGB appearance, a pseudo-NIR prior, intrinsic decomposition, and calibration. It requires RGB images + XYZ geometry + camera/LiDAR calibration. Existing nuScenes research example from the project page. The first portable demo targets Waymo-style exports; this figure is not a new release-validation run. Runtime destinations are relative to the chosen weights directory; files on the Hub are at the repository root. ARTIFACTS.json and SHA256SUMS provide sizes and checksums. Follow the source demo guide with source revision 8803e880e08662b07409f4bc9a9f6fb487101c99 or a compatible later revision. Python 3.10 and PyTorch 2.6 or…
NSFW Wan 1.3b T2V is a powerful, 1.3 billion parameter text-to-video generation model, specifically fine-tuned for generating Not Safe For Work (NSFW) content. The model has undergone multiple training methodologies to create a model that has a solid understanding across the entire NSFW spectrum and can generate videos with coherent motion natively. The primary goal of this model is to provide a research and creative tool capable of generating thematically relevant short video clips based on text prompts within the adult content domain. It aims to understand and render a wide array of NSFW scenarios, aesthetics, and actions described in natural language, now with improved temporal…
Language Intelligence layer of the NULLXES Intelligence Stack. SHINRA Our llm. Release line 1. NULLXES SHINRA-4B-BASE — pretrain 2. NULLXES SHINRA-4B-INSTRUCT — instruction tuning ← this model 3. NULLXES SHINRA-4B-INSTRUCT (aligned) — DPO / preference optimization Parameter budget Architecture source: configs/shinra4b.yaml at 903a639e03bc. The total also matches the published safetensors metadata. Custom SentencePiece Unigram, trained in-house on a web + wiki + code mix. Special tokens trustremotecode=True is required: SHINRA ships a custom ShinraConfig and modeling code, not a reused architecture class. The HF inference widget cannot load custom code, hence inference: false. Data pipeline…
NULLXES SHINRA-4B-INSTRUCT is the Language Intelligence Layer of the NULLXES system. SHINRA is responsible for multilingual understanding, coding intelligence, instruction following, structured outputs, and agent preparation. This checkpoint is the instruction-tuned (and optionally DPO-aligned) 4B-class dense decoder. Proprietary ShinraForCausalLM (not a Llama / Mistral / Qwen / GPT-NeoX wrapper). RMSNorm → GQA+RoPE → residual → RMSNorm → SwiGLU → residual then final RMSNorm and tied LM head. Special tokens:. Generation stop is. Document stop is. Three stages. Pretrain → NULLXES SHINRA-4B-BASE SHINRAPRETRAINV1: 40% FineWeb-Edu, 20% code (python-edu + licensed Stack), 15% math/science…
The code for using the ReaSyn model checkpoint is available in the official Github repository. ReaSyn is a model for predicting the synthesis pathway, reaction steps from reactants to final product(s), for a target product molecule. When the target molecule cannot be synthesized directly using known reaction steps, ReaSyn will generate the pathways for the most structurally similar, synthesizable analog of the target molecule.The model uses an encoder-decoder Transformer architecture, where a full synthetic pathway is represented as a text sequence. ReaSyn v2 improves the reconstruction and projection capabilities of ReaSyn v1 using a more advanced search (by combining top-down and…
Native MLX port of NVIDIA NV-Reason-CT for Apple Silicon. Run 3D CT reasoning and report generation locally on Mac. - Local NIfTI input with Apple Silicon GPU acceleration - FP32 inference using MLX; no PyTorch or CUDA required MLX port by Joseph Sandoval. Requires the companion nv-reason-ct-mlx runtime. Generic mlx-vlm and mlx-lm examples do not support this 3D CT model. Apple Silicon Mac, macOS, Python 3.12 and uv. Weights occupy 17.4 GB. Measured peak MLX memory use is about 22.4 GB; allow additional unified memory for preprocessing, macOS and other apps. To ask a question about the CT, set --prompt to your question. Input is one 3D CT volume in NIfTI format (.nii or.nii.gz) with…
Native compressed-tensors W6A8 checkpoint published by Xananthium. Credit and upstream terms: trohrbaugh/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16-heretic. The source model card is preserved in READMEORIGINAL.md. Quantization provenance and retained high-precision components are recorded in conversion-receipt.json. I wanted W6A8, so that is what we built: six-bit stored weights, dynamic eight-bit activations, and vLLM doing the serving. Humming packs the INT6 values across INT32 word boundaries and unpacks them for the 3090's supported INT8 matrix math. The cards do not need a native six-bit Tensor Core instruction. The extra unpacking and scale work is part of the cost, and the speed tests…
GGUF quantized version of nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16. 1. Upstream llama.cpp: Model probably works. 2. Forked version of llama.cpp: RPC with split mode tensor, MTP, and model works - Generate Korean-focused imatrix calibration dataset (Total 512 samples, Chat template applied for messages) - Create imatrix using llama-imatrix with --parse-special flag Unfortunately, I don't have PPL/KLD for quant, due to out of compute resource and target budget. 2x DGX Sparks, RPC, -sm tensor with MTP enabled (-ub 2048 -b 2048 -c 262144)
An uncensored edit of NVIDIA Nemotron-3.5-Bolt-74B-A9B, a hybrid Mamba/MoE model, produced by editing NVIDIA's shipped 24GB 2-bit GGUF (nvidia/NVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB-GGUF) directly — no BF16 requantize at any point. That matters for this model specifically. Bolt is a LatentMoE: every routed expert sums through one dense ffnlatentup per layer, and the experts themselves are ~17GB of IQ2XXS weights produced by NVIDIA's quantization-aware distillation. A conventional abliterate-then-requantize pipeline would throw that QAD work away. Instead the projection is applied in place and codes are re-picked under the original block scales, so at λ=0 the output file is byte-identical to…
Model · Image and text to text
NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-heretic-ara-i1-GGUF
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.