# Open-Weight Models Under Apache License 2.0, Page 52 Source: https://savrn.com/models/licenses/apache-2-0/page/52 Markdown alternate of the page above; the site index is https://savrn.com/llms.txt --- SAVRN Model Hub · Models by License # Open-Weight Models Under Apache License 2.0 3,285 open-weight models released under Apache License 2.0 in the SAVRN Model Hub, with Team Mradermacher, Qwen and Google publishing the most. 3,285Models 1587Publishers 12K to 952.4BParameter range 1Licenses YesCommercial use ## What Apache License 2.0 Allows The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors. [Read the license text](https://www.apache.org/licenses/LICENSE-2.0). ## SAVRN's Take Apache 2.0 asks little of the operator. Commercial use, modification and redistribution are all permitted. What it requires is housekeeping: keep the license and copyright notices, carry any NOTICE file along with the weights, and state the significant changes if you pass a modified version on. It also includes an express patent grant from contributors, the clause a procurement reviewer looks for before a model serves paying customers. So an organization can take the files from the publisher, fine-tune on its own data, run the result in its own facility and bill for the output. On our hub 1,018 models carry this license. Qwen leads with 81, Google 50, Convergent Intelligence 39, PyTorch Image Models 32, OpenMed 31, Helsinki-NLP Research Group 27, and Unsloth AI and PaddlePaddle 22 each. Text generation accounts for 164 of the models, image and text to text for 89, and robotics for 43, so it covers more than chat. Downloads show where it runs. all-MiniLM-L6-v2 from Sentence Transformers, a 23M parameter sentence similarity model, pulls 255,050,544 downloads a month and needs 0.1 GB at 16-bit, small enough to ride on a GPU already serving. ms-marco-MiniLM-L6-v2 follows at 88,642,387 for text ranking. The one text generation model in the top eight is Qwen3-0.6B at 22,498,727 downloads a month: 752M parameters, a 40,960 token context, 1.8 GB at 16-bit, and the cheapest host on the Index for it is one MI300X at $1.85 an hour. SAVRN Research, 2026-09-18 ## Most Downloaded | Model | Publisher | Parameters | License | Monthly downloads | Cheapest GPUs at 16-bit | | --- | --- | --- | --- | --- | --- | | [all-MiniLM-L6-v2](https://savrn.com/models/all-minilm-l6-v2) | Sentence Transformers | 23M | apache-2.0 | 203.2M | 1x MI300X, $1.85/hr | | [ms-marco-MiniLM-L6-v2](https://savrn.com/models/ms-marco-minilm-l6-v2) | Sentence Transformers - Cross-Encoders | 23M | apache-2.0 | 74.7M | 1x MI300X, $1.85/hr | | [paraphrase-multilingual-MiniLM-L12-v2](https://savrn.com/models/paraphrase-multilingual-minilm-l12-v2) | Sentence Transformers | 118M | apache-2.0 | 43M | 1x MI300X, $1.85/hr | | [electra-base-discriminator](https://savrn.com/models/electra-base-discriminator) | Google | — | apache-2.0 | 41M | — | | [bert-base-uncased](https://savrn.com/models/bert-base-uncased) | BERT community | 110M | apache-2.0 | 29.7M | 1x MI300X, $1.85/hr | | [Qwen3-0.6B](https://savrn.com/models/qwen3-0-6b) | Qwen | 752M | apache-2.0 | 28.2M | 1x MI300X, $1.85/hr | | [t5-small](https://savrn.com/models/t5-small) | T5 community | 61M | apache-2.0 | 24.5M | 1x MI300X, $1.85/hr | | [mobilenetv3_small_100.lamb_in1k](https://savrn.com/models/mobilenetv3-small-100-lamb-in1k) | PyTorch Image Models | 3M | apache-2.0 | 18.5M | 1x MI300X, $1.85/hr | | [all-mpnet-base-v2](https://savrn.com/models/all-mpnet-base-v2) | Sentence Transformers | 109M | apache-2.0 | 18.3M | 1x MI300X, $1.85/hr | | [wav2vec2-large-xlsr-53-japanese](https://savrn.com/models/wav2vec2-large-xlsr-53-japanese) | Jonatas Grosman | — | apache-2.0 | 16.2M | — | ## Who Publishes Them | Publisher | Models | | --- | --- | | [Team Mradermacher](https://savrn.com/model-publishers/mradermacher) | 162 | | [Qwen](https://savrn.com/model-publishers/qwen) | 81 | | [Google](https://savrn.com/model-publishers/google) | 50 | | [Quan](https://savrn.com/model-publishers/wq2012) | 44 | | [RKNNAI](https://savrn.com/model-publishers/rknnai) | 43 | | [Convergent Intelligence](https://savrn.com/model-publishers/reaperdoesntknow) | 39 | ## All 3,285 Models, Page 52 of 55 Model · Image classification ### [species-classifier-arachnids](https://savrn.com/models/species-classifier-arachnids) [Michael G](https://savrn.com/model-publishers/mlgraham) A 153-species arachnid classifier (spiders, scorpions, ticks, harvestmen) small enough for microcontrollers and the Coral Edge TPU. MobileNetV2 at full width and 160 px, full-integer int8, trained on the iNaturalist 2021 competition data. It fills the gap people hit first with Google's AIY insect model, which has no spiders. Part of species-classifiers. Measured on the int8 file over the 1,530 iNaturalist 2021 validation images for these species (10 per species), with an 87.5% centre crop: The PyTorch checkpoint scores 64.4% top-1. For comparison, Google's stock AIY insect model scores 59.1% top-1 on the iNat2021 species it shares, choosing among 1,022 classes. - Input [1, 160, 160, 3]… Open weights apache-2.0 tflite [View model](https://savrn.com/models/species-classifier-arachnids) Model · Image classification ### [species-classifier-mammals](https://savrn.com/models/species-classifier-mammals) [Michael G](https://savrn.com/model-publishers/mlgraham) A 246-species mammal classifier small enough for microcontrollers and the Coral Edge TPU. MobileNetV2 at full width and 160 px, full-integer int8, trained on the iNaturalist 2021 competition data. Part of species-classifiers, which also rescues Google's AIY Vision Kit insect, plant and bird models into the same shape. Measured on the int8 file over the 2,460 iNaturalist 2021 validation images for these species (10 per species), with an 87.5% centre crop: The PyTorch checkpoint scores 56.1% top-1; int8 costs about a point. Grayscale is reported because the model was trained with 20% random grayscale so it holds up on monochrome cameras such as the Coral Dev Board Micro's. For comparison… Open weights apache-2.0 tflite [View model](https://savrn.com/models/species-classifier-mammals) Model · Robotics ### [spiral-so101-rlt12-stage1-beta-teleop57-dagger44-rm1fix12-fresh-1500](https://savrn.com/models/spiral-so101-rlt12-stage1-beta-teleop57-dagger44-rm1fix12-fresh-1500) [Nadeem Shoukath](https://savrn.com/model-publishers/nadeem-shoukath115113) SPIRAL round 1 as Algorithm 1 stage 2b defines it (π₁'s own fine-tuning data, labelled by RM1), repeated at four values of betabc, one run per beta, everything else fixed. A residual actor and a 5-critic ensemble on top of a FROZEN π₀.₅ + RLT policy; actor and critics both start fresh. Not robot-evaluated; the offline critic probes have not been run. All four are identical except betabc (checked in each rlconfig.json). Caveat for beta 70 and 100: with a fresh actor the residual starts large (random init) and has to shrink to the value set by the critic. At beta 5 and 15 it settles (residual × beta ≈ 0.035, the critic's slope on this data), but at beta 70 and 100 it was still falling at step… Open weights apache-2.0 [View model](https://savrn.com/models/spiral-so101-rlt12-stage1-beta-teleop57-dagger44-rm1fix12-fresh-1500) Model · Text classification ### [sporadic-cystitis-two-stage](https://savrn.com/models/sporadic-cystitis-two-stage) [David Martinez](https://savrn.com/model-publishers/jibmaird) Open weights apache-2.0 pytorch [View model](https://savrn.com/models/sporadic-cystitis-two-stage) Model ### [spr-qwen2.5-7b-trained](https://savrn.com/models/spr-qwen2-5-7b-trained) [Mahakisore](https://savrn.com/model-publishers/mahakisore77) This qwen2 model was trained 2x faster with Unsloth Open weights apache-2.0 transformers [View model](https://savrn.com/models/spr-qwen2-5-7b-trained) Model · Text generation ### [SRA-RiskGate-4B-GGUF](https://savrn.com/models/sra-riskgate-4b-gguf-2) [Team Mradermacher](https://savrn.com/model-publishers/mradermacher) Open weights apache-2.0 transformers [View model](https://savrn.com/models/sra-riskgate-4b-gguf-2) Model · Image and text to text ### [sriq-heretic-spatial-9b-MIXED-10020-preview](https://savrn.com/models/sriq-heretic-spatial-9b-mixed-10020-preview) [Sparse Reasoning Intelligence & Quality](https://savrn.com/model-publishers/sriq-ai) Merged bf16 weights. Reasons in compressed Simplified Chinese and answers in the language it was asked in. No benchmarks were run for this release. Nothing here reports rows that reason in Chinese, verification rate, or looping against the base model. Evaluate it on your own workload before relying on it. - Training saw 2,940 of the dataset's 2,940 rows at 700 steps on an effective batch of 8, about 1.90 of one epoch. - Inherits the base model's limitations. Do not pass enablethinking=False: the template prefills an empty and the trace disappears. Wall clock and cost are measured by the agent from its own start to publish, so they cover the base-model download and the merge upload as well… Open weights apache-2.0 9.7B parameters 262,144 tokens transformers [View model](https://savrn.com/models/sriq-heretic-spatial-9b-mixed-10020-preview) Model ### [Starkid](https://savrn.com/models/starkid) [TRƯỜNG MẦM NON NGÔI SAO NHỎ](https://savrn.com/model-publishers/ngoisao) Open weights apache-2.0 [View model](https://savrn.com/models/starkid) Model · Tabular regression ### [stocka3](https://savrn.com/models/stocka3) [BAAR](https://savrn.com/model-publishers/aixk) ~425K 파라미터 4-Layer Dynamic GAU 기반 실전 주식 트레이딩 파운데이션 모델입니다. ATR 연동 동적 배리어 라벨링과 고확신 타점 필터링으로 손익비 1.5:1 이상을 타겟팅합니다. Open weights apache-2.0 490,602 parameters safetensors [View model](https://savrn.com/models/stocka3) Model · Text generation ### [stormdesk-triage](https://savrn.com/models/stormdesk-triage) [Nathan DeMoss](https://savrn.com/model-publishers/ndemoss28) This qwen3 model was trained 2x faster with Unsloth and Huggingface's TRL library. Open weights apache-2.0 1.7B parameters 40,960 tokens transformers [View model](https://savrn.com/models/stormdesk-triage) Model · Text generation ### [stride-qwen3-1.7b-allref-correctonly-kl0-local-positive-20260925-resume](https://savrn.com/models/stride-qwen3-1-7b-allref-correctonly-kl0-local-positive-20260925-resume) [Kumar](https://savrn.com/model-publishers/sandeep123) This repository contains the final epoch-4 LoRA adapter from the 1.7B all-reference STRIDE experiment. The base model is Qwen/Qwen3-1.7B, used with thinking disabled during training and evaluation. Each prompt produced eight rollouts. Novelty was computed against eligible steps from all eight rollouts, including correct and incorrect trajectories. Only correct trajectories received the auxiliary STRIDE diversity bonus; incorrect trajectories retained their ordinary outcome/GRPO signal. This adapter uses local-positive step allocation. Use the Qwen chat template with enablethinking=False to reproduce the non-thinking evaluation setting. This is a research adapter for mathematical reasoning.… Open weights apache-2.0 peft [View model](https://savrn.com/models/stride-qwen3-1-7b-allref-correctonly-kl0-local-positive-20260925-resume) Model ### [Structure-Over-Scale](https://savrn.com/models/structure-over-scale) [Convergent Intelligence](https://savrn.com/model-publishers/reaperdoesntknow) Roy C March 2026 We present a methodology for training small language models on CPU at FP32 precision that achieves capability-per-dollar efficiency orders of magnitude beyond GPU-based training. Across 15 models spanning four novel architecture families — Mixture of Attentions (MoA), cross-architecture fusion (Qemma), swarm intelligence (SAGI), and metric-space causal language models (DiscoverLM) — total compute cost was $24 on a single AMD EPYC 9454P processor. We introduce seven methodological pillars: (1) FP32 precision preservation, with experiments demonstrating 5,810x single-operation error and 23,225x compounding error ratio for FP16 at network depth; (2) sparse cognitive… Open weights apache-2.0 [View model](https://savrn.com/models/structure-over-scale) Model ### [study-classification](https://savrn.com/models/study-classification) [Tony Y.](https://savrn.com/model-publishers/tonygyoshida) A small Mocov3 implementation for Classification, packaged with an explicit configuration and an initialization checkpoint. The small variant is a reproducible starting point, not a trained model release. - The Python file contains the model and runnable example or training entry point. - config.json records the generated architecture settings. - trainingargs.json records the default experiment recipe. - model.safetensors is a valid initialization checkpoint for smoke tests; it is not presented as a trained benchmark checkpoint. - No benchmark score is claimed in this repository. The included configuration uses adam with a constant warmup schedule. These are starting values in the script… Open weights apache-2.0 33,088 parameters 128 tokens [View model](https://savrn.com/models/study-classification) Model · Text generation ### [SUPERBRAIN-1](https://savrn.com/models/superbrain-1) [Ahmad Odeh](https://savrn.com/model-publishers/abuz8studios) SUPERBRAIN-1 is a 9B Arabic model by ABUZ8, fine-tuned from QCRI/Fanar-1-9B to do Arabic root, conjugate, derive masdars and participles, and name a verb's form. It is a completion model, not a chat model: give it one of the prompt formats below and read the first line of the output. Packet B — held-out test. 1,343 items over 258 roots that were Quranic Arabic Corpus) and modern words from the Arramooz lexicon. Root items are asked both with a few-shot prompt and as a bare question. Root accuracy by root class (bare): sound 79%, hollow 60%, doubled 56%, hamzated 51%, defective 45%, assimilated 44%. 85-item gate (v3) — 50 root words, 20 arithmetic/logic questions, 15 Generalization probe.… Open weights apache-2.0 gguf [View model](https://savrn.com/models/superbrain-1) Model · Text classification ### [superfast-tiny-home-robotic](https://savrn.com/models/superfast-tiny-home-robotic) [Sr Aivante](https://savrn.com/model-publishers/sraivante) An offline English and Hinglish command interpreter for home devices and registered industrial/robotics devices. It converts one instruction into a catalog-validated JSON intent. The package includes a trained 87,181-byte action model, a 1,002,414-device SQLite catalog, a Python API, an HTTP service, a browser playground, Docker files, and reproducible training and evaluation data. The runtime was previously packaged locally as edge-command-docker-v3-ui. The system resolves device names through explicit identifiers, aliases and SQLite FTS5 search. It prioritizes action patterns and capability checks, with a stored count-based unigram/bigram classifier for the fallback scoring path. The… Open weights apache-2.0 [View model](https://savrn.com/models/superfast-tiny-home-robotic) Model ### [Supernova-NepaliFast-V5](https://savrn.com/models/supernova-nepalifast-v5) [Supernova](https://savrn.com/model-publishers/supernova11c) next Supernova NepaliFast V5 is a custom tokenizer architecture designed primarily for Nepali, English, Devanagari, and multilingual Unicode text. V5 is not simply a new vocabulary placed on top of an existing BPE, Unigram, or WordPiece engine. The core V5 system includes a custom main tokenizer, custom trie-based matching, a dedicated Unicode fallback expert, grapheme-aware routing, exact reconstruction validation, a global token-ID namespace, and deterministic decoding. Supernova V5 uses a tokenizer-level Mixture-of-Experts (MoE) architecture. This document provides an authoritative security audit and bytecode forensic verification for all compiled pickle (.pkl) serialized files in the… Open weights apache-2.0 tokenizers,supernova-v5 [View model](https://savrn.com/models/supernova-nepalifast-v5) Model ### [Supernova-NepaliFast-V6](https://savrn.com/models/supernova-nepalifast-v6) [Supernova](https://savrn.com/model-publishers/supernova11c) This directory contains the completed, optimized Rust Native Tokenizer Engine (supernovarust) for Supernova NepaliFast V5. It is built using PyO3 and Maturin, delivering 15M+ characters per second (running 3.86x faster than OpenAI's highly optimized tiktoken engine). OpenAI Tiktoken (cl100kbase): ~3.9M chars/sec (~3.96 ms average on sample document) Follow these steps to compile and install the high-performance native module directly from this source code. Run this inside the supernovarust directory containing Cargo.toml: After building, install the newly created.whl package using pip: Once compiled and installed, you can use the Rust Engine in your Python workspace as follows Open weights apache-2.0 [View model](https://savrn.com/models/supernova-nepalifast-v6) Model · Any to any ### [support](https://savrn.com/models/support) [David Belton](https://savrn.com/model-publishers/davidau) https://huggingface.co/collections/DavidAU/turbo-brilliance-ai-constructs-50x-reasoning-instruct To support our team, sponsor, "Buy Me A Coffee", commission project(s), access premium item(s)/model(s)/module(s), model modifications, GET HELP, see a sorted list of models (380+) and more please see below. We build precision models at scale for the community. We are user-funded, and use local hardware to create, tune and deploy models worldwide. We also create systems for fine tuning, and specialized model systems too as well as "augment" current model/AI technology. - Model fine tuning across a diverse arch type of models and sizes, including MOEs/sparse moes. (over 1000) - Model… Open weights apache-2.0 transformers [View model](https://savrn.com/models/support) Model ### [surfPRM-qwen3.5-4b](https://savrn.com/models/surfprm-qwen3-5-4b) [PoloWitty](https://savrn.com/model-publishers/polowitty) A process reward model (PRM) for web agents, built by LoRA SFT on Qwen3.5-4B and merged into full weights. It evaluates candidate next actions of web agents at intermediate steps, given the task intent, the current page's accessibility tree (AXTree), and the previous action trajectory. Trained on the surfPRM-data dataset: 9,821 pairwise step-preference samples for web agents. Each sample contains the task intent, the current page's accessibility tree (AXTree), the previous action trajectory, the start/current URLs, and two candidate next actions grounded on the current page; the model outputs structured / / / reasoning and selects the better action. Requires transformers >= 5.6. Load as a… Open weights apache-2.0 4.5B parameters 262,144 tokens [View model](https://savrn.com/models/surfprm-qwen3-5-4b) Model · Text to image ### [Surjo-Image-Preview-ONNX](https://savrn.com/models/surjo-image-preview-onnx) [Mrfakename](https://savrn.com/model-publishers/mrfakename) ONNX export of SurjoLabs/Surjo-Image-Preview for the browser (onnxruntime-web, WebGPU), together with its text encoder LiquidAI/LFM2.5-Encoder-230M and VAE decoder fp16w means weights are stored in fp16 and computed in fp32 (image PSNR about 60 dB against PyTorch fp32). An int4 text encoder was tried and dropped: it changed the pictures too much. Sampling: start from standard normal noise, t = i/steps, dt = 1/steps, 50 steps, cfg 3.0. onnx/parity.json has the image PSNR against the PyTorch fp32 pipeline. Open weights apache-2.0 onnxruntime [View model](https://savrn.com/models/surjo-image-preview-onnx) Model · Text generation ### [svd-safety-mis7_swift_remove30_swapgapiter_b010](https://savrn.com/models/svd-safety-mis7-swift-remove30-swapgapiter-b010) [Park](https://savrn.com/model-publishers/jeesup) A Mistral-7B-Instruct-v0.2 checkpoint compressed with SVD-LLM to 70.0% of dense parameters, then edited by 10 of 10 rounds of iterative parameter-neutral swap selected by the gapiter rule (up to 0.1% of dense parameters per round; the full run's budget is 1.0%). This is a research artifact from a study of how SVD compression damages safety behaviour and which component-selection rule best repairs it. It is one cell of a grid over selection rules and budgets; it is not a general-purpose chat model. This checkpoint exists to measure safety/utility trade-offs under compression. Several arms in the grid are deliberately safety-degraded relative to Mistral-7B-Instruct-v0.2: compression alone… Open weights apache-2.0 7.2B parameters 32,768 tokens transformers [View model](https://savrn.com/models/svd-safety-mis7-swift-remove30-swapgapiter-b010) Model · Text generation ### [swift-1.5-27b-coding-GGUF](https://savrn.com/models/swift-1-5-27b-coding-gguf) [Alejandro Melo](https://savrn.com/model-publishers/bjivanovich) Official GGUF quantized weights for swift-1.5-27b-coding, a specialized 27B parameter coding model fine-tuned using DoRA on bjivanovich/code-py-rust-cpp-50k (covering Python, Rust, C++, and multi-step reasoning). All quantizations were generated using an importance matrix (imatrix) calibrated directly on domain-specific programming samples. Open weights apache-2.0 [View model](https://savrn.com/models/swift-1-5-27b-coding-gguf) Model ### [swin-t-contrastive-medium](https://savrn.com/models/swin-t-contrastive-medium) [Nathan Reid](https://savrn.com/model-publishers/bmwo-zniak) This is an experimental Swin T codebase for Contrastive. It keeps the xlarge setup intentionally manageable so architecture changes can be inspected before a full training run. - The Python file contains the model and runnable example or training entry point. - config.json records the generated architecture settings. - trainingargs.json records the default experiment recipe. - model.safetensors is a valid initialization checkpoint for smoke tests; it is not presented as a trained benchmark checkpoint. - No benchmark score is claimed in this repository. The included configuration uses lamb with a linear warmup schedule. These are starting values in the script, not evidence of a completed… Open weights apache-2.0 33,088 parameters 128 tokens [View model](https://savrn.com/models/swin-t-contrastive-medium) Model ### [swin-t-experiment](https://savrn.com/models/swin-t-experiment) [Matthew Moore](https://savrn.com/model-publishers/nicholassmith1220) This is an experimental Swin T codebase for Matching. It keeps the nano setup intentionally manageable so architecture changes can be inspected before a full training run. - The Python file contains the model and runnable example or training entry point. - config.json records the generated architecture settings. - trainingargs.json records the default experiment recipe. - model.safetensors is a valid initialization checkpoint for smoke tests; it is not presented as a trained benchmark checkpoint. - No benchmark score is claimed in this repository. The included configuration uses adamw with a step schedule. These are starting values in the script, not evidence of a completed run. For a… Open weights apache-2.0 49,600 parameters 256 tokens [View model](https://savrn.com/models/swin-t-experiment) Model ### [swin-t-matching-light](https://savrn.com/models/swin-t-matching-light) [Brianna S. Turner](https://savrn.com/model-publishers/turnerbria) A small Swin T implementation for Matching, packaged with an explicit configuration and an initialization checkpoint. The giant variant is a reproducible starting point, not a trained model release. - The Python file contains the model and runnable example or training entry point. - config.json records the generated architecture settings. - trainingargs.json records the default experiment recipe. - model.safetensors is a valid initialization checkpoint for smoke tests; it is not presented as a trained benchmark checkpoint. - No benchmark score is claimed in this repository. The included configuration uses sgd with a step schedule. These are starting values in the script, not evidence of a… Open weights apache-2.0 24,832 parameters 512 tokens [View model](https://savrn.com/models/swin-t-matching-light) Model · Image classification ### [swin-tiny-patch4-window7-224-finetuned-eurosat](https://savrn.com/models/swin-tiny-patch4-window7-224-finetuned-eurosat) [User Name](https://savrn.com/model-publishers/username-6370) This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on an unknown dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 5e-05 - trainbatchsize: 32 - evalbatchsize: 32 - gradientaccumulationsteps: 4 - totaltrainbatchsize: 128 - lrschedulertype: linear - numepochs: 3 - Transformers 5.18.0 - Pytorch 2.11.0+cu130 - Datasets 4.8.5 - Tokenizers 0.23.2 Open weights apache-2.0 28M parameters transformers [View model](https://savrn.com/models/swin-tiny-patch4-window7-224-finetuned-eurosat) Model · Image classification ### [swin-tiny-patch4-window7-224-finetuned-eurosat](https://savrn.com/models/swin-tiny-patch4-window7-224-finetuned-eurosat-2) [Brysa Nicks](https://savrn.com/model-publishers/brysanicks) This model is a fine-tuned version of microsoft/swin-tiny-patch4-window7-224 on an unknown dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 5e-05 - trainbatchsize: 32 - evalbatchsize: 32 - lrschedulertype: linear - numepochs: 3 - Transformers 5.18.0 - Pytorch 2.11.0+cu130 - Datasets 4.8.5 - Tokenizers 0.23.2 Open weights apache-2.0 28M parameters transformers [View model](https://savrn.com/models/swin-tiny-patch4-window7-224-finetuned-eurosat-2) Model · Text generation ### [Swish-26B-A4B-GGUF](https://savrn.com/models/swish-26b-a4b-gguf) [Ready.Art](https://savrn.com/model-publishers/readyart) Open weights apache-2.0 [View model](https://savrn.com/models/swish-26b-a4b-gguf) Model ### [sy-test-adapter-0918](https://savrn.com/models/sy-test-adapter-0918) [Sheng Yao](https://savrn.com/model-publishers/sheng-yao) Open weights apache-2.0 [View model](https://savrn.com/models/sy-test-adapter-0918) Model ### [SyncReward](https://savrn.com/models/syncreward) [NaiveBayes](https://savrn.com/model-publishers/guan123) Open weights apache-2.0 [View model](https://savrn.com/models/syncreward) Model ### [szl-formulas](https://savrn.com/models/szl-formulas) [SZL Holdings](https://savrn.com/model-publishers/szlholdings) GitHub source: szl-holdings/szl-formulas @ 977f344ccf248b55299a623d91db162f266eda29. Artifacts: BENCH.laptop-blackwell.json, OPERATIONAL.json. Software kernel for SZL formula composition. Not a model. No weights. IS: a pure-Python, stdlib-only governance kernel — offline replay of the 21 canonical formulas plus the governed-loop composer. Public surface includes registrycount(), lambdaaggregate, LOCKEDPROVENFORMULAIDS, rungovernedloop, PROOFSTATUS (mirrored verbatim). IS NOT: trained weights, LoRA, GGUF,.pt, or a classifier. Hub model.joblib is QUARANTINED executable serialization (and is not on this tree) — do not joblib.load it. Not a CUDA/Triton speedup. Not lutar-lean and not the… Open weights apache-2.0 kernels [View model](https://savrn.com/models/szl-formulas) Model · Text classification ### [szl-triage-linear-development-v1](https://savrn.com/models/szl-triage-linear-development-v1) [SZL Holdings](https://savrn.com/model-publishers/szlholdings) NOTPROMOTABLE. No production admission or runtime integration. This is a newly trained, independently implemented six-class Python softmax classifier with 744 train-only features. It is a custom JSON weight format, not a Transformers checkpoint or a hosted inference endpoint. The complete study source, original weights, records, negative control and sealed thresholds are distributed from immutable Git blobs. SOURCEBINDING.json lists their hashes; unsigned hashes require trust in the declared canonical source. The fixed seed 11 study trained on 527 engine-labelled rows, excluding 101 held rows. On the previously exposed 42-row development challenge, joint label/state accuracy was 8/42… Open weights apache-2.0 [View model](https://savrn.com/models/szl-triage-linear-development-v1) Model ### [ta-certo-milistral-gguf](https://savrn.com/models/ta-certo-milistral-gguf) [Fernandes](https://savrn.com/model-publishers/dnfs26) Open weights apache-2.0 [View model](https://savrn.com/models/ta-certo-milistral-gguf) Model · Text to image ### [TAE-Krea-2](https://savrn.com/models/tae-krea-2) [J.M. Ruiz](https://savrn.com/model-publishers/academiasd) ComfyUI previews Krea 2 with Latent2RGB, a linear 16→3 color projection that looks blurry and blocky. This decoder is distilled from the Krea 2 VAE: it turns the sampler's latents into a real preview image in about 12 ms, with no need for the full VAE. 1. Download TAEKrea2AcademiaSD.safetensors to ComfyUI/models/vaeapprox/. 2. Install ComfyUI-KJNodes. 3. Add the Model Preview Override node between your Krea 2 model and the sampler. 4. In its tinyvae input, choose TAEKrea2AcademiaSD.safetensors. - Decoder only, there is no encoder. - Trained and tested with Krea 2 only. Tiny AutoEncoder (decoder tipo TAESD) para previsualizar Krea 2 en ComfyUI con nitidez y en tiempo real. Sustituye a… Open weights apache-2.0 comfyui [View model](https://savrn.com/models/tae-krea-2) Model ### [take2](https://savrn.com/models/take2) [W Joshua](https://savrn.com/model-publishers/checkmoteur) Open weights apache-2.0 [View model](https://savrn.com/models/take2) Model ### [tanpo-deals-LoRA](https://savrn.com/models/tanpo-deals-lora) [Ninja](https://savrn.com/model-publishers/d4rkninja) LoRA adapter for tanpo-deals (Deal Cracker) — B2B deal-closing and distributor/channel specialist. Open weights apache-2.0 peft [View model](https://savrn.com/models/tanpo-deals-lora) Model · Text generation ### [tanpo-retention-LoRA](https://savrn.com/models/tanpo-retention-lora) [Ninja](https://savrn.com/model-publishers/d4rkninja) PEFT LoRA adapter that specializes LiquidAI/LFM2.5-1.2B-Instruct for retention workflows (trained via Unsloth hub id unsloth/LFM2.5-1.2B-Instruct). Tanpo is a family of compact domain-specialized business models for local / edge / inexpensive deployment. Different specialists cover different workflows. Original upstream: LiquidAI/LFM2.5-1.2B-Instruct (~1.17B parameters, 32,768-token context, designed for edge/on-device deployment). Fine-tuning: Unsloth-compatible loading of that checkpoint via hub id unsloth/LFM2.5-1.2B-Instruct (LoRA / PEFT). Full eval, examples, and responsible-use notes: d4rkninja/tanpo-retention. - Onboarding, activation, churn-save, renewal, expansion, escalation, and… Open weights apache-2.0 peft [View model](https://savrn.com/models/tanpo-retention-lora) Model ### [TeamFolklore_Model](https://savrn.com/models/teamfolklore-model) [Sher](https://savrn.com/model-publishers/sherthecoder) Open weights apache-2.0 [View model](https://savrn.com/models/teamfolklore-model) Model · Speech recognition ### [techno-whisper-myanmar](https://savrn.com/models/techno-whisper-myanmar) [Myanmar Pyi Ai Techno](https://savrn.com/model-publishers/aitechno-burma) This model is a fine-tuned version of openai/whisper-medium on the chuuhtetnaing/myanmar-speech-dataset-openslr-80 dataset. It achieves the following results on the evaluation set: The following hyperparameters were used during training: - learningrate: 0.0003 - trainbatchsize: 40 - evalbatchsize: 40 - lrschedulertype: linear - lrschedulerwarmupsteps: 200 - numepochs: 50 - mixedprecisiontraining: Native AMP - Transformers 4.35.2 - Pytorch 2.1.1+cu121 - Datasets 2.14.5 - Tokenizers 0.15.1 Open weights apache-2.0 764M parameters transformers [View model](https://savrn.com/models/techno-whisper-myanmar) Model · Text generation ### [TeenyTinyStories-1M](https://savrn.com/models/teenytinystories-1m) [YF](https://savrn.com/model-publishers/ableian) GPT-Neo storyteller (nominal 1M; actual 3,745,984 parameters including the 50,257-token embedding) trained on TeenyTinyStories (TTS): 2,491,490 stories (498.5M tokens with EOS) generated by roneneldan/TinyStories-Instruct-33M. batch 1280 x 512, constant LR 5e-4, weight decay 0.1, betas (0.9, 0.95), clip 1.0. - TS-valid capped to stories of at most 400 tokens (20,711 stories): 2.369 / ppl 10.69 Research on small-scale synthetic pretraining data. Generates simple children's stories; sample at temperature 0.8-1.0. See the TeenyTinyStories paper for the corpus recipe and judge evaluations. Apache-2.0. Open weights apache-2.0 4M parameters 2,048 tokens transformers [View model](https://savrn.com/models/teenytinystories-1m) Model · Text generation ### [TeenyTinyStories-33M](https://savrn.com/models/teenytinystories-33m) [YF](https://savrn.com/model-publishers/ableian) GPT-Neo storyteller (nominal 33M; actual 68,514,048 parameters including the 50,257-token embedding) trained on TeenyTinyStories (ableian/TeenyTinyStories) batch 1280 x 512, constant LR 5e-4, weight decay 0.1, betas (0.9, 0.95), clip 1.0. - TS-valid capped to stories of at most 400 tokens (20,711 stories): 1.539 / ppl 4.66 Research on small-scale synthetic pretraining data. Generates simple children's stories; sample at temperature 0.8-1.0. See the TeenyTinyStories paper for the corpus recipe and judge evaluations. Apache-2.0. Open weights apache-2.0 69M parameters 2,048 tokens transformers [View model](https://savrn.com/models/teenytinystories-33m) Model ### [TerminalHorizon-122B-A10B-i1-GGUF](https://savrn.com/models/terminalhorizon-122b-a10b-i1-gguf) [Team Mradermacher](https://savrn.com/model-publishers/mradermacher) weighted/imatrix quants of https://huggingface.co/shunzou05/TerminalHorizon-122B-A10B For a convenient overview and download list, visit our model page for this model. static quants are available at https://huggingface.co/mradermacher/TerminalHorizon-122B-A10B-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… Open weights apache-2.0 transformers [View model](https://savrn.com/models/terminalhorizon-122b-a10b-i1-gguf) Model ### [tern-tc-100m](https://savrn.com/models/tern-tc-100m) [Dmitrii Vasilev](https://savrn.com/model-publishers/playra) A 100M-parameter ternary code model (BitNet b1.58 linears), trained on 10B tokens next to a full-precision twin, with the twin's scores beside it -- the ternary model is the weaker of the two. The ternary model is weaker than its full-precision twin. This is a research checkpoint, not a coding assistant. Every fact and every number about this model lives in modelcard.t27 in this repository, and nowhere else: it is a t27 module that is compiled and tested by t27c. This README states none of them. The checkpoint is final.pt, loaded with the model.py next to it, using tokenizer.json. Open weights apache-2.0 pytorch [View model](https://savrn.com/models/tern-tc-100m) Model · Text generation ### [Ternary-Bonsai-2-27B-Abliterated-PTQ1_0-GGUF](https://savrn.com/models/ternary-bonsai-2-27b-abliterated-ptq1-0-gguf) [Josh Bolding](https://savrn.com/model-publishers/boldingbuilds) Refusal-ablated Ternary Bonsai 2 27B, edited directly on the ternary lattice and shipped in PrismML's native PTQ10 pack. - 5,946,648,928 bytes — byte-for-byte the size of the official PTQ10 release - No requantization, no BF16 bake, no runtime steering, no control vector at serve time - Requires the PrismML llama.cpp fork Differs from PrismML's release in exactly 98 tensors and nowhere else. The other 753 tensors are byte-identical, including tokenembd and output. Anyone with both files can check that without trusting us. Paired against the untouched PrismML release: same card, same runtime, same flags, same seed, enablethinking: false on both arms. Judge is Qwen3.8-27B-OBLITERATED-Q80 with… Open weights apache-2.0 [View model](https://savrn.com/models/ternary-bonsai-2-27b-abliterated-ptq1-0-gguf) Model · Text generation ### [Ternary-Bonsai-2-27B-gguf](https://savrn.com/models/ternary-bonsai-2-27b-gguf) [Nick Pario](https://savrn.com/model-publishers/npario) Full 27B-class reasoning in ternary transformer weights, for llama.cpp (CUDA, Metal, CPU) - \~5.9 GB language model (down from \~54 GB FP16) — full 27B-class reasoning on a standard laptop or a single GPU - 98.2% of FP16 intelligence retained: 84.78 average across 14 thinking-mode benchmarks — far above the conventional IQ2XXS build (72.59) at less than two-thirds of its footprint, and within 0.4 points of UD-Q4KXL at three times the footprint - Retains thinking, reasoning, and agentic behavior deep in the sub-4-bit regime, where conventional low-bit representations collapse: math within half a point of full precision (96.57), coding level with the baseline (89.42), agentic tool calling at… Open weights apache-2.0 llama.cpp [View model](https://savrn.com/models/ternary-bonsai-2-27b-gguf) Model · Text generation ### [Ternary-Bonsai-2-27B-Refusal-Steering-GGUF](https://savrn.com/models/ternary-bonsai-2-27b-refusal-steering-gguf) [Mithril](https://savrn.com/model-publishers/com-kotobalabs) Experimental refusal-direction control vector for This repository does not contain a fully uncensored model. It contains a 1.3 MB GGUF control vector and the reproducibility material used to test refusal steering. The held-out evaluation did not support a reliable uncensored claim, so this artifact is published under its measured scope. normalized independently for layers 1–63 - At -2.0 one training prompt changed internal reasoning toward conditional compliance, but its final answer still refused - At -3.0 over layers 10–63 output quality collapsed - At -3.0 over layers 20–40 benign arithmetic remained correct (17 x 23 = 391), but 4/4 held-out prompts still refused These are small… Open weights apache-2.0 [View model](https://savrn.com/models/ternary-bonsai-2-27b-refusal-steering-gguf) Model · Text generation ### [Ternary-Bonsai-2-27B-vllm](https://savrn.com/models/ternary-bonsai-2-27b-vllm) [Fraser Price](https://savrn.com/model-publishers/fraserprice) Bonsai 2 27B by Prism ML, repacked for vLLM. Unofficial; not affiliated with Prism ML. It needs the prismternary vLLM plugin from fraserprice/bonsai-vllm, which has the run command, kernels and throughput numbers: Built and tested for the RTX PRO 6000 Blackwell only; other NVIDIA GPUs are untested. Problems: open an issue. - The ternary weights of the MLX pack, bit for bit: the same 2-bit codes and FP16 group scales (g128), in the same Hadamard-rotated basis. MLX's redundant per-group biases (-scale) are dropped, and the rotation signs move into config.json. - The embedding table is dequantized to BF16. - Norms and the linear-attention state path are Prism ML's own tensors from the MLX… Open weights apache-2.0 3.5B parameters 262,144 tokens vllm [View model](https://savrn.com/models/ternary-bonsai-2-27b-vllm) Model · Image and text to text ### [Ternary-Bonsai-2-4B-onwx](https://savrn.com/models/ternary-bonsai-2-4b-onwx) [Ryu Gyo](https://savrn.com/model-publishers/ryugyosoft) Prism ML の Ternary-Bonsai-2-4B（Qwen3.5-4B を三値の重みで学習し直したモデル）を、onwx 用に変換したものです。onwx は LLM を AMD Ryzen AI の NPU（XDNA2）だけで 動かすアプリで、このリポジトリにはモデルと、このモデル専用の NPU プログラムが入っています（アプリは別にインストールします）。 重みは三値（−1・0・+1）を 2 ビットのまま NPU に載せ、行列計算・DeltaNet・アテンション・LM ヘッドまで 1 トークンの計算をすべて NPU の 32 個の AI エンジンで行います。同じ大きさの int4 のモデルよりダウンロードもメモリも小さく、生成も速くなります。画像のエンコーダー（ViT）は CPU で動きます。NPU 向けのプログラムはビルド済みなので、PC の上でコンパイルを待つことはありません。 （onwx 0.7.0、Ryzen AI Max+ 395（Windows）で NPU の電力モードを performance にして測定。Ryzen AI 300 シリーズ（HX 370 など）も NPU は同じ 32 個の AI エンジンで、同じ方式で動きます。標準の電力モードでは NPU のクロックが下がって 2 割ほど遅くなります。電力モードは管理者の PowerShell で C:\Windows\System32\AMD\xrt-smi.exe configure --pmode performance。Windows の Copilot+ の AI 機能が NPU… Open weights apache-2.0 262,144 tokens [View model](https://savrn.com/models/ternary-bonsai-2-4b-onwx) Model ### [Ternary-Bonsai-2-9B-hf-bf16-GGUF](https://savrn.com/models/ternary-bonsai-2-9b-hf-bf16-gguf) [Team Mradermacher](https://savrn.com/model-publishers/mradermacher) static quants of https://huggingface.co/jonathan-kellerai/Ternary-Bonsai-2-9B-hf-bf16 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… Open weights apache-2.0 transformers [View model](https://savrn.com/models/ternary-bonsai-2-9b-hf-bf16-gguf) Model · Image and text to text ### [Ternary-Bonsai-2-9B-onwx](https://savrn.com/models/ternary-bonsai-2-9b-onwx) [Ryu Gyo](https://savrn.com/model-publishers/ryugyosoft) Prism ML の Ternary-Bonsai-2-9B（Qwen3.5-9B を三値の重みで学習し直したモデル）を、onwx 用に変換したものです。onwx は LLM を AMD Ryzen AI の NPU（XDNA2）だけで 動かすアプリで、このリポジトリにはモデルと、このモデル専用の NPU プログラムが入っています（アプリは別にインストールします）。 重みは三値（−1・0・+1）を 2 ビットのまま NPU に載せ、行列計算・DeltaNet・アテンション・LM ヘッドまで 1 トークンの計算をすべて NPU の 32 個の AI エンジンで行います。同じ大きさの int4 のモデルよりダウンロードもメモリも小さく、生成も速くなります。画像のエンコーダー（ViT）は CPU で動きます。NPU 向けのプログラムはビルド済みなので、PC の上でコンパイルを待つことはありません。 （onwx 0.7.0、Ryzen AI Max+ 395（Windows）で NPU の電力モードを performance にして測定。Ryzen AI 300 シリーズ（HX 370 など）も NPU は同じ 32 個の AI エンジンで、同じ方式で動きます。標準の電力モードでは NPU のクロックが下がって 2 割ほど遅くなります。電力モードは管理者の PowerShell で C:\Windows\System32\AMD\xrt-smi.exe configure --pmode performance。Windows の Copilot+ の AI 機能が NPU… Open weights apache-2.0 262,144 tokens [View model](https://savrn.com/models/ternary-bonsai-2-9b-onwx) Model · Text generation ### [Ternary-Bonsai-8B-gguf-fork](https://savrn.com/models/ternary-bonsai-8b-gguf-fork) [Pythai.net](https://savrn.com/model-publishers/pythai) Ternary (1.58-bit) language model in GGUF Q20 format for llama.cpp - Demo repo — examples for serving, benchmarking, and integrating Bonsai - Discord — community support and updates Each weight takes a value from {-1, 0, +1}, with one shared FP16 scale per group of 128 weights: Q20 encodes each weight as a 2-bit code q in {0, 1, 2, 3}, dequantized via w = (q - 1) scale. One 128-element block is 34 bytes (2 bytes FP16 scale + 32 bytes of packed 2-bit codes) for an effective 2.125 bits/weight. The fourth code point (q = 3, reconstructing to +2 scale) is reserved for future extensions; for ternary weights it is unused. Evaluated with EvalScope v1.4.2 + vLLM 0.15.1 on NVIDIA H100 under… Open weights apache-2.0 gguf [View model](https://savrn.com/models/ternary-bonsai-8b-gguf-fork) Model ### [terramind-qwen2.5-coder-1.5b-terraform](https://savrn.com/models/terramind-qwen2-5-coder-1-5b-terraform) [Shubh Om Yadav](https://savrn.com/model-publishers/shubh-3011) A small, local Terraform (HCL) generator that runs on a laptop CPU or GPU with no cloud API. This is the model bundled with TerraMind, a Code-OSS fork for Terraform. On 12 held-out examples through the shipped GGUF engine: ~83 % parse, ~33 % pass terraform validate. This is a review-required draft assistant, not a correctness guarantee. A larger retrain (v3) did not improve on it. Always run terraform validate / terraform plan before applying generated code. TerraMind is the reliable path for structured requests. - Trained on real-world Terraform that leans AWS; Azure/GCP coverage is thinner. - Not a security, reliability, cost, or deployment guarantee. in the TerraMind repository… Open weights apache-2.0 [View model](https://savrn.com/models/terramind-qwen2-5-coder-1-5b-terraform) Model ### [Tess-3-Mixtral-8x22B-GGUF](https://savrn.com/models/tess-3-mixtral-8x22b-gguf) [Team Mradermacher](https://savrn.com/model-publishers/mradermacher) static quants of https://huggingface.co/migtissera/Tess-3-Mixtral-8x22B For a convenient overview and download list, visit our model page for this model. weighted/imatrix quants are available at https://huggingface.co/mradermacher/Tess-3-Mixtral-8x22B-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… Open weights apache-2.0 transformers [View model](https://savrn.com/models/tess-3-mixtral-8x22b-gguf) Model · Image and text to text ### [Tess-4.5-27B-exl3](https://savrn.com/models/tess-4-5-27b-exl3) [Everson Alves da Silva](https://savrn.com/model-publishers/khronnuz) Open weights apache-2.0 [View model](https://savrn.com/models/tess-4-5-27b-exl3) Model · Image to text ### [tesseract-kaa-cyrl](https://savrn.com/models/tesseract-kaa-cyrl) [Kaa ML](https://savrn.com/model-publishers/kaa-ml) Tesseract 5 LSTM model for printed Karakalpak in the Cyrillic alphabet, including the letters Ә Ғ Қ Ң Ө Ү Ў Ҳ. Fine-tuned from Tesseract's Kazakh kaz (tessdatabest). For Latin-script Karakalpak use kaa-ml/tesseract-kaa-latn (CER 4.7%). Before this model, the usual choice for Karakalpak Cyrillic was the Kazakh model. It has no ў and ҳ, so it silently writes у and х instead — the text looks right but is misspelled. The Russian model loses all eight Karakalpak letters. 10 scanned pages of an old printed Karakalpak book (never used in training), --psm 4. CER = character error rate (lower is better). Letter recall = share of ә ғ қ ң ө ү ў ҳ in the reference that the model output contains. \ on… Open weights apache-2.0 tesseract [View model](https://savrn.com/models/tesseract-kaa-cyrl) Model · Image to text ### [tesseract-kaa-latn](https://savrn.com/models/tesseract-kaa-latn) [Kaa ML](https://savrn.com/model-publishers/kaa-ml) Tesseract LSTM model for printed Karakalpak in the current Latin alphabet, including á ǵ ı ń ó ú. Fine-tuned from Tesseract's English model, with a Karakalpak word list, punctuation and number dictionaries built in. For Cyrillic Karakalpak use kaa-ml/tesseract-kaa-cyrl. 12 pages from digital Karakalpak (Latin) academic PDFs, rendered at 300 DPI; reference = the PDF's own text layer. --psm 3. CER = character error rate (lower is better). Letter recall = share of á ó ú ı ń ǵ in the reference found in the output. These are clean digital documents; old or poor scans will score worse. Speed: ~3.5 s per A4 page on a MacBook CPU. Fine-tuned in 2023 from eng (Tesseract 4.1.1 format, works with… Open weights apache-2.0 tesseract [View model](https://savrn.com/models/tesseract-kaa-latn) Model ### [test](https://savrn.com/models/test-2) [Yangyingjie](https://savrn.com/model-publishers/yangyingjie1997) Open weights apache-2.0 [View model](https://savrn.com/models/test-2) Model ### [test-contrastive](https://savrn.com/models/test-contrastive) [Adrian Kowalski](https://savrn.com/model-publishers/ackowalski) A small DeiT implementation for Contrastive, packaged with an explicit configuration and an initialization checkpoint. The nano variant is a reproducible starting point, not a trained model release. - The Python file contains the model and runnable example or training entry point. - config.json records the generated architecture settings. - trainingargs.json records the default experiment recipe. - model.safetensors is a valid initialization checkpoint for smoke tests; it is not presented as a trained benchmark checkpoint. - No benchmark score is claimed in this repository. The included configuration uses adamw with a step schedule. These are starting values in the script, not evidence of a… Open weights apache-2.0 24,832 parameters 512 tokens [View model](https://savrn.com/models/test-contrastive) Model · Text generation ### [testr-100k](https://savrn.com/models/testr-100k) [Compactbot](https://savrn.com/model-publishers/compactbot) A 100,936-parameter character-level GPT trained from scratch. Fulfils model request #17 from @GGUFGuy. A minimal nanoGPT-style causal transformer operating at the character level (128-char vocabulary). The point of this specific model is a memorization experiment: train a very small model on a held-out validation set and measure how much it actually memorizes versus generalizing. (52 tensors in the file; the head is tied to the embedding, so counting it once gives 100,936. The earlier "106,568" double-counted the tied head.) The question: does a 100K-parameter model trained only on a 19 MB validation set actually memorize it, or does it just learn character statistics? Measured perplexity… Open weights apache-2.0 106,568 parameters [View model](https://savrn.com/models/testr-100k) Model ### [TestSubject-2](https://savrn.com/models/testsubject-2) [Mr. Black Beard](https://savrn.com/model-publishers/cyanmonkey) Open weights apache-2.0 99M parameters [View model](https://savrn.com/models/testsubject-2) ## Questions ### Can I use Apache License 2.0 models commercially? Yes. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors. ### Which Apache License 2.0 models are most downloaded? By monthly downloads reported by the Hugging Face Hub: . ## Other Licenses - [mit](https://savrn.com/models/licenses/mit) 1305 - [cc-by-nc-4.0](https://savrn.com/models/licenses/cc-by-nc-4-0) 209 - [cc-by-4.0](https://savrn.com/models/licenses/cc-by-4-0) 164 - [bsd-3-clause](https://savrn.com/models/licenses/bsd-3-clause) 108 - [gemma](https://savrn.com/models/licenses/gemma) 63 - [cc-by-nc-sa-4.0](https://savrn.com/models/licenses/cc-by-nc-sa-4-0) 52 - [llama3.2](https://savrn.com/models/licenses/llama3-2) 40 - [agpl-3.0](https://savrn.com/models/licenses/agpl-3-0) 32 - [unknown](https://savrn.com/models/licenses/unknown) 32 - [llama3.1](https://savrn.com/models/licenses/llama3-1) 28 - [cc-by-sa-4.0](https://savrn.com/models/licenses/cc-by-sa-4-0) 20 - [gpl-3.0](https://savrn.com/models/licenses/gpl-3-0) 20 [See all](https://savrn.com/models/licenses)