Yolo26 Safetensors
For all runtime usage, Rust APIs, and examples for every YOLO26 / YOLOE-26 task, see .
Fonte do modelo
- Tipo
- Other
- Tamanho do arquivo
- 6,30–329 MB
- Modelo base
- Ultralytics/YOLO26
- Versão
- vRevision cd6aa1a
- Atualizado
- 17 de set. de 2026
- Downloads
- 0
- Fonte
- Hugging Face
Trecho da fonte
For all runtime usage, Rust APIs, and examples for every YOLO26 / YOLOE-26 task, see .
Fontes
1 fonte
Fontes
1 fonteVerificado 2 de set.
Artefatos de modelo
40 artefatos
Artefatos de modelo
40 artefatosTrechos de fonte
2 trechos
Trechos de fonte
2 trechos--- license: agpl-3.0 base_model: Ultralytics/YOLO26 tags: - yolo - yolo26 - yoloe - safetensors - object-detection - image-segmentation - image-classification - pose-estimation - oriented-object-detection - semantic-segmentation --- # YOLO26 SafeTensors [中文](README_CN.md) For all runtime usage, Rust APIs, and examples for every YOLO26 / YOLOE-26 task, see <https://github.com/wzh19960613/yolo26-rs>. This repository contains safetensors-converted weights from the official Ultralytics YOLO26 release. Tensor keys, shapes, dtypes, and values are preserved from the upstream `state_dict`; only the serialization format is changed. **Original source:** https://huggingface.co/Ultralytics/YOLO26 **License:** AGPL-3.0 (same as upstream) ## Weights File naming convention: ``` yolo26{size}.safetensors # detection yolo26{size}-{task}.safetensors # cls / seg / pose / sem / obb yoloe-26{size}-seg.safetensors # YOLOE text / visual prompt yoloe-26{size}-seg-pf.safetensors # YOLOE prompt-free LRPC ``` **Sizes:** n, s, m, l, x **Tasks:** detect, cls, seg, pose, sem, obb, yoloe-seg, yoloe-seg-pf ## Re-export Requires Python 3.10+ with `ultralytics` and `safetensors` installed: ```bash pip install ultralytics safetensors ``` Export specific models: ```bash # single: yolo26n.pt -> .safetensors python export-from-ultralytics.py -s n # multiple: yolo26n.pt, yolo26s.pt, yolo26n-seg.pt, yolo26s-seg.pt -> .safetensors python export-from-ultralytics.py -s n s -t detect seg # YOLOE text/visual and prompt-free: # yoloe-26n-seg.pt, yoloe-26n-seg-pf.pt -> .safetensors python export-from-ultralytics.py -s n -t yoloe-seg yoloe-seg-pf # all combinations python export-from-ultralytics.py -s all -t all # output directory (default: current directory) python export-from-ultralytics.py -s n -t detect -o public/models # delete .pt files downloaded during this run after successful export python export-from-ultralytics.py -s n -t yoloe-seg --clean ``` Existing valid `.safetensors` files are skipped automatically. Re-run to re-export corrupted files. By default, official `.pt` files downloaded by Ultralytics are kept in the current working directory. Pass `--clean` to delete only the `.pt` files downloaded during that run after their `.safetensors` export succeeds.
Source context: 0 downloads · 0 likes · Pipeline image-classification · Repo wzh19960613/yolo26-safetensors
