Improve model card and surface related exports

README.md

@@ -14,22 +14,19 @@ tags:
 
 # YOLOX-RKNN2
 
-Rockchip Model Zoo conversion of YOLOX Nano for RK3576 and RK3588. YOLOX Nano FP16 and INT8 models for RK3576 and RK3588. Download `model.rknn` and `metadata.yaml` from the same directory under `models/v2.5.0/detect/yolox_nano/<chip>/<precision>/`. The [manifest](manifest.json) records each model hash.
+> **Looking for Ultralytics RKNN exports?** Browse the [Ultralytics exports collection](https://huggingface.co/collections/HanzoHuang/ultralytics-rknn-exports-6af1d088ba04ccf2dcbf81c2).
 
-The [original YOLOX Nano checkpoint](https://github.com/Megvii-BaseDetection/YOLOX/releases/download/0.1.1rc0/yolox_nano.pth) is included under [source](tree/main/source). It was exported to ONNX with the [Rockchip YOLOX exporter](https://github.com/airockchip/YOLOX/tree/fc69c1e2ccb30fce7e18250bddc13a68090dca09) and compiled with the [Model Zoo YOLOX converter](https://github.com/airockchip/rknn_model_zoo/tree/837db8e992815ff8647eadb709f07067f4bc3a71/examples/yolox) using RKNN Toolkit2 2.5.0. The conversion changed the model format and quantized the INT8 variant.
+Rockchip Model Zoo conversions of YOLOX Nano for **RK3576** and **RK3588**, with FP16 and INT8 variants.
 
-YOLOX is from [Megvii-BaseDetection](https://github.com/Megvii-BaseDetection/YOLOX). The [Apache-2.0 license](LICENSE) applies to this distribution.
+## Choose a model
 
-Models are organized by RKNN Toolkit2 version under `models/v<version>/`. Each model and its metadata share a folder.
+For Toolkit 2.5.0, select a directory under `models/v2.5.0/detect/yolox_nano/<chip>/<precision>/` and download `model.rknn` and `metadata.yaml` together. The [manifest](manifest.json) records model hashes.
 
-## Model Zoo downloads
+- [Toolkit 2.5.0 builds](tree/main/models/v2.5.0) use RKNN Toolkit2 2.5.0. Conversion receipts identify the original ONNX model and converter.
+- [Toolkit 2.3.2 builds](tree/main/models/v2.3.2) were tested on their target board. Each receipt identifies its source graph and compiler.
 
-The `models/v2.5.0/` downloads use RKNN Toolkit2 2.5.0. The conversion receipts identify the original ONNX and converter. See [source and license notices](LICENSES.md).
+## Conversion and license
 
-## Toolkit 2.3.2 builds
+The [original YOLOX Nano checkpoint](https://github.com/Megvii-BaseDetection/YOLOX/releases/download/0.1.1rc0/yolox_nano.pth) is included in [source](tree/main/source). It was exported to ONNX with the [Rockchip YOLOX exporter](https://github.com/airockchip/YOLOX/tree/fc69c1e2ccb30fce7e18250bddc13a68090dca09) and compiled with the [Model Zoo YOLOX converter](https://github.com/airockchip/rknn_model_zoo/tree/837db8e992815ff8647eadb709f07067f4bc3a71/examples/yolox) and RKNN Toolkit2 2.5.0. The INT8 variant was quantized during conversion.
 
-See [models/v2.3.2](tree/main/models/v2.3.2) for builds tested on their target board. Each conversion receipt identifies its source graph and compiler.
-
-## Related RKNN exports
-
-This repository is part of the RKNN Model Zoo YOLO exports collection, which includes Rockchip conversions and custom extensions. For models exported with the Ultralytics workflow, see the [Ultralytics RKNN exports collection](https://huggingface.co/collections/HanzoHuang/ultralytics-rknn-exports-6af1d088ba04ccf2dcbf81c2).
+YOLOX is from [Megvii-BaseDetection](https://github.com/Megvii-BaseDetection/YOLOX). See [LICENSE](LICENSE) and [source and license notices](LICENSES.md) for the applicable terms.