docs: curate public model references

README.md

@@ -30,7 +30,7 @@ cargo add oar-ocr --features cuda,auto-download
 
 This keeps the default `download-binaries` and `simd` features enabled, makes the ONNX Runtime CUDA execution provider available for selection, and downloads missing registered model files from ModelScope into `$OAR_HOME` when they are first used.
 
-See the [Cargo feature guide](docs/features.md) for all available features and the [model guide](docs/models.md#auto-download-via-the-auto-download-feature) for model download and cache behavior.
+See the [Cargo feature guide](docs/features.md) for all available features and the [model guide](docs/models.md#auto-download) for model download and cache behavior.
 
 Builders also accept raw ONNX bytes such as `include_bytes!`, allowing models to be embedded in a single binary. See [Loading Models from Memory](docs/usage.md#loading-models-from-memory).
 
@@ -113,7 +113,7 @@ The classic pipeline runs ONNX models through ONNX Runtime and supports the foll
 
 Available text-recognition checkpoints cover Chinese, Traditional Chinese, English, Arabic, Cyrillic, Devanagari, Greek, Eastern Slavic, Japanese, Georgian, Korean, Latin, Tamil, Telugu, and Thai scripts or languages.
 
-### Vision-Language Models (VLM)
+### Vision-Language Models
 
 The [`oar-ocr-vl`](oar-ocr-vl/README.md) crate provides native [Candle](https://github.com/huggingface/candle) inference for compact document VLMs on CPU, CUDA, and Metal.
 

docs/FAQ.md

@@ -4,29 +4,18 @@ Common build and runtime questions, distilled from GitHub issues.
 
 ## Windows: linking fails with `LNK2001: unresolved external symbol __std_find_trivial_8`
 
-Unresolved `__std_*` symbols (`__std_find_trivial_*`, `__std_max_element_*`,
-`__std_search_*`, ...) mean the linker's MSVC STL is too old. The prebuilt
-ONNX Runtime binaries that `ort-sys` downloads are compiled with Visual
-Studio 2022 (MSVC v143) and reference vectorized STL helpers that do not
-exist in the VS 2019 (v142) link libraries.
+Unresolved `__std_*` symbols (`__std_find_trivial_*`, `__std_max_element_*`, `__std_search_*`, ...) mean the linker's MSVC STL is too old. The prebuilt ONNX Runtime binaries that `ort-sys` downloads are compiled with Visual Studio 2022 (MSVC v143) and reference vectorized STL helpers that do not exist in the VS 2019 (v142) link libraries.
 
 Fix:
 
-1. Install **Visual Studio 2022** or **Build Tools for Visual Studio 2022**
-   with the "Desktop development with C++" workload (MSVC v143 + Windows SDK).
-2. Run `cargo clean` and rebuild — rustc picks the newest installed MSVC
-   toolset automatically.
+1. Install **Visual Studio 2022** or **Build Tools for Visual Studio 2022** with the "Desktop development with C++" workload (MSVC v143 + Windows SDK).
+2. Run `cargo clean` and rebuild. Rustc picks the newest installed MSVC toolset automatically.
 
-VS 2019 and VS 2022 build tools can coexist; only the newer one needs to be
-present for linking. (See issue [#105](https://github.com/GreatV/oar-ocr/issues/105).)
+VS 2019 and VS 2022 build tools can coexist. Only the newer one needs to be present for linking. See issue [#105](https://github.com/GreatV/oar-ocr/issues/105).
 
 ## GPU inference is slower than CPU for PP-OCRv6 tiny/small
 
-Expected. The tiny/small models are so small that per-call overhead —
-host↔device tensor copies, kernel launches, CPU/GPU synchronization, plus
-pre/post-processing that always runs on the CPU — outweighs the compute the
-GPU saves. Measured on an RTX 4090 + i9-13900KF (single image, warmup
-excluded):
+Expected. The tiny and small models are so compact that per-call overhead, including host-to-device tensor copies, kernel launches, CPU/GPU synchronization, and CPU pre/post-processing, can outweigh the computation saved by the GPU. The following results were measured on an RTX 4090 with an i9-13900KF using a single image and excluding warmup:
 
 | Model  | CPU        | GPU (CUDA EP)             |
 | ------ | ---------- | ------------------------- |
@@ -37,13 +26,7 @@ excluded):
 Guidelines:
 
 - For tiny/small, use the default CPU mode (the `simd` feature is on by default).
-- Use the medium model, or batch several images per `predict()` call, when
-  GPU acceleration matters.
-- Exclude the first call when benchmarking: it includes cuDNN initialization
-  and algorithm selection (~5× slower than steady state).
-
-Also note that requesting `OrtExecutionProvider::CUDA` without building with
-`--features cuda` makes the pipeline builder return an error — check the
-`Result` of `.build()`. Without the `cuda` feature, the downloaded ONNX
-Runtime is CPU-only and the GPU is never used.
-(See issue [#151](https://github.com/GreatV/oar-ocr/issues/151).)
+- Use the medium model, or batch several images per `predict()` call, when GPU acceleration matters.
+- Exclude the first call when benchmarking because it includes cuDNN initialization and algorithm selection (~5× slower than steady state).
+
+Also note that requesting `OrtExecutionProvider::CUDA` without building with `--features cuda` makes the pipeline builder return an error. Check the `Result` of `.build()`. Without the `cuda` feature, the downloaded ONNX Runtime is CPU-only and the GPU is never used. See issue [#151](https://github.com/GreatV/oar-ocr/issues/151).

docs/environment-variables.md

@@ -28,7 +28,7 @@ Runtime environment variables read by the oar-ocr crates. Project-specific varia
 
 ## `OAR_HOME`
 
-Directory where the `auto-download` feature caches model files (default `~/.oar`). Bare model file names passed to the builders are downloaded here and verified against their expected SHA-256. See [models.md](models.md#auto-download-via-the-auto-download-feature) for the exact path resolution rules.
+Directory where the `auto-download` feature caches model files (default `~/.oar`). Bare model file names passed to the builders are downloaded here and verified against their expected SHA-256. See [models.md](models.md#auto-download) for the exact path resolution rules.
 
 ```bash
 OAR_HOME=/data/oar-models cargo run --release --example ocr -- doc.jpg

docs/features.md

@@ -159,7 +159,7 @@ If an execution provider feature is enabled, the supplied ONNX Runtime build mus
 cargo add oar-ocr --features auto-download
 ```
 
-Existing files and explicit paths remain under caller control. In-memory ONNX sources also bypass the download resolver. See the [model guide](models.md#auto-download-via-the-auto-download-feature) for the complete path-resolution and cache rules.
+Existing files and explicit paths remain under caller control. In-memory ONNX sources also bypass the download resolver. See the [model guide](models.md#auto-download) for the complete path-resolution and cache rules.
 
 `auto-download` supplies model files at runtime. It does not install ONNX Runtime or any accelerator dependencies.