v0.9.2

What's Changed

• fix: correct PaddleOCR-VL and PP-DocLayout on Metal by @GreatV in #178
• refactor(vl): query Metal memory via Candle by @GreatV in #179
• Optimize OCR batching and simplify pipeline APIs by @GreatV in #180
• fix(vl): stop CUDA-graph disposal from poisoning the stream-ordered allocator by @GreatV in #181
• build: bump version to 0.9.2 by @GreatV in #182

Full Changelog: v0.9.1...v0.9.2

v0.9.1

What's Changed

• build(deps): update fontdb requirement from 0.23 to 0.24 by @dependabot[bot] in #172
• build: prepare v0.9.1 dependency updates by @GreatV in #175

Full Changelog: v0.9.0...v0.9.1

v0.9.0

What's Changed

• Make oar-ocr-vl standalone with native PP-DocLayout by @GreatV in #171

Full Changelog: v0.8.1...v0.9.0

v0.8.1

What's Changed

• feat: support HunyuanOCR 1.5 by @GreatV in #156
• feat(vl): accelerate HunyuanOCR 1.5 with DFlash by @GreatV in #158
• perf(vl): optimize HunyuanOCR decoding by @GreatV in #159
• perf(vl): accelerate GLM-OCR, MinerU, and PaddleOCR-VL by @GreatV in #160
• docs: refresh project documentation by @GreatV in #161
• docs: curate public model references by @GreatV in #162
• feat(vl): add OvisOCR2 support by @GreatV in #163
• Add MonkeyOCRv2 S/B Parsing support by @GreatV in #164
• Improve Metal and CoreML inference performance by @GreatV in #166
• Optimize Windows OCR performance by @GreatV in #165
• Add native HPD-Parsing inference by @GreatV in #168
• chore: bump version to 0.8.1 by @GreatV in #169

Full Changelog: v0.8.0...v0.8.1

v0.8.0

What's Changed

• fix: remove text recognition length limit by @GreatV in #150
• refactor(predictors): centralize auto-download path resolution by @GreatV in #152
• feat: support loading models from memory by @GreatV in #153
• docs: add FAQ, document in-memory model loading; bump version to 0.8.0 by @GreatV in #154

Full Changelog: v0.7.4...v0.8.0

v0.7.4

What's Changed

• fix(vl): avoid OOM in PaddleOCR-VL vision attention on low-VRAM GPUs by @GreatV in #148
• chore: bump version to 0.7.4, document environment variables, trim verbose comments by @GreatV in #149

Full Changelog: v0.7.3...v0.7.4

v0.7.3

What's Changed

• fix(vl): avoid BF16 on unsupported CUDA devices by @GreatV in #144
• ci: pin macOS runner to macos-26 to avoid image drift breaking cached builds by @GreatV in #145
• bump version to 0.7.3 by @GreatV in #146

Full Changelog: v0.7.2...v0.7.3

v0.7.2

What's Changed

• build(deps): update candle-transformers requirement from 0.10.2 to 0.11.0 by @dependabot[bot] in #141
• build(deps): update candle-core requirement from 0.10.2 to 0.11.0 by @dependabot[bot] in #140
• build(deps): update itertools requirement from 0.14 to 0.15 by @dependabot[bot] in #137
• refactor: replace crop_rotated_bounding_box with get_rotate_crop_image in BBoxCrop by @GreatV in #142

Full Changelog: v0.7.1...v0.7.2

v0.7.1

What's Changed

• feat: add PP-OCRv6 model support and optimize inference performance by @GreatV in #135
• build(deps): update imageproc requirement from 0.26.0 to 0.27.0 by @dependabot[bot] in #134
• chore(release): bump version to 0.7.1 by @GreatV in #136

Full Changelog: v0.7.0...v0.7.1

v0.7.0

What's Changed

• Add Windows CI for tests by @Copilot in #111
• perf: reduce image copies in OCR pipeline by @GreatV in #112
• build(deps): update tokenizers requirement from 0.22 to 0.23 by @dependabot[bot] in #113
• Support hierarchical speculative decoding and align inference outputs by @GreatV in #116
• build(deps): replace clipper2 with clipper2-rust by @GreatV in #117
• feat: auto-download ONNX models from ModelScope by @GreatV in #118
• Update deps and adapt Hayro 0.6 rendering by @GreatV in #119
• chore: update hayro dependency to version 0.7 and adjust related depe… by @GreatV in #121
• fix: improve clarity and formatting in HSD documentation by @GreatV in #122
• Fix math formatting in hsd.md by @GreatV in #123
• Fix link to ModelScope in README by @GreatV in #124
• Enhance image processing and OCR capabilities by @GreatV in #126
• build(deps): update nalgebra requirement from 0.34 to 0.35 by @dependabot[bot] in #125
• feat: update dependencies to use workspace features for improved management by @GreatV in #127
• refactor(oar-ocr-vl): dedupe weight loading, config, and vision RoPE helpers by @GreatV in #128
• fix: stabilize formula/layout inference, dedup VL masks, and tidy code & docs by @GreatV in #129
• refactor(oar-ocr-vl)!: remove Hierarchical Speculative Decoding (HSD) by @GreatV in #130
• Optimize DB layout postprocessing and align detection with official implementations by @GreatV in #131
• feat: add PP-OCRv6 models with direct links from PaddlePaddle's official channels by @GreatV in #132
• feat(core): SIMD-accelerate hot CPU normalization and CTC-decode kernels by @GreatV in #133

Full Changelog: v0.6.3...v0.7.0