AISpot

llama.cpp 新增 MiniCPM-OCR-Jev 支援

llama.cpp releases版本更新地端推論開源

llama.cpp 建置 b11928 加入 MiniCPM-OCR-Jev,這是一個只打分、不生成文字的 OCR 模型。它以 Qwen3.5-VL 為骨幹,在最後一個 token 上讀取 value head,為影像與其渲染文字的匹配度打分,通過 POST /v1/systemone 的 openbmb-minicpm-ocr-jev 決策型別提供服務。checkpoint 沒有 LM head,因此文字生成請求會被拒絕;

llama.cpp 接入打分式而非生成式的 OCR 模型,走 /v1/systemone 新決策型別,並保留每圖 16384 token 的影像預算,對在本地搭建 OCR 結果校驗流程的開發者最直接。

原標題:b11928
閱讀原文

評分58 / 58(平均 58,門檻 60)
狀態未入選

原文

model: Add MiniCPM-OCR-Jev ( #30980 ) model : add MiniCPM-OCR-Jev support The model is a Qwen3.5-VL trunk with a value head read at the last token: it scores an OCR text against an image rendered from it, it does not generate. Add the openbmb-minicpm-ocr-jev decision type, served by POST /v1/systemone. The class scores are the pooled output of the sequence, read with last pooling, so the conversion turns score.weight into cls.output and sets embedding_length_out. The checkpoint has no LM head, so text generation is rejected instead of using the output weights duplicated from the token embeddings. The task is fixed, so its prompt is entirely in the systemone template, the same way h2o-lightning keeps its default there. The processor of the checkpoint allows up to 16384 merged tokens per image, more than the Qwen-VL default of mtmd, so the mmproj keeps the image budget of the preprocessor config. The checkpoint declares MiniCPMOCRForSequenceClassification in its config.json, as other models do. Signed-off-by: tc-mb [email protected] fix code Signed-off-by: tc-mb [email protected] use gguf meta classifier_output_labels remove redudant code changes Signed-off-by: tc-mb [email protected] Co-authored-by: Xuan Son Nguyen [email protected] Website: https://llama.app Attestations: https://github.com/ggml-org/llama.cpp/attestations/60366505 macOS/iOS: macOS Apple Silicon (arm64) macOS Apple Silicon (arm64, KleidiAI enabled) DISABLED macOS Intel (x64) iOS XCFramework Linux: Ubuntu x64 (CPU) Ubuntu arm64 (CPU) Ubuntu s390x (CPU) Ubuntu x64 (Vulkan) Ubuntu arm64 (Vulkan) Ubuntu x64 (CUDA 12) - CUDA 12.8 libraries Ubuntu x64 (CUDA 13) - CUDA 13.4 libraries Ubuntu arm64 (CUDA 13) - CUDA 13.4 libraries Ubuntu x64 (ROCm 10.0) Ubuntu x64 (OpenVINO) Ubuntu x64 (SYCL FP32) Ubuntu x64 (SYCL FP16) Linux arm64 (Snapdragon: CPU, Adreno GPU, Hexagon NPU) - setup guide Android: Android arm64 (CPU) Android arm64 (Snapdragon: CPU, Adreno GPU, Hexagon NPU) - setup guide Windows: Windows x64 (CPU) Windows arm64 (CPU) Windows arm64 (OpenCL Adreno) Windows x64 (CUDA 12) - CUDA 12.4 DLLs Windows x64 (CUDA 13) - CUDA 13.4 DLLs Windows arm64 (CUDA 13) - CUDA 13.4 DLLs Windows x64 (Vulkan) Windows arm64 (Vulkan) Windows x64 (OpenVINO) Windows x64 (SYCL) Windows x64 (ROCm 10.0) openEuler: DISABLED openEuler x86 (310p) openEuler x86 (910b, ACL Graph) openEuler aarch64 (310p) openEuler aarch64 (910b, ACL Graph) UI: UI

相關報導

llama.cpp releasesAI 評分53

llama.cpp b11826 新增 MiniCPM-OCR 1.0 支援

llama.cpp 建置 b11826 在 mtmd 多模態模組中加入 MiniCPM-OCR 1.0 支援,本地推理可直接用上這一 OCR 模型。同批預編譯包覆蓋 macOS(Apple Silicon 與 Intel)、iOS、Android、Windows、Linux 及 openEuler,包含 CUDA 12/13、Vulkan、ROCm 10.0、SYCL、OpenVINO 與驍龍 CPU/Adreno GPU/Hexagon NPU 後端。

量子位● 精選AI 評分67

1.3B 的 OXR 開源登頂三項文件解析榜,速度超 PaddleOCR、MinerU

1.3B 引數的開源文件解析模型 OXR 在 OmniDocBench、olmOCR-Bench、DocParsingBench 三項基準上分別取得 97.18、87.84、87.76 分,均列第一。它覆蓋文字、表格、公式、化學結構式等 13 類元素,輸出 Markdown / JSON 並保留版面座標,公式轉 LaTeX、表格轉 HTML。

量子位● 精選AI 評分79

百度開源文心 Unlimited OCR 2,峰值 KV 快取降約 96%

百度發布並開源文心 Unlimited OCR 2,百頁級單序列峰值 KV 快取從 47GB 降到 1.77GB,減少約 96%。它靠稀疏參考滑動視窗注意力 SR-SWA 與分塊無限解碼,可連續解析 3600 頁以上文件,16 路併發吞吐從 360 提升到 1315 Token/秒。MarathonDocBench 得 77.06、MPDocBench 得 84.59,單步解碼速度分別為全注意力和上代 R-SWA 的 7.0 倍與 5.6 倍。