Unlimited OCR Mixed NVFP4

--- license: mit language: - en - zh tags: - ocr - document-parsing - mlx - krill - deepseek-ocr - nvfp4 pipeline_tag: image-to-text base_model: baidu/Unlimited-OCR library_name: krill --- # Unlimited-OCR — Krill mixed-nvfp4 (MLX) A **Mac-native (Apple Silicon / MLX)** mixed-precision build of [`baidu/Unlimited-OCR`](https://huggingface.co/baidu/Unlimited-OCR) (built on DeepSeek-OCR), packaged for the [**Krill**](https://github.com/srvsngh99/Krill) runtime. It parses documents and images to **grounded text** with **no Python and no `trust_remote_code`** — the DeepSeek-MoE language backbone *and* the DeepEncoder vision tower (SAM-ViT-B + CLIP-L + projector) run natively in Swift + MLX. ## Run with Krill ```bash krill pull unlimited-ocr krill run unlimited-ocr --image your_page.png "document parsing." ``` It emits grounded OCR, e.g.: ``` <|det|>title [48, 74, 402, 130]<|/det|>Invoice 2026 <|det|>text [33, 229, 370, 290]<|/det|>Bill to: Acme Corporation ... ``` ## Format Mixed-precision, single-file `model.safetensors` (~2.3 GB, from the 6.67 GB bf16 source): | Modules | Precision | |---|---| | MoE experts (the residency-dominant bulk) | **nvfp4** (group 16) | | attention q/k/v/o, dense + shared FFN, embed, lm_head | 8-bit affine (group 64) | | DeepEncoder vision Linears (SAM / CLIP / projector) | 8-bit affine | | Conv2d kernels, norms, position/learned embeddings, router gate | unquantized | This format is **Krill-specific** (the native Swift+MLX runtime resolves the per-module quantization); it is not a drop-in `transformers` checkpoint. ## Scope Ships the **base view** (single 1024 global view → 273 image tokens), which reads full pages including wide layouts. The "gundam" tiling mode (local crops for very dense/large scans) is a follow-up. ## Credit & license All credit to the original authors, [`baidu/Unlimited-OCR`](https://huggingface.co/baidu/Unlimited-OCR) and DeepSeek-OCR. MIT licensed, as upstream. --- ## About this build Quantized and tested by [Sourav Singh](https://souravailabs.ai) for [Krill](https://github.com/srvsngh99/Krill), a Mac-native LLM runtime and coding agent in one Swift + MLX binary. More Apple Silicon builds: [huggingface.co/srv-sngh](https://huggingface.co/srv-sngh). Notes on building with local models and frontier AI: [souravailabs.ai](https://souravailabs.ai) · [GitHub](https://github.com/srvsngh99) · [LinkedI...

--- license: mit language: - en - zh tags: - ocr - document-parsing - mlx - krill - deepseek-ocr - nvfp4 pipeline_tag: image-to-text base_model: baidu/Unlimited-OCR library_name: krill --- # Unlimited-OCR — Krill mixed-nvfp4 (MLX) A **Mac-native (Apple Silicon / MLX)** mixed-precision build of [`baidu/Unlimited-OCR`](https://huggingface.co/baidu/Unlimited-OCR) (built on DeepSeek-OCR), packaged for the [**Krill**](https://github.com/srvsngh99/krill) runtime. It parses documents and images to **grounded text** with **no Python and no `trust_remote_code`** — the DeepSeek-MoE language backbone *and* the DeepEncoder vision tower (SAM-ViT-B + CLIP-L + projector) run natively in Swift + MLX. ## Run with Krill ```bash krill pull unlimited-ocr krill run unlimited-ocr --image your_page.png "document parsing." ``` It emits grounded OCR, e.g.: ``` <|det|>title [48, 74, 402, 130]<|/det|>Invoice 2026 <|det|>text [33, 229, 370, 290]<|/det|>Bill to: Acme Corporation ... ``` ## Format Mixed-precision, single-file `model.safetensors` (~2.3 GB, from the 6.67 GB bf16 source): | Modules | Precision | |---|---| | MoE experts (the residency-dominant bulk) | **nvfp4** (group 16) | | attention q/k/v/o, dense + shared FFN, embed, lm_head | 8-bit affine (group 64) | | DeepEncoder vision Linears (SAM / CLIP / projector) | 8-bit affine | | Conv2d kernels, norms, position/learned embeddings, router gate | unquantized | This format is **Krill-specific** (the native Swift+MLX runtime resolves the per-module quantization); it is not a drop-in `transformers` checkpoint. ## Scope Ships the **base view** (single 1024 global view → 273 image tokens), which reads full pages including wide layouts. The "gundam" tiling mode (local crops for very dense/large scans) is a follow-up. ## Credit & license All credit to the original authors, [`baidu/Unlimited-OCR`](https://huggingface.co/baidu/Unlimited-OCR) and DeepSeek-OCR. MIT licensed, as upstream.

Source context: 91 downloads · 0 likes · Pipeline image-to-text · Library krill · Repo srv-sngh/Unlimited-OCR-mixed-nvfp4