GraySoft
Projects Models Compare Cloud benchmarks FAQ Download guIDE →
Model Intelligence Sheet

mlx-community/PP-OCRv6-small-rec-GGUF overview

PP OCRv6 small recognizer — CrispEmbed GGUF Files: PP OCRv6 small rec f16.gguf and PP OCRv6 small rec crispasr q4 k policy.gguf . The policy q4 container inten…

crispembedggufocrpp-ocrv6license:apache-2.0region:us

Runs locally from ~9.5 MB disk (4 GB VRAM class GPUs with llama.cpp / guIDE).

Downloads
0
Likes
0
Pipeline

Repository Files & Downloads

3 GGUF files detected
Direct downloads for local inference
FileTypeQuantizationSizeLink
PP-OCRv6_small_rec-crispasr-q4_k-policy.ggufGGUFQ4_K10.5 MBDownload
PP-OCRv6_small_rec-f16.ggufGGUFF1610.5 MBDownload
PP-OCRv6_small_rec-fox-ref.ggufGGUFGGUF9.5 MBDownload

Model Details

Model IDmlx-community/PP-OCRv6-small-rec-GGUF
Authormlx-community
Pipeline
Licenseapache-2.0
Base model
Last modified2026-07-31T14:44:54.000Z

Model README

---

library_name: crispembed

license: apache-2.0

tags: [ocr, pp-ocrv6, gguf, crispembed]

---

PP-OCRv6 small recognizer — CrispEmbed GGUF

Files: PP-OCRv6_small_rec-f16.gguf and PP-OCRv6_small_rec-crispasr-q4_k-policy.gguf.

The policy-q4 container intentionally keeps the complete PP-OCRv6 recognizer graph in F16: quantizing intermediate CNN/SVTR weights caused compounding CTC drift. Source: PaddlePaddle PP-OCRv6, Apache-2.0.

Parity on tests/regression/images/fox.png: F16 and policy-q4 both pass the 0.999 cosine floor across the full native diff path; worst stage cosine is 0.999265 and logits cosine is 0.999530.

Run mlx-community/PP-OCRv6-small-rec-GGUF with guIDE

Download guIDE — the AI-native code editor with local LLM inference and 69 built-in tools.

Download guIDE → · Browse 524k+ models · Compare models

Source: Hugging Face · Compare models