---
title: PP-OCRv6-tiny-rec-GGUF
canonical_url: "https://www.modelscope.cn/models/mlx-community/PP-OCRv6-tiny-rec-GGUF"
md_url: "https://www.modelscope.cn/models/mlx-community/PP-OCRv6-tiny-rec-GGUF.md"
repository: mlx-community/PP-OCRv6-tiny-rec-GGUF
last_updated: 2026-07-31
license: apache-2.0
library_name:
  - gguf
  - pytorch
frameworks:
  - pytorch
downloads: 23
stars: 0
tags:
  - ocr
  - pp-ocrv6
  - gguf
  - crispembed
---

# PP-OCRv6-tiny-rec-GGUF

> PP-OCRv6-tiny-rec-GGUF - mlx-community 在 ModelScope 开源的模型。PP-OCRv6 tiny recognizer — CrispEmbed GGUF

mlx-community/PP-OCRv6-tiny-rec-GGUF 是 ModelScope 魔搭社区上的机器学习模型，采用 apache-2.0 许可。

- **Repository**: mlx-community/PP-OCRv6-tiny-rec-GGUF
- **License**: apache-2.0
- **Tags**: ocr, pp-ocrv6, gguf, crispembed
- **Downloads**: 23
- **Stars**: 0
- **Last updated**: 2026-07-31

Source: https://www.modelscope.cn/models/mlx-community/PP-OCRv6-tiny-rec-GGUF

---

# PP-OCRv6 tiny recognizer — CrispEmbed GGUF

Files: `PP-OCRv6_tiny_rec-f16.gguf` and `PP-OCRv6_tiny_rec-crispasr-q4_k-policy.gguf`.

The policy-q4 container intentionally keeps the complete PP-OCRv6 detector/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` using the CrispEmbed diff harness: input 0.999999, stage4 0.999982, head_input 0.999988, logits 0.999992 (F16 and policy-q4; all reported stages pass the 0.999 threshold).