add deepseek-ocr2 support

This commit is contained in:
jhqxxx
2026-03-18 19:15:21 +08:00
parent ca57a7245c
commit 6890f86030
22 changed files with 708 additions and 89 deletions
+3
View File
@@ -5,6 +5,9 @@ All notable changes to aha will be documented in this file.
The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
### 2026-03-18
- add DeepSeek-OCR-2
### 2026-03-17
- add PaddleOCR-VL1.5 model
- fix qwen3.5 position_ids create bug
+3
View File
@@ -5,6 +5,9 @@
格式基于 [Keep a Changelog](https://keepachangelog.com/zh-CN/1.0.0/)
本项目遵循 [语义化版本](https://semver.org/lang/zh-CN/spec/v2.0.0.html)。
### 2026-03-18
- 新增 DeepSeek-OCR-2
### 2026-03-17
- 新增 PaddleOCR-VL1.5 模型
- 修复 qwen3.5 position_ids 创建错误
+13
View File
@@ -7,6 +7,10 @@
"OCR:" | "Table Recognition:" | "Formula Recognition:" | "Chart Recognition:" | "Spotting:" | "Seal Recognition:"
## DeepSeekOCR
#### Main Prompts
1. document: <image>\n<|grounding|>Convert the document to markdown.
2. without layouts: <image>\nFree OCR.
#### Metadata
* base_size: 512 | 640 | 1024 | 1280
* image_size: 512 | 640 | 1024 | 1280
@@ -25,6 +29,15 @@
| Large | 1280 | 1280 | false |
| Gundam | 1024 | 640 | true |
## DeepSeekOCR2
#### Main Prompts
1. document: <image>\n<|grounding|>Convert the document to markdown.
2. without layouts: <image>\nFree OCR.
#### Metadata
* crop_mode: false | true
## GLM-OCR
#### Prompt Limited
##### Document Parsing prompt
+12
View File
@@ -7,6 +7,10 @@
"OCR:" | "Table Recognition:" | "Formula Recognition:" | "Chart Recognition:" | "Spotting:" | "Seal Recognition:"
## DeepSeekOCR
#### 主要提示词
1. 文档: <image>\n<|grounding|>Convert the document to markdown.
2. 无布局信息: <image>\nFree OCR.
#### Metadata
* base_size: 512 | 640 | 1024 | 1280
* image_size: 512 | 640 | 1024 | 1280
@@ -25,6 +29,14 @@
| Large | 1280 | 1280 | false |
| Gundam | 1024 | 640 | true |
## DeepSeekOCR2
#### 主要提示词
1. 文档: <image>\n<|grounding|>Convert the document to markdown.
2. 无布局信息: <image>\nFree OCR.
#### Metadata
* crop_mode: false | true
## GLM-OCR
#### 限定提示词
##### 文档解析: