初始化项目，由ModelHub XC社区提供模型

Model: xuanwulab/HaS_Text_0209_0.6B Source: Original Platform
2026-07-01 16:57:01 +08:00
commit 4dc78f3859
11 changed files with 152387 additions and 0 deletions
--- a/.gitattributes
+++ b/.gitattributes
@@ -0,0 +1,40 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tar filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text
+Figure1.png filter=lfs diff=lfs merge=lfs -text
+Table1.png filter=lfs diff=lfs merge=lfs -text
+Table2.png filter=lfs diff=lfs merge=lfs -text
+Table3.png filter=lfs diff=lfs merge=lfs -text
+tokenizer.json filter=lfs diff=lfs merge=lfs -text
--- a/README.md
+++ b/README.md
@@ -0,0 +1,560 @@
+---
+license: mit
+language:
+  - zh
+  - en
+  - pt
+  - fr
+  - es
+  - de
+  - ko
+  - ja
+tags:
+  - privacy
+  - anonymization
+  - ner
+  - pii
+  - hide-and-seek
+  - on-device
+  - qwen3
+base_model: Qwen/Qwen3-0.6B-Base
+pipeline_tag: text-generation
+---
+
+# HaS Text Model (Full Precision)
+
+**HaS (Hide and Seek)** is an on-device privacy model providing a complete pipeline from entity recognition to anonymization and restoration.
+
+- 📦 **0.6B parameters**, full precision (FP16), 1.2 GB
+- 🔒 **Data never leaves device** — local inference, no network required
+- 🌍 **8 languages natively supported**: Chinese, English, Portuguese, French, Spanish, German, Korean, Japanese
+- ⚡ **Apple M4 Pro benchmark**: prefill ~4,900 tok/s (llama.cpp), decode ~156 tok/s (mlx_lm)
+
+> This is the **full-precision base model** for fine-tuning and research. For production deployment, use the quantized versions: [Q8_0](https://huggingface.co/xuanwulab/HaS_Text_0209_0.6B_Q8) (recommended) or [Q4_K_M](https://huggingface.co/xuanwulab/HaS_Text_0209_0.6B_Q4).
+
+## 1. Core Capabilities
+
+Traditional anonymization (regex, Presidio, etc.) only does pattern matching. HaS is an **on-device Agentic privacy pipeline** — a set of composable atomic capabilities that solve multi-turn consistency, reversible restoration, and post-anonymization data usability.
+
+| Capability | Description |
+|------------|-------------|
+| **3-Level Semantic Tags** | Instead of `[REDACTED]`, produces tags like `<Amount[1].ContractAmount.NumberSymbol>` — LLMs understand "this is a contract amount", preserving data usability |
+| **Coreference Resolution** | "CloudGenius Inc.", "CloudGenius", "云创智能" → all unified as `<Organization[1].Company.Name>`. Different forms, same ID |
+| **Multi-turn Consistency** | Carries historical mapping dictionaries for incremental anonymization. Entity IDs stay consistent across turns. Same mechanism supports recursive chunking for long documents |
+| **Reversible Restoration** | Anonymized text can be processed by cloud LLMs (translation, rewriting, etc.), then Seek restores the tags back to original values |
+| **Open-set Entity Types** | Trained on ~70,000 entity types. Users can freely specify any type name without being limited to predefined categories |
+| **Public/Private Distinction** | "Industrial and Commercial Bank of China" preserved, "Li Hong 138-xxxx" anonymized — only redacts what should be redacted |
+
+## 2. Six Atomic Capabilities
+
+| # | Capability | Description |
+|---|------------|-------------|
+| 1 | **NER** | Recognize named entities of specified types |
+| 2 | **Hide_with** | Anonymize using an existing mapping dictionary (maintains cross-text consistency) |
+| 3 | **Hide_without** | First-time anonymization (no mapping, model generates tags autonomously) |
+| 4 | **Pair** | Extract mapping relationships from original and anonymized text pairs |
+| 5 | **Split** | Split composite tags into atomic single-entity mappings |
+| 6 | **Seek** | Restore tagged text using a mapping dictionary |
+
+## 3. Structured Semantic Tags & Coreference Resolution
+
+### 3-Level Semantic Tags
+
+Tags use a `<EntityType[ID].Category.Attribute>` three-level structure:
+
+```
+<Address[1].City.CityName>             ← identifies this as a city name
+<Address[2].StreetAddress.FullAddress>  ← identifies this as a detailed address
+<Amount[1].ContractAmount.NumberSymbol> ← identifies this as a contract amount
+<Phone[1].Mobile.FullNumber>           ← identifies this as a mobile number
+```
+
+Comparison with traditional approaches:
+
+| Traditional | HaS 3-Level Tag |
+|-------------|-----------------|
+| `[ADDRESS]` | `<Address[1].City.CityName>` |
+| `[ADDRESS]` | `<Address[2].StreetAddress.FullAddress>` |
+| `[MONEY]` | `<Amount[1].ContractAmount.NumberSymbol>` |
+
+### Coreference Resolution
+
+The same entity often appears in multiple forms. HaS automatically recognizes they refer to the same object and unifies them under one ID:
+
+```
+Original forms               Unified tag
+───────────────────           ───────────────────────
+CloudGenius Inc.          →   <Organization[1].Company.Name>
+CloudGenius               →   <Organization[1].Company.Name>
+云创智能                   →   <Organization[1].Company.Name>
+CG                        →   <Organization[1].Company.Name>
+```
+
+This ensures anonymized text remains logically coherent — LLMs seeing multiple `<Organization[1]>` know it's the same company. Critical for multi-turn conversations and long document chunking: entity IDs remain globally consistent across turns and chunks.
+
+## 4. Quick Start
+
+Recommended deployment with llama.cpp:
+
+```bash
+llama-server -m has_text_model.gguf -ngl 999 -c 8192 -np 1 -fa on
+```
+
+- Listens on `http://127.0.0.1:8080/v1` by default
+- OpenAI Chat Completions compatible API
+- ~**2.4 GB** total memory with default settings
+
+## 5. Usage Scenarios
+
+The 6 atomic capabilities can be composed into various privacy pipelines:
+
+| Scenario | Description | Capabilities Used |
+|----------|-------------|-------------------|
+| **Redacted Sharing** | Auto-anonymize files, emails, code before sending; retain mapping for restoration | Hide → Pair |
+| **Privacy Scanning** | Scan files/directories, list all sensitive entities, assess exposure risk | NER |
+| **Privacy Knowledge Base** | Anonymize documents before ingestion; restore query results via mapping | Hide → Pair (write), Seek (read) |
+| **Log Redaction** | Batch-anonymize ops logs before handing to support teams | Hide → Pair |
+| **Secure Cloud Chat** | Anonymize text before sending to cloud LLM; restore LLM responses | NER → Hide → Pair → Seek |
+| **AI Memory Privacy** | Store Agent long-term memory in anonymized form; restore on demand | Hide → Pair (store), Seek (recall) |
+
+## 6. Prompt Templates
+
+> ⚠️ **Templates must match character-for-character** — the model was trained on these exact templates. Any deviation may degrade output quality.
+
+### NER
+
+```
+Recognize the following entity types in the text.
+Specified types:{types_json_array}
+<text>{text}</text>
+```
+
+### Hide_with (with mapping)
+
+**Turn 1**: Same as NER template
+
+**Turn 2**:
+```
+Replace the above-mentioned entity types in the text according to the existing mapping pairs:{mapping_json}
+```
+
+### Hide_without (without mapping)
+
+**Turn 1**: Same as NER template
+
+**Turn 2** (fixed text, no variables):
+```
+Replace the above-mentioned entity types in the text.
+```
+
+### Pair
+
+```
+<original>{original_text}</original>
+<anonymized>{anonymized_text}</anonymized>
+Extract the mapping from anonymized entities to original entities.
+```
+
+### Split
+
+```
+Split each composite anonymized key into atomic keys.
+Composite mapping:
+{composite_mapping_json_array}
+```
+
+### Seek
+
+```
+The mapping from anonymized entities to original entities:
+{mapping_json}
+Restore the original text based on the above mapping:
+{text_with_tags}
+```
+
+## 7. Speed Benchmarks
+
+Test platform: Apple M4 Pro (48 GB RAM)
+
+| Metric | llama.cpp | mlx_lm | mlc_llm |
+|--------|:---------:|:------:|:-------:|
+| **FP16 model size** | 1.2 GB | 1.2 GB | 1.2 GB |
+| **FP16 prefill (tok/s)** | **4,904** | 4,272 | 1,818 |
+| **FP16 decode (tok/s)** | 128 | **156** | 118 |
+| **Q4 model size** | 0.4 GB | 0.4 GB | 0.4 GB |
+| **Q4 prefill (tok/s)** | **4,828** | 3,183 | 2,236 |
+| **Q4 decode (tok/s)** | 238 | **345** | 172 |
+
+> All performance figures are rounded. **Bold** indicates best in class.
+
+## 8. Quantization Versions
+
+| Version | Quantization | File Size | Runtime Memory | Notes |
+|---------|:---:|:---:|:---:|-------|
+| **Full Precision** | FP16 | 1.2 GB | ~2.4 GB | Base model for fine-tuning and research |
+| [Q8_0](https://huggingface.co/xuanwulab/HaS_Text_0209_0.6B_Q8) | 8.50 BPW | 639 MB | ~1.56 GB | **Recommended for production**, best output quality |
+| [Q4_K_M](https://huggingface.co/xuanwulab/HaS_Text_0209_0.6B_Q4) | 5.24 BPW | 397 MB | ~1.29 GB | Faster inference, lower memory, for resource-constrained environments |
+
+---
+
+<details>
+<summary><b>中文版</b></summary>
+
+# HaS Text Model（全量模型）
+
+**HaS（Hide and Seek）** 是一个端侧部署的隐私模型，提供从实体识别到脱敏还原的完整管线。
+
+- 📦 **0.6B 参数**，全精度（FP16），1.2 GB
+- 🔒 **数据不出设备**，本地推理，无需联网
+- 🌍 **8 语言原生支持**：中、英、葡、法、西、德、韩、日
+- ⚡ **Apple M4 Pro 实测**：prefill ~4,900 tok/s（llama.cpp），decode ~156 tok/s（mlx_lm）
+
+> 这是**全精度基座模型**，适用于微调和研究。生产部署请使用量化版本：[Q8_0](https://huggingface.co/xuanwulab/HaS_Text_0209_0.6B_Q8)（推荐）或 [Q4_K_M](https://huggingface.co/xuanwulab/HaS_Text_0209_0.6B_Q4)。
+
+## 一、核心能力
+
+传统脱敏方案（正则、Presidio 等）只做模式匹配。HaS 的定位是**端侧 Agentic 隐私管线**——用一组可组合的原子能力解决多轮一致、可逆还原和脱敏后数据可用性的问题。
+
+| 能力 | 说明 |
+|------|------|
+| **三级语义标签** | 脱敏后不是 `[REDACTED]`，而是 `<金额[1].合同金额.数字符号>` 这样携带语义的标签——LLM 一看就知道"这是一笔合同金额"，保持脱敏后数据可用性 |
+| **指代消解** | "云创智能有限公司"、"云创智能"、"CloudGenius"→ 全部归为 `<组织[1].企业.名称>`。不同写法，同一编号 |
+| **多轮一致** | 携带历史映射字典做增量脱敏，跨轮次实体编号一致。同一机制支持递归分块处理超长文档 |
+| **可逆还原** | 脱敏后的文本可先交给云端 LLM 处理（翻译、改写等），Seek 能对处理后文本中的标签进行还原 |
+| **开集指定** | 训练覆盖约 7 万种实体类型，用户可自由指定任意类型名称，不受预定义类别限制 |
+| **公私区分** | "中国工商银行"保留，"李红 138-xxxx"脱敏——只脱该脱的，不过度脱敏 |
+
+## 二、6 个原子能力
+
+| # | 能力 | 说明 |
+|---|------|------|
+| 1 | **NER** | 识别指定类型的命名实体 |
+| 2 | **Hide_with** | 使用已有映射字典脱敏（保持跨文本一致） |
+| 3 | **Hide_without** | 首次脱敏（无映射，模型自主生成标签） |
+| 4 | **Pair** | 从原文和脱敏文本对中提取映射关系 |
+| 5 | **Split** | 拆分复合标签为原子单实体映射 |
+| 6 | **Seek** | 根据映射字典还原含标签的文本 |
+
+## 三、结构化语义标签与指代消解
+
+### 三级语义标签
+
+脱敏后的标签采用 `<实体类型[编号].分类.属性>` 三级结构：
+
+```
+<地址[1].城市.市名>          ← 知道这是一个城市名
+<地址[2].街道门牌.完整地址>   ← 知道这是一个详细地址
+<金额[1].合同金额.数字符号>   ← 知道这是一笔合同金额，不只是普通数字
+<电话[1].手机号.完整号码>     ← 知道这是手机号，不是座机或传真
+```
+
+对比传统脱敏方案：
+
+| 传统方案 | HaS 三级标签 |
+|---------|-------------|
+| `[ADDRESS]` | `<地址[1].城市.市名>` |
+| `[ADDRESS]` | `<地址[2].街道门牌.完整地址>` |
+| `[MONEY]` | `<金额[1].合同金额.数字符号>` |
+
+### 指代消解
+
+同一实体在文本中往往以多种形式出现。HaS 会自动识别它们指向同一对象，统一归为同一编号：
+
+```
+原文中的写法              脱敏后统一为
+───────────────────       ───────────────────────
+云创智能科技有限公司   →   <组织[1].企业.名称>
+云创智能              →   <组织[1].企业.名称>
+CloudGenius           →   <组织[1].企业.名称>
+云创                  →   <组织[1].企业.名称>
+```
+
+这确保了脱敏后的文本逻辑自洽——LLM 看到多处 `<组织[1]>` 就知道是同一家公司，而不会误以为是不同实体。在多轮对话和长文档分块中尤为关键：跨轮次、跨分块的实体编号全局一致。
+
+## 四、快速开始
+
+推荐使用 llama.cpp 推理框架：
+
+```bash
+llama-server -m has_text_model.gguf -ngl 999 -c 8192 -np 1 -fa on
+```
+
+- 默认监听 `http://127.0.0.1:8080/v1`
+- API 兼容 OpenAI Chat Completions 格式
+- 默认配置下总内存约 **2.4 GB**
+
+## 五、使用场景
+
+6 个原子能力可以组合成多种隐私管线：
+
+| 场景 | 说明 | 使用能力 |
+|------|------|----------|
+| **脱敏分享** | 文件、邮件、代码在外发前自动脱敏，保留映射表可随时还原 | Hide → Pair |
+| **全量隐私扫描** | 扫描文件或目录，列出所有敏感实体，评估泄露风险 | NER |
+| **隐私知识库** | 文档先脱敏再入库，查询结果通过映射表还原原文 | Hide → Pair（写入）、Seek（读取） |
+| **日志脱敏** | 运维日志在交给支持团队前批量脱敏 | Hide → Pair |
+| **安全云端对话** | 脱敏后文本发给云端 LLM 处理，LLM 返回结果再还原 | NER → Hide → Pair → Seek |
+| **AI 记忆隐私** | Agent 的长期记忆以脱敏形式存储，使用时按需还原 | Hide → Pair（存储）、Seek（召回） |
+
+## 六、提示词模板
+
+> ⚠️ **模板必须逐字符精确匹配**，模型基于这些模板训练。任何偏差（空格、换行、标点）都可能降低输出质量。
+
+### NER
+
+```
+Recognize the following entity types in the text.
+Specified types:{types_json_array}
+<text>{text}</text>
+```
+
+### Hide_with（带映射脱敏）
+
+**第 1 轮**：与 NER 模板相同
+
+**第 2 轮**：
+
+```
+Replace the above-mentioned entity types in the text according to the existing mapping pairs:{mapping_json}
+```
+
+### Hide_without（无映射脱敏）
+
+**第 1 轮**：与 NER 模板相同
+
+**第 2 轮**（固定文本，无变量）：
+
+```
+Replace the above-mentioned entity types in the text.
+```
+
+### Pair（提取映射）
+
+```
+<original>{original_text}</original>
+<anonymized>{anonymized_text}</anonymized>
+Extract the mapping from anonymized entities to original entities.
+```
+
+### Split（拆分复合标签）
+
+```
+Split each composite anonymized key into atomic keys.
+Composite mapping:
+{composite_mapping_json_array}
+```
+
+### Seek（还原）
+
+```
+The mapping from anonymized entities to original entities:
+{mapping_json}
+Restore the original text based on the above mapping:
+{text_with_tags}
+```
+
+## 7. Speed Benchmarks
+
+Test platform: Apple M4 Pro (48 GB RAM)
+
+| Metric | llama.cpp | mlx_lm | mlc_llm |
+|--------|:---------:|:------:|:-------:|
+| **FP16 model size** | 1.2 GB | 1.2 GB | 1.2 GB |
+| **FP16 prefill (tok/s)** | **4,904** | 4,272 | 1,818 |
+| **FP16 decode (tok/s)** | 128 | **156** | 118 |
+| **Q4 model size** | 0.4 GB | 0.4 GB | 0.4 GB |
+| **Q4 prefill (tok/s)** | **4,828** | 3,183 | 2,236 |
+| **Q4 decode (tok/s)** | 238 | **345** | 172 |
+
+> All performance figures are rounded. **Bold** indicates best in class.
+
+## 8. Quantization Versions
+
+| Version | Quantization | File Size | Runtime Memory | Notes |
+|---------|:---:|:---:|:---:|-------|
+| **Full Precision** | FP16 | 1.2 GB | ~2.4 GB | Base model for fine-tuning and research |
+| [Q8_0](https://huggingface.co/xuanwulab/HaS_Text_0209_0.6B_Q8) | 8.50 BPW | 639 MB | ~1.56 GB | **Recommended for production**, best output quality |
+| [Q4_K_M](https://huggingface.co/xuanwulab/HaS_Text_0209_0.6B_Q4) | 5.24 BPW | 397 MB | ~1.29 GB | Faster inference, lower memory, for resource-constrained environments |
+
+---
+
+<details>
+<summary><b>中文版</b></summary>
+
+# HaS Text Model（全量模型）
+
+**HaS（Hide and Seek）** 是一个端侧部署的隐私模型，提供从实体识别到脱敏还原的完整管线。
+
+- 📦 **0.6B 参数**，全精度（FP16），1.2 GB
+- 🔒 **数据不出设备**，本地推理，无需联网
+- 🌍 **8 语言原生支持**：中、英、葡、法、西、德、韩、日
+- ⚡ **Apple M4 Pro 实测**：prefill ~4,900 tok/s（llama.cpp），decode ~156 tok/s（mlx_lm）
+
+> 这是**全精度基座模型**，适用于微调和研究。生产部署请使用量化版本：[Q8_0](https://huggingface.co/xuanwulab/HaS_Text_0209_0.6B_Q8)（推荐）或 [Q4_K_M](https://huggingface.co/xuanwulab/HaS_Text_0209_0.6B_Q4)。
+
+## 一、核心能力
+
+传统脱敏方案（正则、Presidio 等）只做模式匹配。HaS 的定位是**端侧 Agentic 隐私管线**——用一组可组合的原子能力解决多轮一致、可逆还原和脱敏后数据可用性的问题。
+
+| 能力 | 说明 |
+|------|------|
+| **三级语义标签** | 脱敏后不是 `[REDACTED]`，而是 `<金额[1].合同金额.数字符号>` 这样携带语义的标签——LLM 一看就知道"这是一笔合同金额"，保持脱敏后数据可用性 |
+| **指代消解** | "云创智能有限公司"、"云创智能"、"CloudGenius"→ 全部归为 `<组织[1].企业.名称>`。不同写法，同一编号 |
+| **多轮一致** | 携带历史映射字典做增量脱敏，跨轮次实体编号一致。同一机制支持递归分块处理超长文档 |
+| **可逆还原** | 脱敏后的文本可先交给云端 LLM 处理（翻译、改写等），Seek 能对处理后文本中的标签进行还原 |
+| **开集指定** | 训练覆盖约 7 万种实体类型，用户可自由指定任意类型名称，不受预定义类别限制 |
+| **公私区分** | "中国工商银行"保留，"李红 138-xxxx"脱敏——只脱该脱的，不过度脱敏 |
+
+## 二、6 个原子能力
+
+| # | 能力 | 说明 |
+|---|------|------|
+| 1 | **NER** | 识别指定类型的命名实体 |
+| 2 | **Hide_with** | 使用已有映射字典脱敏（保持跨文本一致） |
+| 3 | **Hide_without** | 首次脱敏（无映射，模型自主生成标签） |
+| 4 | **Pair** | 从原文和脱敏文本对中提取映射关系 |
+| 5 | **Split** | 拆分复合标签为原子单实体映射 |
+| 6 | **Seek** | 根据映射字典还原含标签的文本 |
+
+## 三、结构化语义标签与指代消解
+
+### 三级语义标签
+
+脱敏后的标签采用 `<实体类型[编号].分类.属性>` 三级结构：
+
+```
+<地址[1].城市.市名>          ← 知道这是一个城市名
+<地址[2].街道门牌.完整地址>   ← 知道这是一个详细地址
+<金额[1].合同金额.数字符号>   ← 知道这是一笔合同金额，不只是普通数字
+<电话[1].手机号.完整号码>     ← 知道这是手机号，不是座机或传真
+```
+
+对比传统脱敏方案：
+
+| 传统方案 | HaS 三级标签 |
+|---------|-------------|
+| `[ADDRESS]` | `<地址[1].城市.市名>` |
+| `[ADDRESS]` | `<地址[2].街道门牌.完整地址>` |
+| `[MONEY]` | `<金额[1].合同金额.数字符号>` |
+
+### 指代消解
+
+同一实体在文本中往往以多种形式出现。HaS 会自动识别它们指向同一对象，统一归为同一编号：
+
+```
+原文中的写法              脱敏后统一为
+───────────────────       ───────────────────────
+云创智能科技有限公司   →   <组织[1].企业.名称>
+云创智能              →   <组织[1].企业.名称>
+CloudGenius           →   <组织[1].企业.名称>
+云创                  →   <组织[1].企业.名称>
+```
+
+这确保了脱敏后的文本逻辑自洽——LLM 看到多处 `<组织[1]>` 就知道是同一家公司，而不会误以为是不同实体。
+
+## 四、快速开始
+
+推荐使用 llama.cpp 推理框架：
+
+```bash
+llama-server -m has_text_model.gguf -ngl 999 -c 8192 -np 1 -fa on
+```
+
+- 默认监听 `http://127.0.0.1:8080/v1`
+- API 兼容 OpenAI Chat Completions 格式
+- 默认配置下总内存约 **2.4 GB**
+
+## 五、使用场景
+
+6 个原子能力可以组合成多种隐私管线：
+
+| 场景 | 说明 | 使用能力 |
+|------|------|----------|
+| **脱敏分享** | 文件、邮件、代码在外发前自动脱敏，保留映射表可随时还原 | Hide → Pair |
+| **全量隐私扫描** | 扫描文件或目录，列出所有敏感实体，评估泄露风险 | NER |
+| **隐私知识库** | 文档先脱敏再入库，查询结果通过映射表还原原文 | Hide → Pair（写入）、Seek（读取） |
+| **日志脱敏** | 运维日志在交给支持团队前批量脱敏 | Hide → Pair |
+| **安全云端对话** | 脱敏后文本发给云端 LLM 处理，LLM 返回结果再还原 | NER → Hide → Pair → Seek |
+| **AI 记忆隐私** | Agent 的长期记忆以脱敏形式存储，使用时按需还原 | Hide → Pair（存储）、Seek（召回） |
+
+## 六、提示词模板
+
+> ⚠️ **模板必须逐字符精确匹配**，模型基于这些模板训练。任何偏差（空格、换行、标点）都可能降低输出质量。
+
+### NER
+
+```
+Recognize the following entity types in the text.
+Specified types:{types_json_array}
+<text>{text}</text>
+```
+
+### Hide_with（带映射脱敏）
+
+**第 1 轮**：与 NER 模板相同
+
+**第 2 轮**：
+
+```
+Replace the above-mentioned entity types in the text according to the existing mapping pairs:{mapping_json}
+```
+
+### Hide_without（无映射脱敏）
+
+**第 1 轮**：与 NER 模板相同
+
+**第 2 轮**（固定文本，无变量）：
+
+```
+Replace the above-mentioned entity types in the text.
+```
+
+### Pair（提取映射）
+
+```
+<original>{original_text}</original>
+<anonymized>{anonymized_text}</anonymized>
+Extract the mapping from anonymized entities to original entities.
+```
+
+### Split（拆分复合标签）
+
+```
+Split each composite anonymized key into atomic keys.
+Composite mapping:
+{composite_mapping_json_array}
+```
+
+### Seek（还原）
+
+```
+The mapping from anonymized entities to original entities:
+{mapping_json}
+Restore the original text based on the above mapping:
+{text_with_tags}
+```
+
+## 七、速度评估
+
+测试平台：Apple M4 Pro（48 GB 内存）
+
+| 指标 | llama.cpp | mlx_lm | mlc_llm |
+|------|:---------:|:------:|:-------:|
+| **FP16 模型大小** | 1.2 GB | 1.2 GB | 1.2 GB |
+| **FP16 prefill（tok/s）** | **4,904** | 4,272 | 1,818 |
+| **FP16 decode（tok/s）** | 128 | **156** | 118 |
+| **Q4 模型大小** | 0.4 GB | 0.4 GB | 0.4 GB |
+| **Q4 prefill（tok/s）** | **4,828** | 3,183 | 2,236 |
+| **Q4 decode（tok/s）** | 238 | **345** | 172 |
+
+> 所有性能数据均已取整。**粗体**表示该项最佳表现。
+
+## 八、量化版本
+
+| 版本 | 量化 | 文件大小 | 运行内存 | 说明 |
+|------|:---:|:---:|:---:|------|
+| **全量模型** | FP16 | 1.2 GB | ~2.4 GB | 基座模型，适用于微调和研究 |
+| [Q8_0](https://huggingface.co/xuanwulab/HaS_Text_0209_0.6B_Q8) | 8.50 BPW | 639 MB | ~1.56 GB | **推荐生产使用**，输出质量最佳 |
+| [Q4_K_M](https://huggingface.co/xuanwulab/HaS_Text_0209_0.6B_Q4) | 5.24 BPW | 397 MB | ~1.29 GB | 推理更快，内存更省，适合资源受限场景 |
+
+</details>
--- a/chat_template.jinja
+++ b/chat_template.jinja
@@ -0,0 +1,85 @@
+{%- if tools %}
+    {{- '<|im_start|>system\n' }}
+    {%- if messages[0].role == 'system' %}
+        {{- messages[0].content + '\n\n' }}
+    {%- endif %}
+    {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
+    {%- for tool in tools %}
+        {{- "\n" }}
+        {{- tool | tojson }}
+    {%- endfor %}
+    {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
+{%- else %}
+    {%- if messages[0].role == 'system' %}
+        {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
+    {%- endif %}
+{%- endif %}
+{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
+{%- for message in messages[::-1] %}
+    {%- set index = (messages|length - 1) - loop.index0 %}
+    {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
+        {%- set ns.multi_step_tool = false %}
+        {%- set ns.last_query_index = index %}
+    {%- endif %}
+{%- endfor %}
+{%- for message in messages %}
+    {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
+        {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
+    {%- elif message.role == "assistant" %}
+        {%- set content = message.content %}
+        {%- set reasoning_content = '' %}
+        {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
+            {%- set reasoning_content = message.reasoning_content %}
+        {%- else %}
+            {%- if '</think>' in message.content %}
+                {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
+                {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
+            {%- endif %}
+        {%- endif %}
+        {%- if loop.index0 > ns.last_query_index %}
+            {%- if loop.last or (not loop.last and reasoning_content) %}
+                {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
+            {%- else %}
+                {{- '<|im_start|>' + message.role + '\n' + content }}
+            {%- endif %}
+        {%- else %}
+            {{- '<|im_start|>' + message.role + '\n' + content }}
+        {%- endif %}
+        {%- if message.tool_calls %}
+            {%- for tool_call in message.tool_calls %}
+                {%- if (loop.first and content) or (not loop.first) %}
+                    {{- '\n' }}
+                {%- endif %}
+                {%- if tool_call.function %}
+                    {%- set tool_call = tool_call.function %}
+                {%- endif %}
+                {{- '<tool_call>\n{"name": "' }}
+                {{- tool_call.name }}
+                {{- '", "arguments": ' }}
+                {%- if tool_call.arguments is string %}
+                    {{- tool_call.arguments }}
+                {%- else %}
+                    {{- tool_call.arguments | tojson }}
+                {%- endif %}
+                {{- '}\n</tool_call>' }}
+            {%- endfor %}
+        {%- endif %}
+        {{- '<|im_end|>\n' }}
+    {%- elif message.role == "tool" %}
+        {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
+            {{- '<|im_start|>user' }}
+        {%- endif %}
+        {{- '\n<tool_response>\n' }}
+        {{- message.content }}
+        {{- '\n</tool_response>' }}
+        {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
+            {{- '<|im_end|>\n' }}
+        {%- endif %}
+    {%- endif %}
+{%- endfor %}
+{%- if add_generation_prompt %}
+    {{- '<|im_start|>assistant\n' }}
+    {%- if enable_thinking is defined and enable_thinking is false %}
+        {{- '<think>\n\n</think>\n\n' }}
+    {%- endif %}
+{%- endif %}
--- a/config.json
+++ b/config.json
@@ -0,0 +1,30 @@
+{
+  "architectures": [
+    "Qwen3ForCausalLM"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "head_dim": 128,
+  "hidden_act": "silu",
+  "hidden_size": 1024,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "max_position_embeddings": 32768,
+  "max_window_layers": 28,
+  "model_type": "qwen3",
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "num_key_value_heads": 8,
+  "rms_norm_eps": 1e-06,
+  "rope_scaling": null,
+  "rope_theta": 1000000,
+  "sliding_window": null,
+  "tie_word_embeddings": true,
+  "torch_dtype": "bfloat16",
+  "transformers_version": "4.52.3",
+  "use_cache": false,
+  "use_sliding_window": false,
+  "vocab_size": 151936
+}
--- a/generation_config.json
+++ b/generation_config.json
@@ -0,0 +1,6 @@
+{
+  "bos_token_id": 151643,
+  "eos_token_id": 151643,
+  "max_new_tokens": 2048,
+  "transformers_version": "4.52.3"
+}
--- a/merges.txt
+++ b/merges.txt
--- a/model.safetensors
+++ b/model.safetensors
@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c507df88475199370d85d89e37d098e6dd7395aa5e16e8dd7cf57120cc54bb06
+size 1192135096
--- a/special_tokens_map.json
+++ b/special_tokens_map.json
@@ -0,0 +1,31 @@
+{
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "eos_token": {
+    "content": "<|im_end|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}
--- a/tokenizer.json
+++ b/tokenizer.json
--- a/tokenizer_config.json
+++ b/tokenizer_config.json
@@ -0,0 +1,240 @@
+{
+  "add_bos_token": false,
+  "add_prefix_space": false,
+  "added_tokens_decoder": {
+    "151643": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151644": {
+      "content": "<|im_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151645": {
+      "content": "<|im_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151646": {
+      "content": "<|object_ref_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151647": {
+      "content": "<|object_ref_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151648": {
+      "content": "<|box_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151649": {
+      "content": "<|box_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151650": {
+      "content": "<|quad_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151651": {
+      "content": "<|quad_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151652": {
+      "content": "<|vision_start|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151653": {
+      "content": "<|vision_end|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151654": {
+      "content": "<|vision_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151655": {
+      "content": "<|image_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151656": {
+      "content": "<|video_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "151657": {
+      "content": "<tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151658": {
+      "content": "</tool_call>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151659": {
+      "content": "<|fim_prefix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151660": {
+      "content": "<|fim_middle|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151661": {
+      "content": "<|fim_suffix|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151662": {
+      "content": "<|fim_pad|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151663": {
+      "content": "<|repo_name|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151664": {
+      "content": "<|file_sep|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151665": {
+      "content": "<tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151666": {
+      "content": "</tool_response>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151667": {
+      "content": "<think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    },
+    "151668": {
+      "content": "</think>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": false
+    }
+  },
+  "additional_special_tokens": [
+    "<|im_start|>",
+    "<|im_end|>",
+    "<|object_ref_start|>",
+    "<|object_ref_end|>",
+    "<|box_start|>",
+    "<|box_end|>",
+    "<|quad_start|>",
+    "<|quad_end|>",
+    "<|vision_start|>",
+    "<|vision_end|>",
+    "<|vision_pad|>",
+    "<|image_pad|>",
+    "<|video_pad|>"
+  ],
+  "bos_token": null,
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<|im_end|>",
+  "errors": "replace",
+  "extra_special_tokens": {},
+  "model_max_length": 131072,
+  "pad_token": "<|endoftext|>",
+  "padding_side": "right",
+  "split_special_tokens": false,
+  "tokenizer_class": "Qwen2Tokenizer",
+  "unk_token": null
+}
--- a/vocab.json
+++ b/vocab.json