Qwen3.5-4B-Condenser实战案例：如何处理长文本并保持关键信息不丢失

【免费下载链接】Qwen3.5-4B-Condenser 项目地址: https://ai.gitcode.com/hf_mirrors/twinkle-kit/Qwen3.5-4B-Condenser

Qwen3.5-4B-Condenser是一款基于Qwen3.5-4B模型开发的文本压缩工具，它能够将长文本高效压缩为结构化摘要，同时确保关键信息不丢失。这一工具特别适用于需要处理大量文本数据的场景，帮助用户快速获取核心内容。

为什么选择Qwen3.5-4B-Condenser进行长文本处理

在信息爆炸的时代，我们每天都会面对海量的长文本内容。传统的文本摘要方法往往会丢失重要信息，而Qwen3.5-4B-Condenser采用了先进的LoRA（Low-Rank Adaptation）技术，通过对Qwen3.5-4B模型进行微调，实现了在压缩文本的同时保留关键信息的目标。

该工具的核心优势在于：

• 高效压缩：能够将文本压缩至原始长度的35%-55%，大大节省阅读时间
• 结构化输出：采用"## Summary"和"## More"两部分结构，使摘要更加清晰易读
• 关键信息保留：通过特殊的训练方法，确保重要事实、数据和实体不被遗漏
• 查询感知排序：能够根据用户查询对信息进行排序，优先展示相关内容

Qwen3.5-4B-Condenser的工作原理

Qwen3.5-4B-Condenser的工作流程基于特定的输入输出契约。系统提示定义了压缩规则，用户提示包含查询、目标长度和待压缩文本，模型输出则是结构化的压缩结果。

输入输出契约

系统提示明确了压缩的目标和规则，要求压缩结果必须保留源文本的主要信息。用户提示则包含三个关键部分：

• 查询：作为排序提示，而非过滤条件
• 目标长度：硬字符限制，计算公式为max(160, int(len(title+body) * 0.5))
• 文本：包含标题和正文的待压缩内容

模型输出格式固定为两部分：

## Summary
<主题 + 2-4个具体核心事实，与查询相关的优先>

## More
<逗号分隔的类别关键词，需要展开才能查看具体值>

压缩规则

Qwen3.5-4B-Condenser遵循以下关键压缩规则：

1. 电报式风格：删除功能词（"the"、"a"、"is"等），冒号和逗号表示"是"或"有"
2. 摘要必须包含文本的主要主题和2-4个具体核心事实
3. 摘要不得包含关于查询的元评论
4. "More"部分是类别关键词的索引，而非内联数据
5. 输出语言必须与源语言匹配
6. 不得编造信息，不得遗漏主要信息

实战案例：使用Qwen3.5-4B-Condenser处理长文本

准备工作

首先，需要克隆仓库并安装必要的依赖：

git clone https://gitcode.com/hf_mirrors/twinkle-kit/Qwen3.5-4B-Condenser
cd Qwen3.5-4B-Condenser
# 安装依赖的命令（根据项目实际情况补充）

基本使用方法

以下是使用Qwen3.5-4B-Condenser的基本Python代码示例：

from openai import OpenAI

client = OpenAI(base_url='http://localhost:8000/v1', api_key='EMPTY')

SYSTEM = '''You are a text compression assistant. A downstream model will read your compressed output to decide whether the detail it needs is inside this block; if yes, it will fetch and read the original passage.

Downstream model workflow:
Read your compressed output -> Decide whether needed info is in this block -> If yes -> Fetch original.

Therefore your compression MUST NOT lose major information from the source.

Output format:

```text
## Summary
Overview plus facts STRONGLY RELATED to the Query, stated explicitly.

## More
A collapsed index; expansion required to see specific information.

Rules:

1. Telegraphic style — drop function words ("the", "a", "is", "are", "of", ...); colons and commas mean "is" / "has".
2. Summary MUST contain the passage's primary topic + 2–4 concrete core facts drawn from the source (entities, numbers, dates, relations). If a Query is given, order Query-relevant facts first, but STILL include other core facts within the budget. A Query is an ORDERING HINT, NOT a filter.
3. Summary MUST NOT be meta-commentary about the Query. Forbidden patterns: "no X mention", "Query info: absent", "passage covers Y only", "does not contain ...", "no relevant info", or summaries that are only abstract category words like "structure/order/usage" with no facts. If the passage is unrelated to the Query, you still summarize the passage normally.
4. More is an INDEX of category keywords, NOT inline data. Enumerate what CAN be recovered from the source (e.g. "birthplace, death place, age"); do NOT paste dates/numbers/names inline. Make sure all category of useful facts are introduced here.
5. Output language MUST match the source language.
6. Do NOT fabricate. Do NOT omit major information. Any fact not in the source MUST NOT appear in your output.

Example: [示例内容省略]

Now begin.'''

USER_TEMPLATE = '''Downstream model will read your compressed block to decide whether to expand it. Compress faithfully: preserve the passage topic + core facts. Do NOT invent facts. Do NOT drop major facts. Do NOT write meta-commentary about the Query (never write "Query info: absent", "no X mention", etc.); if the passage does not address the Query, still summarize the passage.

Query (ordering hint only — still summarize the whole passage)

{query}

Target length

Compress AS MUCH AS faithfully possible. HARD CEILING: {budget} chars (~50% of the source). If core facts fit in far fewer chars, output fewer. Never exceed the ceiling.

Passage

{text}'''

query = 'Which city was Marie Curie born in?' title = 'Marie Curie' body = 'Marie Curie (7 Nov 1867 – 4 Jul 1934), born Maria Sklodowska in Warsaw (then Russian Poland); parents were teachers. Barred from Polish universities, she and her sister agreed to take turns funding each other's overseas study. In 1891 Marie reached Paris and enrolled at the Sorbonne, earning a physics degree (1893) and a mathematics degree (1894), becoming the school's first female physics lecturer. In 1895 she married French physicist Pierre Curie; they spent the rest of their lives on radioactivity research. In July 1898 she discovered polonium, named after her homeland Poland; in December she and Pierre announced the discovery of radium. She coined "radioactivity" and showed it is an atomic property, not a chemical reaction. In 1903 she shared the Nobel Prize in Physics with Pierre and Henri Becquerel. In 1911 she alone won the Nobel Prize in Chemistry for polonium and radium. She is the first woman to win a Nobel, and the only person to win Nobels in two different sciences. After Pierre died in a carriage accident in 1906, Marie took his chair and became the first female professor at the Sorbonne. During World War I she developed mobile X-ray units, called "Petites Curies" in French; about 20 were deployed to the front, examining over 1,000,000 wounded soldiers. She died of aplastic anaemia from radiation exposure on 4 July 1934 in Passy, Haute-Savoie, France, aged 66. Her notebooks remain highly radioactive, kept in lead boxes; researchers must wear protective gear to consult them.' text = f'{title}: {body}' budget = max(160, int(len(text) * 0.5))

resp = client.chat.completions.create( model='qwen3.5-4b-condenser', messages=[ {'role': 'system', 'content': SYSTEM}, {'role': 'user', 'content': USER_TEMPLATE.format( query=query, budget=budget, text=text)}, ], temperature=0.3, max_tokens=max(128, int(budget * 0.6) + 16), ) print(resp.choices[0].message.content)

### 预期输出结果

上述代码的预期输出如下：

```text
## Summary
Marie Curie: French-Polish physicist/chemist, founder of radioactivity research, first female Sorbonne professor.
- Nobel x2 (Physics + Chemistry); first woman Nobel laureate; only person with Nobels in two sciences.
- Discovered polonium + radium; coined "radioactivity"; proved it is an atomic property.

## More
birthplace, death place, age, cause of death, degree years, in-school firsts x2, element naming origin, collaborators, full timeline, Nobel year per prize, co-laureates, citation, device name, deployment scale, patients treated, notebook radioactivity, storage, access conditions

可以看到，输出结果清晰地分为"Summary"和"More"两部分，既保留了核心信息，又大大缩短了文本长度。

Qwen3.5-4B-Condenser的高级应用场景

1. 作为Twinkle Agentic RL管道的压缩后端

Qwen3.5-4B-Condenser可以作为Twinkle Agentic RL管道的压缩后端，通过cookbook/rl/grpo_condensed.py实现更复杂的文本处理流程。

2. 用于后续GRPO/DPO的SFT预热

该模型可以作为后续GRPO（Generalized Relative Policy Optimization）或DPO（Direct Preference Optimization）的SFT（Supervised Fine-Tuning）预热，针对相同的压缩器模式进行进一步优化。

3. 离线数据集预处理

Qwen3.5-4B-Condenser非常适合用于离线数据集预处理，通过预压缩长文本语料，可以在检索时控制上下文预算，提高处理效率。

模型训练参数与配置

Qwen3.5-4B-Condenser基于Qwen3.5-4B模型，使用LoRA技术进行微调。关键训练参数如下：

• 基础模型：Qwen/Qwen3.5-4B
• 数据集：twinkle-kit/hotpotqa-condensed-9k
• LoRA参数：
  • Rank (r)：16
  • lora_alpha：32
  • target_modules：all-linear
• 优化器：AdamW
• 学习率：1e-4
• 批大小：8（每DP rank）
• 梯度累积：4
• 学习率调度器：CosineWarmupScheduler
• 训练轮次：5

详细配置可参考adapter_config.json文件。

已知限制与解决方案

尽管Qwen3.5-4B-Condenser表现出色，但仍有一些已知限制：

1. 短文本压缩效率问题

对于长度低于250字符的文本，压缩率可能会超过0.7，压缩效果不佳。这是因为短文本本身可压缩空间有限。

解决方案：对于短文本，可以考虑不进行压缩，或调整压缩参数以适应短文本特点。

2. 实体属性覆盖问题

低频属性（如"英裔美国人"这样的国籍描述）可能会被完全删除，而不是出现在"## More"部分。

解决方案：增加训练数据量或添加针对性提示可以改善这一问题。

3. 语言限制

目前Qwen3.5-4B-Condenser仅支持英语，多语言处理能力未经测试。

解决方案：使用多语言数据集进行微调，扩展模型的语言支持能力。

4. 非独立QA模型

Qwen3.5-4B-Condenser仅重写文本，下游问答任务需要由其他模型完成。

解决方案：将Qwen3.5-4B-Condenser与专门的QA模型结合使用，构建完整的问答系统。

总结

Qwen3.5-4B-Condenser是一款强大的长文本处理工具，它通过先进的压缩算法和结构化输出，在大幅减少文本长度的同时，确保关键信息不丢失。无论是用于快速阅读、数据预处理还是作为AI系统的一部分，Qwen3.5-4B-Condenser都能提供高效、可靠的文本压缩服务。

通过本文介绍的实战案例，您可以快速上手使用Qwen3.5-4B-Condenser，并根据自己的需求进行调整和优化。随着NLP技术的不断发展，我们有理由相信Qwen3.5-4B-Condenser将在长文本处理领域发挥越来越重要的作用。

【免费下载链接】Qwen3.5-4B-Condenser 项目地址: https://ai.gitcode.com/hf_mirrors/twinkle-kit/Qwen3.5-4B-Condenser