Previously, SiliconCloud went online with a language modelingIn-line LoRA fine-tuning function. By simply uploading corpus data and creating a fine-tuning task, you can obtain a proprietary fine-tuned language model.
Recently, SiliconCloud's LLM online LoRA fine-tuning has been extended with Qwen2.5-32B, Qwen2.5-14B, and Llama-3.1-8B models as fine-tuning base models, which further enriches the playability of the fine-tuning, and further reduces the cost of training and using the fine-tuning models.
In fact, fine-tuning a proprietary big language model is quite simple. Using SiliconCloud's fine-tuning feature, you can prepare just a few dozen training corpora to make "significant" changes to your model from the base model.
preliminary
First, we need to be clear about what fine-tuning is trying to accomplish. Often, fine-tuning can help a model learn knowledge not included in the training, or develop a certain distinctive linguistic style.
Our approach will vary for different purposes. For example, if you want the model to learn certain domain knowledge, you need the training data to contain questions and answers based on these data, while if you want the model to learn a certain typical language style, you need dialog data with this style.
Data preparation
Fine-tuning of the LLM model requires that the data be provided in the form of a dialog with a data structure that is essentially the same as the invocation message of the LLM. It can be said that the text generation interface will be invoked to construct the fine-tuned data.
Below is an example of a typical piece of fine-tuned data:
{"messages": [{"role": "system", "content": "你是消防系统专家"}, {"role": "user", "content": "家里的火灾报警器怎么安装？"}, {"role": "assistant", "content": "根据《家用火灾安全系统》（GB22370—2008）的规定，家用火灾报警器的安装应符合制造商的要求。具体来说，安装时应确保设备处于正常监视状态，并且连接到相应的火灾探测器和其他部件。安装位置应避免高温、潮湿和尘土较多的地方，以保证设备的正常运行。"}]}
To perform a fine-tuning, it is usually necessary to prepare dozens of such data and save them as a "jsonl" file in the form of one per line:
{"messages": [{"role": "system", "content": "你是消防系统专家"}, {"role": "user", "content": "家里的火灾报警器怎么安装？"}, {"role": "assistant", "content": "根据《家用火灾安全系统》（GB22370—2008）的规定，家用火灾报警器的安装应符合制造商的要求。具体来说，安装时应确保设备处于正常监视状态，并且连接到相应的火灾探测器和其他部件。安装位置应避免高温、潮湿和尘土较多的地方，以保证设备的正常运行。"}]}
{"messages": [{"role": "system", "content": "你是消防系统专家"}, {"role": "user", "content": "火灾报警器的电源要求是什么？"}, {"role": "assistant", "content": "根据《家用火灾安全系统》（GB22370—2008）的规定，家用火灾报警控制器的主电源应采用220V、50Hz交流电源，电源线输入端应设接线端子。备用电源应为可多次充放电的蓄电池，在不超过生产厂商规定的极限放电情况下，应能将电池在24小时内充至额定容量80%以上，再充48小时后应能充满。"}]}
{"messages": [{"role": "system", "content": "你是消防系统专家"}, {"role": "user", "content": "火灾报警器的声光信号有什么要求？"}, {"role": "assistant", "content": "根据《家用火灾安全系统》（GB22370—2008）的规定，火灾声警报器的声信号至少在一个方向上3米处的声压级应不小于75dB(A计权)，且在任意方向上3米处的声压级应不大于120dB(A计权)。火灾光警报器的光信号在100lx~500lx环境光线下，25米处应清晰可见。"}]}
{"messages": [{"role": "system", "content": "你是消防系统专家"}, {"role": "user", "content": "火灾报警器的故障信号怎么处理？"}, {"role": "assistant", "content": "根据《家用火灾安全系统》（GB22370—2008）的规定，当发生与所连接部件之间的通讯故障、备用电源充电器与备用电源间连接线断路或短路、备用电源与其负载间连接线断路或短路、主电源欠压等故障时，家用火灾报警控制器应在100秒内发出故障声、光信号，并指示故障类型。故障声信号应能手动消除，再有故障信号输入时，应能再次启动。"}]}
{"messages": [{"role": "system", "content": "你是消防系统专家"}, {"role": "user", "content": "火灾报警器的自检功能怎么用？"}, {"role": "assistant", "content": "根据《家用火灾安全系统》（GB22370—2008）的规定，家用火灾报警控制器应能对本机进行功能检查（自检）。在执行自检功能期间，受其控制的外接设备和输出接点均不应动作。自检时间超过1分钟或不能自动停止自检功能时，其非自检部位和控制器本身的火灾报警功能不应受到影响。"}]}
{"messages": [{"role": "system", "content": "你是消防系统专家"}, {"role": "user", "content": "火灾报警器的信息显示有什么要求？"}, {"role": "assistant", "content": "根据《家用火灾安全系统》（GB22370—2008）的规定，采用数字、字母显示工作状态的家用火灾报警控制器应按显示火灾报警、监管报警及其他状态顺序由高至低排列信息显示等级。高等级的状态信息应优先显示，低等级状态信息显示不应影响高等级状态信息显示。显示的信息应与对应的状态一致且易于辨识。"}]}
{"messages": [{"role": "system", "content": "你是消防系统专家"}, {"role": "user", "content": "火灾报警器的声压级是多少？"}, {"role": "assistant", "content": "根据《家用火灾安全系统》（GB22370—2008）的规定，家用火灾报警控制器在各种报警状态下的声压级（正前方1米处）应不小于75dB(A计权)。"}]}
{"messages": [{"role": "system", "content": "你是消防系统专家"}, {"role": "user", "content": "可燃气体探测器怎么安装？"}, {"role": "assistant", "content": "根据《可燃气体报警控制器》（GB16808—2008）的规定，可燃气体探测器应按制造商的要求进行安装，并连接到可燃气体报警控制器上。安装位置应避免高温、潮湿和尘土较多的地方，以保证设备的正常运行。"}]}
{"messages": [{"role": "system", "content": "你是消防系统专家"}, {"role": "user", "content": "可燃气体探测器的报警功能有什么要求？"}, {"role": "assistant", "content": "根据《可燃气体报警控制器》（GB16808—2008）的规定，可燃气体探测器应具有低限报警或低限、高限两段报警功能。当有可燃气体报警信号输入时，控制器应在10秒内发出报警声、光信号，并保持至手动复位。"}]}
{"messages": [{"role": "system", "content": "你是消防系统专家"}, {"role": "user", "content": "如何检查火灾报警器的绝缘性能？"}, {"role": "assistant", "content": "根据《家用火灾安全系统》（GB22370—2008）的规定，通过绝缘电阻试验装置分别对试样的外部带电端子与机壳之间施加500V±50V直流电压，持续60秒±5秒后测量其绝缘电阻值。试样有绝缘要求的外部带电端子与机壳间的绝缘电阻值应不小于20MΩ；试样的电源输入端与机壳间的绝缘电阻值应不小于50MΩ。"}]}
If we have accumulated a relatively large amount of high-quality data in our daily business, we can start fine-tuning by constructing data files that meet the format requirements through simple data cleaning. If we lack data accumulation, or the cleaning process is more complicated, we can try to fine-tune with the help of big models and construct data with models.
Next, let's use an example to practice the process of constructing data. Let's try to train a large model of a "firefighting expert" by constructing questions and answers using firefighting standards.
Tectonic fine-tuning data
SiliconCloud provides a Qwen2.5 model with 128K contexts, which is more than enough to cover quite rich content.
First, we need to prepare locally the file to be used to construct the data, for example, GB22370-2008 Domestic Fire Safety Systems.pdf.
Next, we need to extract the contents of the file, write the prompt, and have the Qwen Big Model generate data that meets the above requirements. This can be done either in code, or using the Dify and other interface tools to complete.
Take the example of configuring a workflow in Dify:
1. The Start node configuration file is uploaded, along with the fill-in fields for the System Prompt in the training corpus;
2. Use the document extraction node and code node that comes with Dify to merge the contents of the file into text;
3. Using the LLM node, call SiliconCloud's Qwen/Qwen2.5-72B-Instruct-128K model (if the length of the file is within 32K, you can choose another model provided by the platform), if you can't find this model, you can add it by the following steps:
-
-
- Settings - Model Vendor, select SiliconFlow;
- Click Add Model, refer to the image to fill in the required parameters and save.
-
- A batch of corpus data can be obtained after the run by using the output of the big model as the output of the workflow;
- Next simply copy, save the contents as a .jsonl file and upload it in SiliconCloud to create the fine-tuning task.
If the data generated at one time is not sufficient, multiple runs can be made to construct more data.
It is important to note that more fine-tuned data is not necessarily betterEven if we have only 1 piece of data, we can start fine-tuning. Even if there is only 1 piece of data, we can start fine-tuning; on the contrary, if we have a lot of data that is not of good enough quality, the fine-tuning effect may not be as expected.
app: description: '上传文件，基于文件内容，使用 SiliconCloud 128K 上下文的 Qwen2.5 模型，生成日常问答内容，JSONL 格式的语料数据 ⚠️ 注： - 由于 Dify 限制，超过 80000 字符的文件内容会被截断 - 生成内容仅供参考，可能存在幻觉或内容错漏、格式错误，请注意甄别' icon: 🤖 icon_background: '#FFEAD5' mode: workflow name: Fine-tune 语料构造器 use_icon_as_answer_icon: false kind: app version: 0.1.5 workflow: conversation_variables: [] environment_variables: [] features: file_upload: allowed_file_extensions: - .JPG - .JPEG - .PNG - .GIF - .WEBP - .SVG allowed_file_types: - image allowed_file_upload_methods: - local_file - remote_url enabled: false fileUploadConfig: audio_file_size_limit: 50 batch_count_limit: 5 file_size_limit: 15 image_file_size_limit: 10 video_file_size_limit: 100 workflow_file_upload_limit: 10 image: enabled: false number_limits: 3 transfer_methods: - local_file - remote_url number_limits: 3 opening_statement: '' retriever_resource: enabled: true sensitive_word_avoidance: enabled: false speech_to_text: enabled: false suggested_questions: [] suggested_questions_after_answer: enabled: false text_to_speech: enabled: false language: '' voice: '' graph: edges: - data: isInIteration: false sourceType: start targetType: document-extractor id: 1735807686274-source-1735807758092-target source: '1735807686274' sourceHandle: source target: '1735807758092' targetHandle: target type: custom zIndex: 0 - data: isInIteration: false sourceType: document-extractor targetType: code id: 1735807758092-source-1735807761855-target source: '1735807758092' sourceHandle: source target: '1735807761855' targetHandle: target type: custom zIndex: 0 - data: isInIteration: false sourceType: code targetType: llm id: 1735807761855-source-1735807764975-target source: '1735807761855' sourceHandle: source target: '1735807764975' targetHandle: target type: custom zIndex: 0 - data: isInIteration: false sourceType: llm targetType: end id: 1735807764975-source-1735807769820-target source: '1735807764975' sourceHandle: source target: '1735807769820' targetHandle: target type: custom zIndex: 0 nodes: - data: desc: '' selected: false title: 开始 type: start variables: - allowed_file_extensions: [] allowed_file_types: - document allowed_file_upload_methods: - local_file - remote_url label: 语料文件 max_length: 10 options: [] required: true type: file-list variable: attachments - allowed_file_extensions: [] allowed_file_types: - image allowed_file_upload_methods: - local_file - remote_url label: 触发词（训练中的 system prompt） max_length: 48 options: [] required: true type: text-input variable: trigger height: 116 id: '1735807686274' position: x: 30 y: 258 positionAbsolute: x: 30 y: 258 selected: false sourcePosition: right targetPosition: left type: custom width: 244 - data: desc: '' is_array_file: true selected: false title: 文档提取器 type: document-extractor variable_selector: - '1735807686274' - attachments height: 92 id: '1735807758092' position: x: 334 y: 258 positionAbsolute: x: 334 y: 258 selected: false sourcePosition: right targetPosition: left type: custom width: 244 - data: code: "def main(articleSections: list) -> dict:\n try:\n # 将列表项合并为字符串\n\ \ combined_text = \"\\n\".join(articleSections)\n \n \ \ # 截取前80000个字符\n truncated_text = combined_text[:80000]\n \ \ \n return {\n \"result\": truncated_text\n \ \ }\n except Exception as e:\n # 错误处理\n return {\n \ \ \"result\": \"\"\n }" code_language: python3 desc: '' outputs: result: children: null type: string selected: false title: 代码执行 type: code variables: - value_selector: - '1735807758092' - text variable: articleSections height: 54 id: '1735807761855' position: x: 638 y: 258 positionAbsolute: x: 638 y: 258 selected: false sourcePosition: right targetPosition: left type: custom width: 244 - data: context: enabled: false variable_selector: [] desc: '' model: completion_params: frequency_penalty: 0.5 max_tokens: 4096 temperature: 0.3 mode: chat name: Qwen/Qwen2.5-72B-Instruct-128K provider: siliconflow prompt_template: - id: b6913d40-d173-45d8-b012-98240d42a196 role: system text: '【角色】 你是一位 LLM 大语言模型科学家，参考用户提供的内容，帮助用户构造符合规范的 Fine-tune（微调）数据 【任务】 - 对于给定的「内容」，你每次回列出 10 个通俗「问题」； - 针对每个「问题」，引用「内容」原文及对内容的合理解释和演绎，做出「解答」； - 并将「问题」「解答」整理为规范的 JSONL 格式 【要求】 1. 问题 **不要** 直接引用「内容」，应该贴近当代现实生活； 2. 问题应该是通俗白话，避免“假、大、空“； 3. 答案应忠于原文，对于原文的解释不能脱离原文的主旨、思想； 【输出规范】 * 输出规范的 JSONL，每行一条数据 * 每条数据应包含一个 message 数组，每个数组都应该包含 role 分别为 system、user 和 assistant 的三条记录 * 其中 role 为 system 的数据，作为训练中的 system prompt 格外重要，其 content 使用用户指定的「触发词」 * role 为 user 的数据对应列出的「问题」 * role 为 assistant 的数据则对应针对「问题」的「解答」 * 示例如下： ``` {"messages": [{"role": "system", "content": "你是当代大儒"}, {"role": "user", "content": "应该怎么学习？"}, {"role": "assistant", "content": "贤贤易色；事父母，能竭其力；事君，能致其身；与朋友交，言而有信。虽曰未学，吾必谓之学矣。"}]} ```' - id: 61530521-14cf-4eaf-8f06-a4bc89db3cb1 role: user text: '「内容」 {{#1735807761855.result#}} 「触发词」 {{#1735807686274.trigger#}}' selected: false title: LLM type: llm variables: [] vision: enabled: false height: 98 id: '1735807764975' position: x: 942 y: 258 positionAbsolute: x: 942 y: 258 selected: true sourcePosition: right targetPosition: left type: custom width: 244 - data: desc: '' outputs: - value_selector: - '1735807764975' - text variable: text selected: false title: 结束 type: end height: 90 id: '1735807769820' position: x: 1246 y: 258 positionAbsolute: x: 1246 y: 258 selected: false sourcePosition: right targetPosition: left type: custom width: 244 - data: author: Dify desc: '' height: 88 selected: false showAuthor: true text: '{"root":{"children":[{"children":[{"detail":0,"format":0,"mode":"normal","style":"","text":"设置较低的 Temperature，提高输出格式的稳定性","type":"text","version":1}],"direction":"ltr","format":"","indent":0,"type":"paragraph","version":1,"textFormat":0}],"direction":"ltr","format":"","indent":0,"type":"root","version":1}}' theme: blue title: '' type: '' width: 240 height: 88 id: '1735808753316' position: x: 951.4285714285714 y: 375.7142857142857 positionAbsolute: x: 951.4285714285714 y: 375.7142857142857 selected: false sourcePosition: right targetPosition: left type: custom-note width: 240 - data: author: Dify desc: '' height: 88 selected: false showAuthor: true text: '{"root":{"children":[{"children":[{"detail":0,"format":0,"mode":"normal","style":"","text":"合并多个文档内容，并截取前 8W 字符","type":"text","version":1}],"direction":"ltr","format":"","indent":0,"type":"paragraph","version":1,"textFormat":0}],"direction":"ltr","format":"","indent":0,"type":"root","version":1}}' theme: blue title: '' type: '' width: 240 height: 88 id: '1735808799815' position: x: 640 y: 338.5714285714286 positionAbsolute: x: 640 y: 338.5714285714286 selected: false sourcePosition: right targetPosition: left type: custom-note width: 240 viewport: x: 0 y: 0 zoom: 0.7
In addition to using tools such as Dify, directly using models such as Qwen/Qwen2.5-Coder-32B-Instruct to assist in writing scripts to run is also an efficient way to construct fine-tuned data, and it can circumvent a number of limitations of the three-party platforms, fully utilize the advantages of the SiliconCloud platform's model-rich, efficient inference, and harvest more high-quality corpus data. Data.
Fine-tuning training and validation
Create a new fine-tuning task on the SiliconCloud platform, upload and select the .jsonl file you just saved, and click Start Fine-Tuning. 
Wait for the fine-tuning task to finish queuing, and once the execution is complete, you can use the fine-tuned model or validate the results through an online experience or API. 
The Compare feature of the Dialog Model Online Experience also allows you to compare model responses and outputs across multiple Checkpoint and base models, choosing the model that works best to continue using. Fine-tuning model vs. base model: 
Comparison between multiple checkpoints: 
At this point, we have used local files to build LLM fine-tuned data and fine-tuned a Fire Expert LLM. of course, with the above steps, you can also try to build your own domain-specific LLM.
English 
简体中文 
日本語 
Deutsch 
Português do Brasil