phi3

我们的模型并非专门为所有下游目的而设计或评估。开发人员在选择用例时应考虑语言模型的常见局限性，并在特定下游用例中使用模型之前评估和缓解准确性、安全性以及公平性，尤其是在高风险场景中。开发人员应了解并遵守与用例相关的适用法律法规（包括隐私、贸易合规法等）。本模型卡片中没有任何内容应被解释为或视为对模型发布所依据的许可证的限制或修改。

负责任的 AI 注意事项

与其他语言模型一样，Phi 系列模型可能会以不公平、不可靠或冒犯的方式运行。需要注意的一些限制行为包括

服务质量：Phi 模型主要针对英文文本进行训练。非英语语言的性能将更差。与训练数据中代表性较低的英语语言变体相比，标准美式英语的性能可能更差。
对伤害的代表性和对刻板印象的延续：这些模型可能过度或不足地代表群体，抹杀某些群体的代表性，或强化贬低或负面的刻板印象。尽管进行了安全后训练，但由于不同群体代表性的差异或训练数据中反映现实世界模式和社会偏见的负面刻板印象的普遍性，这些局限性可能仍然存在。
不当或冒犯性内容：这些模型可能会生成其他类型的不当或冒犯性内容，这可能使它们不适合在敏感的上下文中部署，除非采用针对具体用例的额外缓解措施。
信息可靠性：语言模型可以生成无意义的内容或捏造听起来合理但实际上不准确或过时的内容。
代码范围有限：Phi-3 的大部分训练数据基于 Python，并使用常见的包，例如“typing、math、random、collections、datetime、itertools”。如果模型生成的 Python 脚本利用其他包或其他语言的脚本，我们强烈建议用户手动验证所有 API 使用情况。

开发人员应应用负责任的 AI 最佳实践，并负责确保特定用例遵守相关的法律法规（例如隐私、贸易等）。需要考虑的重要领域包括：+ 分配：如果没有进一步评估和额外的去偏见技术，模型可能不适合可能对法律地位或资源或生活机会（例如住房、就业、信贷等）分配产生重大影响的场景。

高风险场景：开发人员应评估在高风险场景中使用模型的适宜性，在这种场景中，不公平、不可靠或冒犯性的输出可能会造成巨大损失或导致伤害。这包括在准确性和可靠性至关重要的敏感或专业领域提供建议（例如法律或健康建议）。应根据部署环境在应用程序级别实施额外的安全措施。
错误信息：模型可能会生成不准确的信息。开发人员应遵循透明度最佳实践，并告知最终用户他们正在与 AI 系统进行交互。在应用程序级别，开发人员可以构建反馈机制和管道，将响应与特定用例的上下文信息联系起来，这种技术被称为检索增强生成 (RAG)。
有害内容生成：开发人员应根据其上下文评估输出，并使用适用于其用例的可用安全分类器或自定义解决方案。
滥用：其他形式的滥用，例如欺诈、垃圾邮件或恶意软件生产可能是可能的，开发人员应确保其应用程序不违反适用的法律法规。

训练

模型

架构：Phi-3 迷你拥有 3.8B 个参数，是一个密集的仅解码器 Transformer 模型。该模型通过监督微调 (SFT) 和直接偏好优化 (DPO) 进行微调，以确保与人类偏好和安全指南保持一致。
输入：文本。它最适合使用聊天格式的提示。
上下文长度：128K 个标记
GPU：512 个 H100-80G
训练时间：7 天
训练数据：3.3T 个标记
输出：响应输入生成的文本
日期：我们的模型是在 2024 年 2 月至 4 月之间训练的
状态：这是一个在 2023 年 10 月截止日期的离线数据集上训练的静态模型。随着我们改进模型，可能会发布调优模型的未来版本。

数据集

我们的训练数据包括来自各种来源的 3.3 万亿个词元，是以下内容的组合：1) 经过严格质量筛选的公开可用文档，精选的高质量教育数据和代码；2) 为了教学数学、编码、常识推理、对世界的普遍知识（科学、日常活动、心智理论等）而新创建的合成“教科书式”数据；3)涵盖各种主题的高质量聊天格式监督数据，以反映人们对不同方面的偏好，例如指令遵循、真实性、诚实和乐于助人。

软件

许可证

该模型在 MIT 许可证下获得许可。

商标

此项目可能包含项目、产品或服务的商标或徽标。授权使用微软商标或徽标须遵守并遵循 微软商标和品牌指南。在修改版本的此项目中使用微软商标或徽标不得造成混淆或暗示微软赞助。任何使用第三方商标或徽标均受该第三方政策的约束。

资源

Phi-3 is a family of open AI models developed by Microsoft.

## Parameter sizes

- [Phi-3 Mini](https://ollama.org.cn/library/phi3:mini) – 3B parameters – `ollama run phi3:mini`
- [Phi-3 Medium](https://ollama.org.cn/library/phi3:medium) – 14B parameters – `ollama run phi3:medium`

## Context window sizes

> Note: the 128k version of this model requires [Ollama 0.1.39](https://github.com/ollama/ollama/releases/tag/v0.1.39) or later.

- 4k `ollama run phi3:mini` `ollama run phi3:medium`
- 128k `ollama run phi3:medium-128k`

![image.png](https://ollama.org.cn/assets/library/phi3/83b3de66-82d8-4455-9117-256802c1b82e)

## Phi-3 Mini

Phi-3 Mini is a 3.8B parameters, lightweight, state-of-the-art open model trained with the Phi-3 datasets that includes both synthetic data and the filtered publicly available websites data with a focus on high-quality and reasoning dense properties.

The model has underwent a post-training process that incorporates both supervised fine-tuning and direct preference optimization to ensure precise instruction adherence and robust safety measures.

When assessed against benchmarks testing common sense, language understanding, math, code, long context and logical reasoning, Phi-3 Mini-4K-Instruct showcased a robust and state-of-the-art performance among models with less than 13 billion parameters.

## Phi-3 Medium

Phi-3 Medium is a 14B parameter language model, and outperforms Gemini 1.0 Pro.

![image.png](https://ollama.org.cn/assets/library/phi3/2868e29b-3bba-4c4a-a6ed-1a27fb102867)

## Intended Uses

**Primary use cases**

The model is intended for commercial and research use in English. The model provides uses for applications which require
1) memory/compute constrained environments
2) latency bound scenarios
3) strong reasoning (especially math and logic)
4) long context

Our model is designed to accelerate research on language and multimodal models, for use as a building block for generative AI powered features.

**Use case considerations**

Our models are not specifically designed or evaluated for all downstream purposes. Developers should consider common limitations of language models as they select use cases, and evaluate and mitigate for accuracy, safety, and fariness before using within a specific downstream use case, particularly for high risk scenarios.
Developers  should be aware of and adhere to applicable laws or regulations (including privacy, trade compliance laws, etc.) that are relevant to their use case.
Nothing contained in this Model Card should be interpreted as or deemed a restriction or modification to the license the model is released under.

## Responsible AI Considerations
Like other language models, the Phi series models can potentially behave in ways that are unfair, unreliable, or offensive. Some of the limiting behaviors to be aware of include:

+ Quality of Service: the Phi models are trained primarily on English text. Languages other than English will experience worse performance. English language varieties with less representation in the training data might experience worse performance than standard American English.

+ Representation of Harms & Perpetuation of Stereotypes: These models can over- or under-represent groups of people, erase representation of some groups, or reinforce demeaning or negative stereotypes. Despite safety post-training, these limitations may still be present due to differing levels of representation of different groups or prevalence of examples of negative stereotypes in training data that reflect real-world patterns and societal biases.

+ Inappropriate or Offensive Content: these models may produce other types of inappropriate or offensive content, which may make it inappropriate to deploy for sensitive contexts without additional mitigations that are specific to the use case.

+ Information Reliability: Language models can generate nonsensical content or fabricate content that might sound reasonable but is inaccurate or outdated.
+ Limited Scope for Code: Majority of Phi-3 training data is based in Python and use common packages such as "typing, math, random, collections, datetime, itertools". If the model generates Python scripts that utilize other packages or scripts in other languages, we strongly recommend users manually verify all API uses.

Developers should apply responsible AI best practices and are responsible for ensuring that a specific use case complies with relevant laws and regulations (e.g. privacy, trade, etc.). Important areas for consideration include:
+ Allocation: Models may not be suitable for scenarios that could have consequential impact on legal status or the allocation of resources or life opportunities (ex: housing, employment, credit, etc.) without further assessments and additional debiasing techniques.

+ High-Risk Scenarios: Developers should assess suitability of using models in high-risk scenarios where unfair, unreliable or offensive outputs might be extremely costly or lead to harm. This includes providing advice in sensitive or expert domains where accuracy and reliability are critical (ex: legal or health advice). Additional safeguards should be implemented at the application level according to the deployment context.

+ Misinformation: Models may produce inaccurate information. Developers should follow transparency best practices and inform end-users they are interacting with an AI system. At the application level, developers can build feedback mechanisms and pipelines to ground responses in use-case specific, contextual information, a technique known as Retrieval Augmented Generation (RAG).

+ Generation of Harmful Content: Developers should assess outputs for their context and use available safety classifiers or custom solutions appropriate for their use case.

+ Misuse: Other forms of misuse such as fraud, spam, or malware production may be possible, and developers should ensure that their applications do not violate applicable laws and regulations.

## Training

### Model

* Architecture: Phi-3 Mini has 3.8B parameters and is a dense decoder-only Transformer model. The model is fine-tuned with Supervised fine-tuning (SFT) and Direct Preference Optimization (DPO) to ensure alignment with human preferences and safety guidelines.
* Inputs: Text. It is best suited for prompts using chat format.
* Context length: 128K tokens
* GPUS: 512 H100-80G
* Training time: 7 days
* Training data: 3.3T tokens
* Outputs: Generated text in response to the input
* Dates: Our models were trained between February and April 2024
* Status: This is a static model trained on an offline dataset with cutoff date October 2023. Future versions of the tuned models may be released as we improve models.

### Datasets
Our training data includes a wide variety of sources, totaling 3.3 trillion tokens, and is a combination of
1) publicly available documents filtered rigorously for quality, selected high-quality educational data, and code;
2) newly created synthetic, “textbook-like” data for the purpose of teaching math, coding, common sense reasoning, general knowledge of the world (science, daily activities, theory of mind, etc.);
3) high quality chat format supervised data covering various topics to reflect human preferences on different aspects such as instruct-following, truthfulness, honesty and helpfulness.

### Software

* [PyTorch](https://github.com/pytorch/pytorch)
* [DeepSpeed](https://github.com/microsoft/DeepSpeed)
* [Transformers](https://github.com/huggingface/transformers)
* [Flash-Attention](https://github.com/HazyResearch/flash-attention)

### License

The model is licensed under the [MIT license](https://ollama.org.cn/library/phi3:latest/blobs/fa8235e5b48fENSE).

## Trademarks

This project may contain trademarks or logos for projects, products, or services. Authorized use of Microsoft trademarks or logos is subject to and must follow [Microsoft’s Trademark & Brand Guidelines](https://www.microsoft.com/en-us/legal/intellectualproperty/trademarks). Use of Microsoft trademarks or logos in modified versions of this project must not cause confusion or imply Microsoft sponsorship. Any use of third-party trademarks or logos are subject to those third-party’s policies.

## Resources

+ [HuggingFace](https://hugging-face.cn/microsoft/Phi-3-mini-4k-instruct-gguf)
+ [Phi-3 Microsoft Blog](https://aka.ms/phi3-blog)
+ [Phi-3 Technical Report](https://aka.ms/phi3-tech-report)
+ [Phi-3 on Azure AI Studio](https://aka.ms/phi3-azure-ai)
+ [Phi-3 on Hugging Face](https://aka.ms/phi3-hf)
+ Phi-3 ONNX: [4K](https://aka.ms/phi3-mini-4k-instruct-onnx) and [128K](https://aka.ms/phi3-mini-128k-instruct-onnx)

粘贴、拖放或点击上传图像（.png、.jpeg、.jpg、.svg、.gif）