diff --git a/README.md b/README.md
index a54f380..842b20b 100644
--- a/README.md
+++ b/README.md
@@ -1,12 +1,18 @@
+
+> [!NOTE]
+> 本文档由 WeHub 基于上游 README 翻译整理，属于社区翻译，非官方中文文档。
+> [English](./README.en.md) · [原始项目](https://github.com/lyogavin/airllm) · [上游 README](https://github.com/lyogavin/airllm/blob/HEAD/README.md)
+> 原作者、版权与许可证归属以原始项目及本仓库 LICENSE 文件为准。
+
-[**Quickstart**](#quickstart) |
-[**Configurations**](#configurations) |
+[**快速开始**](#quickstart) |
+[**配置项**](#configurations) |
[**MacOS**](#macos) |
-[**Example notebooks**](#example-python-notebook) |
-[**FAQ**](#faq)
+[**示例笔记本**](#example-python-notebook) |
+[**常见问题**](#faq)
-**AirLLM** dramatically reduces inference memory usage, letting 70B large language models run on a single 4GB GPU card — without quantization, distillation, or pruning. You can even run **405B Llama 3.1** on **8GB**, and **DeepSeek-V3 (671B)** on **~12GB**.
+**AirLLM** 可大幅降低推理内存占用，使 70B 大语言模型能在单张 4GB GPU 显卡上运行——无需量化、蒸馏或剪枝。你甚至可以在 **8GB** 上运行 **405B Llama 3.1**，在 **约 12GB** 上运行 **DeepSeek-V3（671B）**。
[](https://pepy.tech/project/airllm)
@@ -22,76 +28,76 @@
[](https://patreon.com/gavinli)
[](https://github.com/sponsors/lyogavin)
-## AI Agents Recommendation:
+## AI 智能体推荐：
-* [Best AI Game Sprite Generator](https://godmodeai.co)
+* [最佳 AI 游戏精灵生成器](https://godmodeai.co)
-* [Best AI Facial Expression Editor](https://crazyfaceai.com)
+* [最佳 AI 面部表情编辑器](https://crazyfaceai.com)
-* [Bloome — build & run AI agent teams in the cloud, zero setup](https://bloome.im/login?ref=G6BYnov0)
+* [Bloome — 在云端构建并运行 AI 智能体团队，零配置](https://bloome.im/login?ref=G6BYnov0)
-## Updates
-[2026/06] **v3.0**: FP8 model support + the latest models. Run **DeepSeek-V3 (671B) on ~12GB** and **Qwen3-235B on ~3GB**, plus Qwen3, Llama 3.x/4, DeepSeek V2/V3, Phi-4, Gemma and more — all through a single `AutoModel`.
+## 更新日志
+[2026/06] **v3.0**：支持 FP8 模型 + 最新模型。在 **约 12GB** 上运行 **DeepSeek-V3（671B）**，在 **约 3GB** 上运行 **Qwen3-235B**，以及 Qwen3、Llama 3.x/4、DeepSeek V2/V3、Phi-4、Gemma 等——全部通过单一的 `AutoModel` 完成。
-[2024/08/20] v2.11.0: Support Qwen2.5
+[2024/08/20] v2.11.0：支持 Qwen2.5
-[2024/08/18] v2.10.1 Support CPU inference. Support non sharded models. Thanks @NavodPeiris for the great work!
+[2024/08/18] v2.10.1 支持 CPU 推理。支持非分片模型。感谢 @NavodPeiris 的出色工作！
-[2024/07/30] Support Llama3.1 **405B** ([example notebook](https://colab.research.google.com/github/lyogavin/airllm/blob/main/air_llm/examples/run_llama3.1_405B.ipynb)). Support **8bit/4bit quantization**.
+[2024/07/30] 支持 Llama3.1 **405B**（[示例笔记本](https://colab.research.google.com/github/lyogavin/airllm/blob/main/air_llm/examples/run_llama3.1_405B.ipynb)). 支持 **8bit/4bit 量化**。
-[2024/04/20] AirLLM supports Llama3 natively already. Run Llama3 70B on 4GB single GPU.
+[2024/04/20] AirLLM 已原生支持 Llama3。可在 4GB 单卡 GPU 上运行 Llama3 70B。
-[2023/12/25] v2.8.2: Support MacOS running 70B large language models.
+[2023/12/25] v2.8.2：支持在 MacOS 上运行 70B 大语言模型。
-[2023/12/20] v2.7: Support AirLLMMixtral.
+[2023/12/20] v2.7：支持 AirLLMMixtral。
-[2023/12/20] v2.6: Added AutoModel, automatically detect model type, no need to provide model class to initialize model.
+[2023/12/20] v2.6：新增 AutoModel，自动检测模型类型，初始化模型时无需再提供模型类。
-[2023/12/18] v2.5: added prefetching to overlap the model loading and compute. 10% speed improvement.
+[2023/12/18] v2.5：加入预取（prefetching）以重叠模型加载与计算。速度提升 10%。
-[2023/12/03] added support of **ChatGLM**, **QWen**, **Baichuan**, **Mistral**, **InternLM**!
+[2023/12/03] 新增对 **ChatGLM**、**QWen**、**Baichuan**、**Mistral**、**InternLM** 的支持！
-[2023/12/02] added support for safetensors. Now support all top 10 models in open llm leaderboard.
+[2023/12/02] 新增对 safetensors 的支持。现已支持 open llm leaderboard 上全部前 10 名模型。
-[2023/12/01] airllm 2.0. Support compressions: **3x run time speed up!**
+[2023/12/01] airllm 2.0。支持压缩：**运行时加速 3 倍！**
-[2023/11/20] airllm Initial version!
+[2023/11/20] airllm 初始版本！
-## Star History
+## Star 历史
-
+
-## Table of Contents
+## 目录
-* [Quick start](#quickstart)
-* [Model Compression](#model-compression---3x-inference-speed-up)
-* [Configurations](#configurations)
-* [Run on MacOS](#macos)
-* [Example notebooks](#example-python-notebook)
-* [Supported Models](#supported-models)
-* [Acknowledgement](#acknowledgement)
-* [FAQ](#faq)
+* [快速开始](#quickstart)
+* [模型压缩](#model-compression---3x-inference-speed-up)
+* [配置项](#configurations)
+* [在 MacOS 上运行](#macos)
+* [示例笔记本](#example-python-notebook)
+* [支持的模型](#supported-models)
+* [致谢](#acknowledgement)
+* [常见问题](#faq)
-## Quickstart
+## 快速开始
-### 1. Install package
+### 1. 安装包
-First, install the airllm pip package.
+首先，安装 airllm 的 pip 包。
```bash
pip install airllm
```
-### 2. Inference
+### 2. 推理
-Then, initialize AirLLMLlama2, pass in the huggingface repo ID of the model being used, or the local path, and inference can be performed similar to a regular transformer model.
+然后，初始化 AirLLMLlama2，传入所用模型的 huggingface 仓库 ID 或本地路径，即可像常规 transformer 模型一样进行推理。
-(*You can also specify the path to save the splitted layered model through **layer_shards_saving_path** when init AirLLMLlama2.*
+（*你也可以在初始化 AirLLMLlama2 时通过 **layer_shards_saving_path** 指定保存按层切分后模型的路径。*
```python
from airllm import AutoModel
@@ -132,20 +138,20 @@ print(output)
```
-Note: During inference, the original model will first be decomposed and saved layer-wise. Please ensure there is sufficient disk space in the huggingface cache directory.
+注意：推理过程中，原始模型会先被按层分解并保存。请确保 huggingface 缓存目录有足够的磁盘空间。
-## Model Compression - 3x Inference Speed Up!
+## 模型压缩 - 推理加速 3 倍！
-We just added model compression based on block-wise quantization-based model compression. Which can further **speed up the inference speed** for up to **3x** , with **almost ignorable accuracy loss!** (see more performance evaluation and why we use block-wise quantization in [this paper](https://arxiv.org/abs/2212.09720))
+我们刚刚加入了基于块级量化（block-wise quantization）的模型压缩。可将**推理速度**进一步提升至最高 **3 倍**，且**精度损失几乎可忽略！**（更多性能评估以及我们为何使用块级量化，请参阅[本文](https://arxiv.org/abs/2212.09720))
-#### How to enable model compression speed up:
+#### 如何启用模型压缩加速：
-* Step 1. make sure you have [bitsandbytes](https://github.com/TimDettmers/bitsandbytes) installed by `pip install -U bitsandbytes `
-* Step 2. make sure airllm verion later than 2.0.0: `pip install -U airllm`
-* Step 3. when initialize the model, passing the argument compression ('4bit' or '8bit'):
+* 步骤 1. 确保已安装 [bitsandbytes](https://github.com/TimDettmers/bitsandbytes)，通过 `pip install -U bitsandbytes `
+* 步骤 2. 确保 airllm 版本高于 2.0.0：`pip install -U airllm`
+* 步骤 3. 初始化模型时传入 compression 参数（'4bit' 或 '8bit'）：
```python
model = AutoModel.from_pretrained("garage-bAInd/Platypus2-70B-instruct",
@@ -153,48 +159,48 @@ model = AutoModel.from_pretrained("garage-bAInd/Platypus2-70B-instruct",
)
```
-#### What are the differences between model compression and quantization?
+#### 模型压缩与量化有何不同？
-Quantization normally needs to quantize both weights and activations to really speed things up. Which makes it harder to maintain accuracy and avoid the impact of outliers in all kinds of inputs.
+量化通常需要对权重和激活值都进行量化才能真正加速，这使保持精度并避免各类输入中异常值的影响变得更困难。
-While in our case the bottleneck is mainly at the disk loading, we only need to make the model loading size smaller. So, we get to only quantize the weights' part, which is easier to ensure the accuracy.
+而在我们的场景中，瓶颈主要在磁盘加载，我们只需让模型加载体积更小。因此，我们只需量化权重部分，这样更容易保证精度。
-## Configurations
+## 配置项
-When initialize the model, we support the following configurations:
+初始化模型时，我们支持以下配置：
-* **compression**: supported options: 4bit, 8bit for 4-bit or 8-bit block-wise quantization, or by default None for no compression
-* **profiling_mode**: supported options: True to output time consumptions or by default False
-* **layer_shards_saving_path**: optionally another path to save the splitted model
-* **hf_token**: huggingface token can be provided here if downloading gated models like: *meta-llama/Llama-2-7b-hf*
-* **prefetching**: prefetching to overlap the model loading and compute. By default, turned on. For now, only AirLLMLlama2 supports this.
-* **delete_original**: if you don't have too much disk space, you can set delete_original to true to delete the original downloaded hugging face model, only keep the transformed one to save half of the disk space.
+* **compression**：支持选项：4bit、8bit 表示 4 位或 8 位块级量化，或默认 None 表示不压缩
+* **profiling_mode**：支持选项：True 输出耗时，或默认 False
+* **layer_shards_saving_path**：可选，用于保存切分后模型的另一路径
+* **hf_token**：若下载如 *meta-llama/Llama-2-7b-hf* 等受限模型，可在此提供 huggingface token
+* **prefetching**：预取以重叠模型加载与计算。默认开启。目前仅 AirLLMLlama2 支持此功能。
+* **delete_original**：若磁盘空间不足，可将 delete_original 设为 true 以删除原始下载的 hugging face 模型，仅保留转换后的模型，从而节省一半磁盘空间。
## MacOS
-Just install airllm and run the code the same as on linux. See more in [Quick Start](#quickstart).
+只需安装 airllm，并像在 Linux 上一样运行代码。详见 [快速开始](#quickstart)。
-* make sure you installed [mlx](https://github.com/ml-explore/mlx?tab=readme-ov-file#installation) and torch
-* you probably need to install python native see more [here](https://stackoverflow.com/a/65432861/21230266)
-* only [Apple silicon](https://support.apple.com/en-us/HT211814) is supported
+* 请确保已安装 [mlx](https://github.com/ml-explore/mlx?tab=readme-ov-file#installation) 和 torch
+* 你可能需要安装原生 Python，详见 [此处](https://stackoverflow.com/a/65432861/21230266)
+* 仅支持 [Apple silicon](https://support.apple.com/en-us/HT211814)
-Example [python notebook] (https://github.com/lyogavin/airllm/blob/main/air_llm/examples/run_on_macos.ipynb)
+示例 [python notebook] (https://github.com/lyogavin/airllm/blob/main/air_llm/examples/run_on_macos.ipynb)
## Example Python Notebook
-Example colabs here:
+示例 Colab 如下：
-#### example of other models (ChatGLM, QWen, Baichuan, Mistral, etc):
+#### 其他模型示例（ChatGLM、QWen、Baichuan、Mistral 等）：
-* ChatGLM:
+* ChatGLM：
```python
from airllm import AutoModel
@@ -215,7 +221,7 @@ generation_output = model.generate(
model.tokenizer.decode(generation_output.sequences[0])
```
-* QWen:
+* QWen：
```python
from airllm import AutoModel
@@ -236,7 +242,7 @@ model.tokenizer.decode(generation_output.sequences[0])
```
-* Baichuan, InternLM, Mistral, etc:
+* Baichuan、InternLM、Mistral 等：
```python
from airllm import AutoModel
@@ -262,19 +268,19 @@ model.tokenizer.decode(generation_output.sequences[0])
-#### To request other model support: [here](https://docs.google.com/forms/d/e/1FAIpQLSe0Io9ANMT964Zi-OQOq1TJmnvP-G3_ZgQDhP7SatN0IEdbOg/viewform?usp=sf_link)
+#### 申请支持其他模型：[此处](https://docs.google.com/forms/d/e/1FAIpQLSe0Io9ANMT964Zi-OQOq1TJmnvP-G3_ZgQDhP7SatN0IEdbOg/viewform?usp=sf_link)
## Supported Models
-AirLLM works out of the box with **virtually every popular open LLM** — just pass its Hugging Face ID to `AutoModel.from_pretrained(...)`. That covers all the major families:
+AirLLM 开箱即用，可支持**几乎所有主流开源 LLM**——只需将其 Hugging Face ID 传给 `AutoModel.from_pretrained(...)`。这覆盖了所有主要模型家族：
-**Llama** (2 / 3 / 3.1 / 3.3 / 4) · **Qwen** (1 / 2 / 2.5 / 3, including MoE and FP8) · **DeepSeek** (V2 / V3 / R1) · **Mistral & Mixtral** · **Phi** · **Gemma** · **ChatGLM** · **Baichuan** · **InternLM** · **Yi** — and most new models the day they're released.
+**Llama**（2 / 3 / 3.1 / 3.3 / 4）· **Qwen**（1 / 2 / 2.5 / 3，含 MoE 与 FP8）· **DeepSeek**（V2 / V3 / R1）· **Mistral & Mixtral** · **Phi** · **Gemma** · **ChatGLM** · **Baichuan** · **InternLM** · **Yi**——以及大多数新模型在发布当天即可使用。
-### Tiny GPU, huge models
+### 小显存，大模型
-The trick: AirLLM only ever keeps **one layer on the GPU at a time**, so the VRAM you need depends on the model's layer size — not its total size. That's how a 671B model fits on a hobbyist card:
+关键在于：AirLLM 始终**一次只在 GPU 上保留一层**，因此所需 VRAM 取决于模型的单层大小，而非模型总大小。这就是 671B 模型能在业余级显卡上运行的原因：
| Model | Size | GPU VRAM |
|---|---|---|
@@ -285,15 +291,15 @@ The trick: AirLLM only ever keeps **one layer on the GPU at a time**, so the VRA
| Llama 3.1 405B | 405B | **~8 GB** |
| DeepSeek-V3 | **671B** | **~12 GB** |
-Same one line of code for all of them — no special setup.
+以上模型均使用同一行代码——无需特殊配置。
## Acknowledgement
-A lot of the code are based on SimJeg's great work in the Kaggle exam competition. Big shoutout to SimJeg:
+大量代码基于 SimJeg 在 Kaggle 考试竞赛中的出色工作。向 SimJeg 致以诚挚感谢：
-[GitHub account @SimJeg](https://github.com/SimJeg),
-[the code on Kaggle](https://www.kaggle.com/code/simjeg/platypus2-70b-with-wikipedia-rag),
-[the associated discussion](https://www.kaggle.com/competitions/kaggle-llm-science-exam/discussion/446414).
+[GitHub 账号 @SimJeg](https://github.com/SimJeg),
+[Kaggle 上的代码](https://www.kaggle.com/code/simjeg/platypus2-70b-with-wikipedia-rag),
+[相关讨论](https://www.kaggle.com/competitions/kaggle-llm-science-exam/discussion/446414).
## FAQ
@@ -302,20 +308,20 @@ A lot of the code are based on SimJeg's great work in the Kaggle exam competitio
safetensors_rust.SafetensorError: Error while deserializing header: MetadataIncompleteBuffer
-If you run into this error, most possible cause is you run out of disk space. The process of splitting model is very disk-consuming. See [this](https://huggingface.co/TheBloke/guanaco-65B-GPTQ/discussions/12). You may need to extend your disk space, clear huggingface [.cache](https://huggingface.co/docs/datasets/cache) and rerun.
+若遇到此错误，最可能的原因是磁盘空间不足。拆分模型的过程非常消耗磁盘空间。参见 [此说明](https://huggingface.co/TheBloke/guanaco-65B-GPTQ/discussions/12). 你可能需要扩容磁盘、清理 huggingface [.cache](https://huggingface.co/docs/datasets/cache) 后重新运行。
### 2. ValueError: max() arg is an empty sequence
-Most likely you are loading QWen or ChatGLM model with Llama2 class. Try the following:
+很可能是你用 Llama2 类加载了 QWen 或 ChatGLM 模型。请尝试以下方式：
-For QWen model:
+对于 QWen 模型：
```python
from airllm import AutoModel #<----- instead of AirLLMLlama2
AutoModel.from_pretrained(...)
```
-For ChatGLM model:
+对于 ChatGLM 模型：
```python
from airllm import AutoModel #<----- instead of AirLLMLlama2
@@ -324,7 +330,7 @@ AutoModel.from_pretrained(...)
### 3. 401 Client Error....Repo model ... is gated.
-Some models are gated models, needs huggingface api token. You can provide hf_token:
+部分模型为受限（gated）模型，需要 huggingface api token。你可以提供 hf_token：
```python
model = AutoModel.from_pretrained("meta-llama/Llama-2-7b-hf", #hf_token='HF_API_TOKEN')
@@ -332,7 +338,7 @@ model = AutoModel.from_pretrained("meta-llama/Llama-2-7b-hf", #hf_token='HF_API_
### 4. ValueError: Asking to pad but the tokenizer does not have a padding token.
-Some model's tokenizer doesn't have padding token, so you can set a padding token or simply turn the padding config off:
+部分模型的 tokenizer 没有 padding token，因此你可以设置 padding token，或直接关闭 padding 配置：
```python
input_tokens = model.tokenizer(input_text,
@@ -367,17 +373,17 @@ BibTex entry:
-### Run AI Agent Teams in the Cloud — Bloome
+### 在云端运行 AI Agent 团队 — Bloome
-Bloome is an AI-agent IM platform: build and run AI agent teams in the cloud with zero setup. Add a skill as an agent in a group chat, run it in one click from web or mobile, and share it with your team — think of it as a group chat where your AI assistants are teammates you can @mention and assign tasks to.
+Bloome 是一个 AI agent IM 平台：无需配置，即可在云端构建并运行 AI agent 团队。在群聊中将 skill 添加为 agent，从网页或移动端一键运行，并与团队共享——可以把它理解为一个群聊，你的 AI 助手就是队友，可以 @mention 并分配任务。
-👉 Try Bloome: https://bloome.im/login?ref=G6BYnov0
+👉 试用 Bloome：https://bloome.im/login?ref=G6BYnov0
## Contribution
-Welcomed contributions, ideas and discussions!
+欢迎贡献、想法与讨论！
-If you find it useful, please ⭐ or buy me a coffee! 🙏
+如果觉得有用，请给个 ⭐ 或请我喝杯咖啡！🙏
[](https://bmc.link/lyogavinQ)
+ 
-#### example of other models (ChatGLM, QWen, Baichuan, Mistral, etc):
+#### 其他模型示例（ChatGLM、QWen、Baichuan、Mistral 等）：
-### Run AI Agent Teams in the Cloud — Bloome
+### 在云端运行 AI Agent 团队 — Bloome
-Bloome is an AI-agent IM platform: build and run AI agent teams in the cloud with zero setup. Add a skill as an agent in a group chat, run it in one click from web or mobile, and share it with your team — think of it as a group chat where your AI assistants are teammates you can @mention and assign tasks to.
+Bloome 是一个 AI agent IM 平台：无需配置，即可在云端构建并运行 AI agent 团队。在群聊中将 skill 添加为 agent，从网页或移动端一键运行，并与团队共享——可以把它理解为一个群聊，你的 AI 助手就是队友，可以 @mention 并分配任务。
-👉 Try Bloome: https://bloome.im/login?ref=G6BYnov0
+👉 试用 Bloome：https://bloome.im/login?ref=G6BYnov0
## Contribution
-Welcomed contributions, ideas and discussions!
+欢迎贡献、想法与讨论！
-If you find it useful, please ⭐ or buy me a coffee! 🙏
+如果觉得有用，请给个 ⭐ 或请我喝杯咖啡！🙏
[](https://bmc.link/lyogavinQ)