---
title: Index-Translate-35B-A3B
canonical_url: "https://www.modelscope.cn/models/IndexTeam/Index-Translate-35B-A3B"
md_url: "https://www.modelscope.cn/models/IndexTeam/Index-Translate-35B-A3B.md"
repository: IndexTeam/Index-Translate-35B-A3B
last_updated: 2026-09-30
license: apache-2.0
model_type:
  - qwen3_5_moe
architectures:
  - Qwen3_5MoeForConditionalGeneration
parameters: 36.0B
tensor_type:
  - BF16
library_name:
  - safetensors
downloads: 5
stars: 0
tags:
  - translation
  - index
---

# Index-Translate-35B-A3B

> Index-Translate-35B-A3B - IndexTeam 在 ModelScope 开源的模型。Index-Translate-35B-A3B

IndexTeam/Index-Translate-35B-A3B 是 ModelScope 魔搭社区上的 36.0B 参数机器学习模型，采用 apache-2.0 许可。

- **Repository**: IndexTeam/Index-Translate-35B-A3B
- **License**: apache-2.0
- **Parameters**: 36.0B
- **Tags**: translation, index
- **Downloads**: 5
- **Stars**: 0
- **Last updated**: 2026-09-30

Source: https://www.modelscope.cn/models/IndexTeam/Index-Translate-35B-A3B

---

# Index-Translate-35B-A3B

[![在线体验](https://img.shields.io/badge/Demo-Online-orange.svg)](https://index-translate.bilibili.com/)
[![代码](https://img.shields.io/badge/GitHub-Index--Translate-yellow.svg)](https://github.com/bilibili/Index-Translate)
[![协议](https://img.shields.io/badge/License-Apache--2.0-lightgrey.svg)](https://github.com/bilibili/Index-Translate/blob/main/LICENSE)

**Index-Translate-35B-A3B** 是 **Index-Translate** 翻译模型家族中最大的多语言文本翻译模型：基于 Qwen3.5 骨干构建的 35B 总参 / 3B 激活 MoE 模型。支持 **150 种语言**，并具备多语言指令遵循能力——术语、格式、风格、结构与输出长度约束。

> 📄 技术报告：*Index-Translate: A Multilingual Translation Model Family — Text, Speech, Controlled Dubbing, and Long-Document Translation*(Index LLM Team,2026 年 9 月)
>
> 💻 代码、快速上手、浏览器扩展与视频配音管线：[github.com/bilibili/Index-Translate](https://github.com/bilibili/Index-Translate)
>
> 🤗 Hugging Face 同名仓库：[IndexTeam/Index-Translate-35B-A3B](https://huggingface.co/IndexTeam/Index-Translate-35B-A3B)

## 亮点

- 在技术报告的评测中，**FLORES COMET-22(0.8794)与 instTrans 指令遵循分(0.8336)均为全部参评系统最优**，参评对象包含前沿大模型。
- **150 种语言** + 指令遵循：通用翻译、受限翻译（术语、结构、风格、上下文）、梗/社区文化表达翻译，单一权重全部支持。
- **训练流程**：多语言中期训练（通用 + 单语 + 平行 + pivot 组织数据；167.77B tokens)→ 专家 SFT（通用 / 指令 / 梗翻译三专家)→ 判定奖励强化学习（GRPO；通用翻译用 XCOMET-XXL + 忠实度门控，指令遵循用 Rubric-as-Reward),RIVAL 对抗评审监督 → **参数插值**(model soup)→ 定向**多教师同策略蒸馏（MOPD)**。
- **低资源语种**：低资源 instTrans 上质量 0.5151 / 指令遵循 0.7715 / off-target 4.05%;FLORES 低资源语向上 COMET-22 0.8314 / XCOMET-XXL 0.7114 / off-target 2.59%。
- 同族小模型：[IndexTeam/Index-Translate-9B](https://modelscope.cn/models/IndexTeam/Index-Translate-9B)、[IndexTeam/Index-Translate-2B](https://modelscope.cn/models/IndexTeam/Index-Translate-2B)。

## 评测结果

文本翻译（技术报告 Table 4;35B-A3B 行为 preview 评测；WMT26 为 0–100 评审分，其余为 0–1):

| 模型 | FLORES COMET-22 | WMT24++ COMET-22 | WMT26 Judge | instTrans 质量 | instTrans IFscore | IFMTBench XCOMET-XXL | IFMTBench IFscore | 垂类均值 | MEME |
|---|---|---|---|---|---|---|---|---|---|
| **Index-Translate-35B-A3B (preview)** | **0.8794** | 0.8586 | 76.76 | 0.6901 | **0.8336** | 0.7926 | 0.8991 | 0.8438 | 0.7405 |
| **Index-Translate-9B** | 0.8789 | 0.8601 | 75.35 | 0.6771 | 0.8209 | 0.7957 | 0.8760 | 0.8451 | 0.7387 |
| **Index-Translate-2B** | 0.8655 | 0.8489 | 60.26 | 0.5391 | 0.7569 | 0.7712 | 0.7584 | 0.8377 | 0.6443 |
| Hy-MT2-1.8B | 0.8522 | 0.8401 | 49.35 | 0.3181 | 0.4932 | 0.7493 | 0.7161 | 0.8314 | 0.3643 |
| Hy-MT2-7B | 0.8747 | 0.8593 | 60.51 | 0.5143 | 0.6079 | 0.8049 | 0.8741 | 0.8335 | 0.5139 |
| Hy-MT2-30B-A3B | 0.8787 | **0.8624** | 66.81 | 0.5725 | 0.6415 | **0.8177** | 0.9029 | **0.8459** | 0.5812 |
| TranslateGemma-12B | 0.8732 | 0.8524 | 71.19 | 0.4515 | 0.3068 | 0.8023 | 0.2892 | 0.8347 | 0.4281 |
| North-Small-Translate (218B-A25B) | 0.8784 | 0.8578 | 68.37 | 0.5697 | 0.5294 | 0.7657 | 0.8635 | 0.8357 | 0.6836 |
| Qwen3.5-2B (base) | 0.6983 | 0.6933 | 32.11 | 0.0999 | 0.2431 | 0.6197 | 0.3836 | 0.7557 | 0.2062 |
| Qwen3.5-9B (base) | 0.8316 | 0.8073 | 60.31 | 0.2467 | 0.0609 | 0.7341 | 0.5980 | 0.8199 | 0.5728 |
| Qwen3.5-35B-A3B | 0.8570 | 0.8290 | 71.33 | 0.3690 | 0.5204 | 0.7589 | 0.7822 | 0.8267 | 0.6447 |
| DeepSeek-V4.1-Flash | 0.8762 | 0.8510 | 83.55 | 0.6068 | 0.6374 | 0.7817 | 0.9090 | 0.8432 | **0.7424** |
| GPT-5.6-Sol | 0.8650 | 0.8469 | **89.10** | **0.6902** | 0.7624 | 0.7946 | **0.9367** | 0.8311 | 0.7194 |
| Gemini 3.5 Flash Lite | 0.8750 | 0.8497 | 79.52 | 0.6068 | 0.6374 | 0.7764 | 0.8854 | 0.8131 | 0.7034 |

垂类翻译（COMET-22;GuoFeng 用 document-COMET):

| 模型 | OPUS-Books | Biomedical | MTNT | MuST-Cinema | GuoFeng |
|---|---|---|---|---|---|
| **Index-Translate-35B-A3B (preview)** | **0.8013** | **0.8681** | 0.8244 | 0.8962 | 0.8291 |
| Index-Translate-9B | 0.8010 | 0.8674 | **0.8246** | **0.8979** | **0.8345** |
| Index-Translate-2B | 0.7845 | 0.8635 | 0.8132 | 0.8930 | **0.8345** |

通用能力（答案准确率 %)——35B-A3B preview 在 GPQA-Diamond、INCLUDE、MMMLU 三项上超过 Hy-MT2-30B-A3B（仅 C-Eval 略低），整体仍低于 Qwen3.5-35B-A3B 基座：

| 模型 | C-Eval | GPQA-Diamond | INCLUDE | MMMLU |
|---|---|---|---|---|
| Index-Translate-35B-A3B (preview) | 77.6 | 49.1 | 68.8 | 71.9 |
| Index-Translate-9B | 69.6 | 36.3 | 63.1 | 66.6 |
| Hy-MT2-30B-A3B | 78.0 | 48.9 | 67.4 | 70.3 |
| Qwen3.5-35B-A3B | **89.8** | **78.8** | **76.7** | **83.0** |

## 使用方法

与家族其他模型一致，推荐 vLLM 部署：

```bash
vllm serve IndexTeam/Index-Translate-35B-A3B --served-model-name Index-Translate-35B-A3B --port 8000
```

```python
from openai import OpenAI

client = OpenAI(base_url="http://localhost:8000/v1", api_key="EMPTY")
resp = client.chat.completions.create(
    model="Index-Translate-35B-A3B",
    messages=[{"role": "user", "content": "Translate the following text into English:\n\n你好，世界！"}],
    temperature=0,
)
print(resp.choices[0].message.content)
```

Transformers 与 SGLang 快速上手见仓库 [inference/llm](https://github.com/bilibili/Index-Translate/tree/main/inference/llm)。

## 模型家族

| 模型 | 规格 | 任务 |
|---|---|---|
| **Index-Translate** | [2B](https://modelscope.cn/models/IndexTeam/Index-Translate-2B) / [9B](https://modelscope.cn/models/IndexTeam/Index-Translate-9B) / **35B-A3B** | 多语言文本翻译与翻译指令 |
| Index-Echo S2TT | [2B](https://modelscope.cn/models/IndexTeam/Index-Echo-S2TT-2B) / [9B](https://modelscope.cn/models/IndexTeam/Index-Echo-S2TT-9B) | 语音转文字翻译 |
| Index-Echo S2ST | [2B](https://modelscope.cn/models/IndexTeam/Index-Echo-S2ST-2B) / [9B](https://modelscope.cn/models/IndexTeam/Index-Echo-S2ST-9B) | 语音到语音翻译（参考源说话人声音） |
| Index-Homura | [2B](https://modelscope.cn/models/IndexTeam/Index-Homura-2B) / [9B](https://modelscope.cn/models/IndexTeam/Index-Homura-9B) | 音节可控翻译（配音） |
| Index-NativeLong (NaiLong) | [2B](https://modelscope.cn/models/IndexTeam/Index-Nailong-2B) / [9B](https://modelscope.cn/models/IndexTeam/Index-Nailong-9B) | 长文档翻译 |

## 引用

```bibtex
@techreport{indextranslate2026,
  title={Index-Translate: A Multilingual Translation Model Family --- Text, Speech, Controlled Dubbing, and Long-Document Translation},
  author={Index LLM Team},
  year={2026},
  month={September}
}
```

## 协议

Apache-2.0。
