docs: align model card with technical report and inference defaults

README.md

@@ -7,89 +7,226 @@ tags:
 
 # Index-Translate-2B
 
-[![Demo](https://img.shields.io/badge/Demo-Online-orange.svg)](https://index-translate.bilibili.com/)
-[![Code](https://img.shields.io/badge/GitHub-Index--Translate-yellow.svg)](https://github.com/bilibili/Index-Translate)
-[![License](https://img.shields.io/badge/License-Apache--2.0-lightgrey.svg)](https://github.com/bilibili/Index-Translate/blob/main/LICENSE)
+[Online demo](https://index-translate.bilibili.com/) · [GitHub](https://github.com/bilibili/Index-Translate) · [Technical report](https://github.com/bilibili/Index-Translate/blob/main/docs/Index_Translate_Series_Technical_Report.pdf) · [Hugging Face collection](https://huggingface.co/collections/IndexTeam/index-translate) · [ModelScope collection](https://www.modelscope.cn/collections/IndexTeam/Index-Translate)
 
-**Index-Translate-2B** is the 2B-parameter multilingual text translation model of the **Index-Translate** family, built on the Qwen3.5 backbone. It supports **150 languages** with multilingual instruction following — terminology, formatting, style, structure, and output-length constraints — and powers the translate page of the [online demo](https://index-translate.bilibili.com/).
+**Index-Translate-2B** is the compact 2B-parameter multilingual text model of the **Index-Translate** family, built on Qwen3.5. It supports text translation across **150 languages** and follows translation instructions involving terminology, formatting, style, structure, context, and output length.
 
-> 📄 Technical report: *Index-Translate: A Multilingual Translation Model Family — Text, Speech, Controlled Dubbing, and Long-Document Translation* (Index LLM Team, September 2026)
->
-> 💻 Code, quickstarts, browser extension & video-dubbing pipeline: [github.com/bilibili/Index-Translate](https://github.com/bilibili/Index-Translate)
+## Capabilities
 
-## Highlights
+- **General multilingual translation:** translate sentences, articles, subtitles, and other text across the supported language inventory.
+- **Translation with instructions:** preserve specified terms, JSON or other structures, code and placeholders; adapt style and resolve meaning from context.
+- **Social and cultural translation:** interpret community aliases, playful spelling, memes, and nonliteral expressions with attention to their intended meaning.
+- Outperforms Hy-MT2-1.8B on every main translation metric reported for both models in the technical report.
+- Scores **0.7569 instTrans IFscore** and **0.6443 MEME**, combining instruction adherence with contextual translation of community expressions.
+- On low-resource instruction translation, scores **0.3050 quality / 0.6586 IFscore**, with **3.97% off-target outputs**.
 
-- **150 languages** with instruction following: general translation, constrained translation (terminology, structure, style, context), and meme/internet-culture translation in a single checkpoint.
-- **Training recipe**: multilingual mid-training (general + monolingual + parallel + pivot-organized data) → specialist SFT (general / instruction / meme experts) → RL with judged rewards (GRPO; XCOMET-XXL + adequacy gates for general translation, Rubric-as-Reward for instruction following) under RIVAL adversarial judge supervision → **parameter interpolation** (model soup, general : instruction : meme = 0.8 : 0.1 : 0.1) → targeted **multi-teacher on-policy distillation (MOPD)** for remaining weaknesses.
-- **262,144-token context window**, greedy decoding recommended.
-- Outperforms the same-tier **Hy-MT2-1.8B** on every metric reported for both models (Table 4 of the report).
-- Larger sibling: [IndexTeam/Index-Translate-9B](https://huggingface.co/IndexTeam/Index-Translate-9B); a **35B-A3B** variant is reported as a preview in the technical report.
+Language coverage is documented in the technical report's language inventory. The speech and long-document models have their own packaged interfaces and language coverage.
 
-## Benchmarks
+## Training
 
-Text translation (Table 4 of the report; WMT26 judge on a 0–100 scale, others 0–1):
+The technical report describes a shared multilingual mid-training recipe, followed by task-specific post-training:
+
+1. **Multilingual mid-training:** general text replay, monolingual text, ordinary parallel translations, and pivot-organized multilingual groups. The constant stage uses general / parallel / monolingual data at **1:1:1**; decay uses general / core-pivot / full-pivot data at **1:4:2**. The shared recipe totals **167.77B tokens**.
+2. **Specialist SFT and RL:** general translation, instruction following, and meme translation specialists receive task-specific supervision. General translation RL combines XCOMET-XXL with language-validity and adequacy judgments; instruction RL uses Rubric-as-Reward with hard checks and graded constraints. RIVAL provides adaptive judge supervision.
+3. **Expert integration and targeted distillation:** parameter interpolation combines complementary specialists, then multi-teacher on-policy distillation (**MOPD**) addresses task types that remain weak after merging.
+
+The evaluated 2B and 9B models combine the general, instruction, and meme specialists with weights **0.8 / 0.1 / 0.1**, then receive targeted MOPD. Reported training-stage ablations use separate configurations; they are not the final model scores below.
+
+## Evaluation
+
+Results below reproduce the latest technical report. Comparisons apply to the report's evaluation data and settings, and metric scales differ. Higher is better except for off-target rates. These are individual benchmark scores, separate from the demo's normalized radar categories.
+
+### Main translation benchmarks
+
+WMT26 Judge uses a **0–100** scale; the other columns use **0–1**. instTrans quality and IFscore measure translation quality and instruction adherence separately. IFMTBench reports reference-based XCOMET-XXL and instruction adherence. Vertical is the equal-weight mean of the five domain scores below. North-Small-Translate is the **218B-A25B** model; TranslateGemma is **translategemma-12b-it**.
 
 | Model | FLORES COMET-22 | WMT24++ COMET-22 | WMT26 Judge | instTrans Quality | instTrans IFscore | IFMTBench XCOMET-XXL | IFMTBench IFscore | Vertical mean | MEME |
-|---|---|---|---|---|---|---|---|---|---|
-| **Index-Translate-2B** | 0.8655 | 0.8489 | 60.26 | 0.5391 | **0.7569** | 0.7712 | 0.7584 | 0.8377 | **0.6443** |
+|---|---:|---:|---:|---:|---:|---:|---:|---:|---:|
+| 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 |
-| Qwen3.5-2B (base) | 0.6983 | 0.6933 | 32.11 | 0.0999 | 0.2431 | 0.6197 | 0.3836 | 0.7557 | 0.2062 |
 | 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 |
+| 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 |
+
+### Low-resource translation and instruction following
+
+**FLORES_minor_pair** contains **104,000 inputs** across **1,040 directions among 62 languages**. **instTrans_minor** contains **2,793 instruction-translation tasks** from Chinese or English into low-resource languages. Off-target is the percentage of outputs in a language other than the requested target; lower is better. The other scores use a 0–1 scale.
+
+| Model | FLORES_minor_pair COMET-22 ↑ | FLORES_minor_pair XCOMET-XXL ↑ | FLORES_minor_pair off-target ↓ | instTrans_minor Quality ↑ | instTrans_minor IFscore ↑ | instTrans_minor off-target ↓ |
+|---|---:|---:|---:|---:|---:|---:|
+| Index-Translate-35B-A3B (preview) | 0.8168 | 0.7164 | 2.4% | 0.5151 | 0.7715 | 4.05% |
+| Index-Translate-9B | 0.7992 | 0.6805 | 4.0% | 0.5222 | 0.7725 | 3.47% |
+| **Index-Translate-2B** | 0.7377 | 0.4817 | 4.2% | 0.3050 | 0.6586 | 3.97% |
+| Hy-MT2-1.8B | 0.3163 | 0.2014 | 54.2% | 0.0412 | 0.1313 | 65.27% |
+| Hy-MT2-7B | 0.4626 | 0.3334 | 35.7% | 0.1121 | 0.2405 | 45.40% |
+| Hy-MT2-30B-A3B | 0.6746 | 0.5359 | 14.5% | 0.2246 | 0.4449 | 15.47% |
+| TranslateGemma-12B | 0.8021 | 0.6217 | 1.7% | 0.2412 | 0.2682 | 5.37% |
+| North-Small-Translate (218B-A25B) | 0.6855 | 0.4919 | 13.2% | 0.2436 | 0.2902 | 26.89% |
+| Qwen3.5-2B (base) | 0.3716 | 0.1852 | 39.1% | 0.0069 | 0.0549 | 73.54% |
+| Qwen3.5-9B (base) | 0.6774 | 0.4502 | 11.2% | 0.1318 | 0.3532 | 18.51% |
+| Qwen3.5-35B-A3B | 0.7854 | 0.6257 | 3.5% | 0.2874 | 0.4458 | 11.60% |
+| DeepSeek-V4.1-Flash | 0.8333 | 0.7297 | 1.3% | 0.4793 | 0.5854 | 5.73% |
+| GPT-5.6-Sol | 0.7669 | 0.6918 | 10.4% | 0.5757 | 0.6866 | 7.73% |
+| Gemini 3.5 Flash Lite | 0.8122 | 0.6995 | 3.3% | 0.3927 | 0.5584 | 12.18% |
+
+### Domain translation
+
+COMET-22 is used for OPUS-Books, WMT Biomedical, MTNT, and MuST-Cinema; GuoFeng-Webnovel uses document-COMET. Higher is better. The complete comparison is in the report.
 
-The 2B model outperforms Hy-MT2-1.8B on all metrics available for both models.
+| Model | 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 |
 
-Vertical-domain translation (COMET-22; GuoFeng uses document-COMET):
+### General capabilities
 
-| Model | OPUS-Books | Biomedical | MTNT | MuST-Cinema | GuoFeng |
-|---|---|---|---|---|---|
-| **Index-Translate-2B** | **0.7845** | **0.8635** | **0.8132** | **0.8930** | **0.8345** |
-| Hy-MT2-1.8B | 0.7773 | 0.8592 | 0.8080 | 0.8838 | 0.8286 |
+These answer-accuracy results (%) use matched evaluation subsets. Translation training can change general capabilities; these scores describe the translation checkpoints rather than the original Qwen3.5 models.
+
+| Model | 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 |
+| **Index-Translate-2B** | 42.6 | 27.4 | 39.1 | 42.7 |
+
+### Evaluation protocol
+
+- The main FLORES evaluation uses **126,000 inputs**, 420 directions among 21 languages; WMT24++ uses 6,000 inputs from English into 20 languages.
+- instTrans has **3,000 Chinese-to-20-language tasks** across ten constraint types. GPT-5.6-Sol judges quality and instruction adherence separately. MEME contains **3,638 Chinese-to-English cases**, covering 703 terms and 857 disambiguated senses.
+- Text translation evaluation uses **greedy decoding and a 16,384-token context window**, except WMT26, which uses zero-shot Chinese instructions, a **3,584-token context**, and a **1,536-token output limit**. WMT26 reference-free GPT-5.6-Sol scores are averaged over 20 target languages.
+- IFMTBench starts with 7,344 examples. Preprocessing removes 280 single-constraint examples with reference or language-label issues, leaving **7,064 inputs**; **7,063 receive valid judgments**. See the [preprocessing note](https://github.com/bilibili/Index-Translate/blob/main/docs/evaluation.md#ifmtbench-preprocessing).
+
+The **1,024-token output budget in the inference client below is a usage default**, not the output budget used to produce every reported evaluation result. Detailed protocols and valid-judgment counts are in the [technical report](https://github.com/bilibili/Index-Translate/blob/main/docs/Index_Translate_Series_Technical_Report.pdf). The report describes instTrans and MEME; benchmark release plans are tracked in the [GitHub TODO](https://github.com/bilibili/Index-Translate#todo).
+
+## Inference
+
+### Default settings
+
+These settings match the released GitHub text client and apply across Index-Translate sizes:
+
+| Setting | Default |
+|---|---|
+| Decoding | Greedy, **`temperature=0`** |
+| Thinking | Disabled: `enable_thinking=False` |
+| Output budget | `max_tokens=1024` |
+| `top_p`, `top_k`, repetition penalties | Not explicitly set by the client; use inference-server defaults |
+
+The shipped configuration sets `max_position_embeddings=262144`. The examples below use **`--max-model-len 32768`**, matching the GitHub 2B/9B Translate serving preset. Input plus output must fit within the served context, and available GPU memory constrains practical lengths. Increase the client output budget for longer translations.
+
+The GitHub inference guide gives an approximate BF16 GPU-memory starting point of **8 GB for 2B**, with additional memory needed for KV cache and longer contexts.
 
-General capabilities (answer accuracy %): C-Eval 42.6 / GPQA-Diamond 27.4 / INCLUDE 39.1 / MMMLU 42.7.
+### Serve with vLLM
 
-## Usage
+Use a CUDA environment and a vLLM build with Qwen3.5 support. Start the server in one terminal:
 
 ```bash
-vllm serve IndexTeam/Index-Translate-2B --served-model-name Index-Translate-2B --port 8000
+pip install -U vllm
+vllm serve IndexTeam/Index-Translate-2B \
+  --served-model-name IndexTeam/Index-Translate-2B \
+  --host 127.0.0.1 --port 8000 --max-model-len 32768
 ```
 
+For a short-text smoke test, `--max-model-len 4096` reduces the requested context. Multi-GPU deployments can add appropriate tensor-parallel arguments. See the [serving guide](https://github.com/bilibili/Index-Translate/blob/main/inference/llm/README.md) for runtime details and other OpenAI-compatible servers.
+
+### OpenAI-compatible Python client
+
+Install the client dependencies in another terminal:
+
+```bash
+pip install openai httpx
+```
+
+This example uses the training-aligned Chinese translation prompt and disables thinking explicitly:
+
 ```python
+import httpx
 from openai import OpenAI
 
-client = OpenAI(base_url="http://localhost:8000/v1", api_key="EMPTY")
-resp = client.chat.completions.create(
-    model="Index-Translate-2B",
-    messages=[{"role": "user", "content": "Translate the following text into English:\n\n你好，世界！"}],
+client = OpenAI(
+    base_url="http://127.0.0.1:8000/v1",
+    api_key="EMPTY",
+    http_client=httpx.Client(trust_env=False),
+)
+response = client.chat.completions.create(
+    model="IndexTeam/Index-Translate-2B",
+    messages=[{
+        "role": "user",
+        "content": "请将以下文本翻译为英语，直接输出翻译结果，不要进行任何解释。\n\n你好，世界！",
+    }],
     temperature=0,
+    max_tokens=1024,
+    extra_body={"chat_template_kwargs": {"enable_thinking": False}},
 )
-print(resp.choices[0].message.content)
+text = response.choices[0].message.content or ""
+# Match the repository client's fallback if a thinking block is emitted.
+if "</think>" in text:
+    text = text.split("</think>", 1)[1]
+print(text.strip().removeprefix("<think>").strip())
 ```
 
-Transformers and SGLang quickstarts: [inference/llm](https://github.com/bilibili/Index-Translate/tree/main/inference/llm).
+### Repository client
+
+After starting the server, use the public client with this model ID:
+
+```bash
+git clone https://github.com/bilibili/Index-Translate.git
+cd Index-Translate
+pip install -r inference/llm/requirements.txt
+python inference/llm/translate.py \
+  "你好，世界。今天天气不错。" --target en \
+  --model IndexTeam/Index-Translate-2B
+```
+
+Use `--source` to specify the input language and `--target` for the output language. For languages without a mapped code in the client, provide the full language name. The client also accepts `--base-url`, `--api-key`, `--max-tokens`, and `--temperature`.
+
+## Examples and practical limits
+
+The following recorded Index-Translate-2B case from the official demo illustrates a community expression:
+
+| Source and context | Recorded English translation |
+|---|---|
+| 狒瘾犯了就去打 — “狒瘾” refers to the urge to play Final Fantasy XIV | Go play when my FFXIV addiction kicks in. |
+
+See the [captured inference cases](https://github.com/bilibili/Index-Translate/blob/main/inference/llm/cases/translate_cases.jsonl) and [prompt examples](https://github.com/bilibili/Index-Translate/blob/main/docs/prompts.md) for more directions and instruction formats.
+
+Scores vary across languages and tasks. Combined constraints and low-resource directions remain harder than core-language translation, and successful individual examples do not guarantee that every instruction will be followed. For explicit target syllable counts, use Index-Homura; for native full-document translation, use Index-NativeLong with its dedicated templates.
 
 ## Model family
 
-| Model | Sizes | Task |
+| Family | Models | Task |
 |---|---|---|
-| **Index-Translate** | [2B](https://huggingface.co/IndexTeam/Index-Translate-2B) / [9B](https://huggingface.co/IndexTeam/Index-Translate-9B) | Multilingual text translation & translation instructions |
-| Index-Echo S2TT | [2B](https://huggingface.co/IndexTeam/Index-Echo-S2TT-2B) / [9B](https://huggingface.co/IndexTeam/Index-Echo-S2TT-9B) | Speech-to-text translation |
-| Index-Echo S2ST | [2B](https://huggingface.co/IndexTeam/Index-Echo-S2ST-2B) / [9B](https://huggingface.co/IndexTeam/Index-Echo-S2ST-9B) | Speech-to-speech translation with voice conditioning |
-| Index-Homura | [2B](https://huggingface.co/IndexTeam/Index-Homura-2B) / [9B](https://huggingface.co/IndexTeam/Index-Homura-9B) | Syllable-controlled translation (dubbing) |
-| Index-NativeLong (NaiLong) | [2B](https://huggingface.co/IndexTeam/Index-Nailong-2B) / [9B](https://huggingface.co/IndexTeam/Index-Nailong-9B) | Long-document translation |
+| Index-Translate | [2B](https://huggingface.co/IndexTeam/Index-Translate-2B) / [9B](https://huggingface.co/IndexTeam/Index-Translate-9B) / [35B-A3B (preview)](https://huggingface.co/IndexTeam/Index-Translate-35B-A3B) | Text translation and translation instructions |
+| Index-Echo S2TT | [2B](https://huggingface.co/IndexTeam/Index-Echo-S2TT-2B) / [9B](https://huggingface.co/IndexTeam/Index-Echo-S2TT-9B) | Speech to translated subtitles |
+| Index-Echo S2ST | [2B](https://huggingface.co/IndexTeam/Index-Echo-S2ST-2B) / [9B](https://huggingface.co/IndexTeam/Index-Echo-S2ST-9B) | Speech to speech with source-voice conditioning |
+| Index-Homura | [2B](https://huggingface.co/IndexTeam/Index-Homura-2B) / [9B](https://huggingface.co/IndexTeam/Index-Homura-9B) | Translation toward a target syllable count |
+| Index-NativeLong | [2B](https://huggingface.co/IndexTeam/Index-Nailong-2B) / [9B](https://huggingface.co/IndexTeam/Index-Nailong-9B) | Full-document translation |
+
+**Naming:** Index-NativeLong retains the published model IDs `IndexTeam/Index-Nailong-2B` and `IndexTeam/Index-Nailong-9B`. The [family repository](https://github.com/bilibili/Index-Translate) includes prompt guides, browser-extension code, and a video-dubbing pipeline. This model's [ModelScope page](https://www.modelscope.cn/models/IndexTeam/Index-Translate-2B) mirrors the model card.
 
 ## Citation
 
 ```bibtex
 @techreport{indextranslate2026,
+  author={Tianjiao Li and Mengran Yu and Chenyu Shi and Lusheng Zhang and
+          Qisi Chen and Yanshan Zhou and Ji Qi and Jingying Liu and
+          Yuang Feng and Ziang Cui and Tianxing Yan},
   title={Index-Translate: A Multilingual Translation Model Family --- Text, Speech, Controlled Dubbing, and Long-Document Translation},
-  author={Index LLM Team},
+  institution={Index LLM Team},
   year={2026},
-  month={September}
+  month={September},
+  url={https://github.com/bilibili/Index-Translate/blob/main/docs/Index_Translate_Series_Technical_Report.pdf}
 }
 ```
 
 ## License
 
-Apache-2.0.
+[Apache-2.0](https://github.com/bilibili/Index-Translate/blob/main/LICENSE). Questions and feedback are welcome through [GitHub Issues](https://github.com/bilibili/Index-Translate/issues).