216 lines
8.0 KiB
Markdown
216 lines
8.0 KiB
Markdown
---
|
|
license: apache-2.0
|
|
language:
|
|
- ar
|
|
- bg
|
|
- bn
|
|
- ca
|
|
- cs
|
|
- da
|
|
- de
|
|
- el
|
|
- es
|
|
- et
|
|
- fa
|
|
- fi
|
|
- fil
|
|
- fr
|
|
- gu
|
|
- he
|
|
- hi
|
|
- hr
|
|
- hu
|
|
- id
|
|
- is
|
|
- it
|
|
- ja
|
|
- kn
|
|
- ko
|
|
- lt
|
|
- lv
|
|
- ml
|
|
- mr
|
|
- nl
|
|
- 'no'
|
|
- pa
|
|
- pl
|
|
- pt
|
|
- ro
|
|
- ru
|
|
- sk
|
|
- sl
|
|
- sr
|
|
- sv
|
|
- sw
|
|
- ta
|
|
- te
|
|
- th
|
|
- tr
|
|
- uk
|
|
- ur
|
|
- vi
|
|
- zh
|
|
- zu
|
|
base_model:
|
|
- winninghealth/WiNGPT-Babel-2
|
|
tags:
|
|
- GGUF
|
|
- multilingual
|
|
datasets:
|
|
- google/wmt24pp
|
|
pipeline_tag: translation
|
|
library_name: transformers
|
|
---
|
|
|
|
# WiNGPT-Babel-2: A Multilingual Translation Language Model
|
|
|
|
[](https://huggingface.co/collections/winninghealth/wingpt-babel-68463d4b2a28d0d675ff3be9)
|
|
[](https://opensource.org/licenses/Apache-2.0)
|
|
|
|
> This is the quantization version (llama.cpp) of [WiNGPT-Babel-2](https://huggingface.co/winninghealth/WiNGPT-Babel-2).
|
|
>
|
|
> Example
|
|
>
|
|
> ```shell
|
|
> ./llama-server -m WiNGPT-Babel-2-GGUF/WiNGPT-Babel-2-IQ4_XS.gguf --jinja --chat-template-file WiNGPT-Babel-2-GGUF/WiNGPT-Babel-2.jinja
|
|
> ```
|
|
>
|
|
> - **--jinja**: This flag activates the Jinja2 chat template processor.
|
|
> - **--chat-template-file**: This flag points the server to the required template file that defines the WiNGPT-Babel-2's custom prompt format.
|
|
|
|
WiNGPT-Babel-2 is a language model optimized for multilingual translation tasks. As an iteration of WiNGPT-Babel, it features significant improvements in language coverage, data format handling, and translation accuracy for complex content.
|
|
|
|
The model continues the "Human-in-the-loop" training strategy, iteratively optimizing through the analysis of log data from real-world application scenarios to ensure its effectiveness and reliability in practical use.
|
|
|
|
## Core Improvements in Version 2.0
|
|
|
|
WiNGPT-Babel-2 introduces the following key technical upgrades over its predecessor:
|
|
|
|
1. **Expanded Language Support:** Through training with the `wmt24pp` dataset, language support has been extended to **55 languages**, primarily enhancing translation capabilities from English (en) to other target languages (xx).
|
|
|
|
2. **Enhanced Chinese Translation:** The translation pipeline from other source languages to Chinese (xx → zh) has been specifically optimized, improving the accuracy and fluency of the results.
|
|
|
|
3. **Structured Data Translation:** The model can now identify and translate text fields embedded within **structured data (e.g., JSON)** while preserving the original data structure. This feature is suitable for scenarios such as API internationalization and multilingual dataset preprocessing.
|
|
|
|
4. **Mixed-Content Handling:** Its ability to handle mixed-content text has been improved, enabling more accurate translation of paragraphs containing **mathematical expressions (LaTeX), code snippets, and web markup (HTML/Markdown)**, while preserving the format and integrity of these non-translatable elements.
|
|
|
|
## Training Methodology
|
|
|
|
The performance improvements in WiNGPT-Babel-2 are attributed to a continuous, data-driven, iterative training process:
|
|
|
|
1. **Data Collection:** Collecting anonymous, real-world translation task logs from integrated applications (e.g., Immersive Translate, Videolingo).
|
|
2. **Data Refinement:** Using a reward model for rejection sampling on the collected data, supplemented by manual review, to filter high-quality, high-value samples for constructing new training datasets.
|
|
3. **Iterative Retraining:** Using the refined data for the model's incremental training, continuously improving its performance in specific domains and scenarios through a cyclical iterative process.
|
|
|
|
## Technical Specifications
|
|
|
|
* **Base Model:** [GemmaX2-28-2B-Pretrain](https://huggingface.co/ModelSpace/GemmaX2-28-2B-Pretrain)
|
|
* **Primary Training Data:** "Human-in-the-loop" in-house dataset, [WMT24++](https://huggingface.co/datasets/google/wmt24pp) dataset
|
|
* **Maximum Context Length:** 4096 tokens
|
|
* **Chat Capability:** Supports multi-turn dialogue, allowing for contextual follow-up and translation refinement.
|
|
|
|
## Language Support
|
|
|
|
| Direction | Description | Supported Languages (Partial List) |
|
|
| :---------------------- | :--------------------------------------------------- | :----------------------------------------------------------- |
|
|
| **Core Support** | Highest quality, extensively optimized. | `en ↔ zh` |
|
|
| **Expanded Support** | Supported via `wmt24pp` dataset training. | `en → 55+ languages`, including: `fr`, `de`, `es`, `ru`, `ar`, `pt`, `ko`, `it`, `nl`, `tr`, `pl`, `sv`... |
|
|
| **Enhanced to Chinese** | Specifically optimized for translation into Chinese. | `xx → zh` |
|
|
|
|
## Performance
|
|
<table>
|
|
<thead>
|
|
<tr>
|
|
<th rowspan="2" align="center">Model</th>
|
|
<th colspan="2" align="center">FLORES-200</th>
|
|
</tr>
|
|
<tr>
|
|
<th align="center">xx → en</th>
|
|
<th align="center">xx → zh</th>
|
|
</tr>
|
|
</thead>
|
|
<tbody>
|
|
<tr>
|
|
<td align="center">WiNGPT-Babel-AWQ</td>
|
|
<td align="center">33.91</td>
|
|
<td align="center">17.29</td>
|
|
</tr>
|
|
<tr>
|
|
<td align="center">WiNGPT-Babel-2-AWQ</td>
|
|
<td align="center">36.43</td>
|
|
<td align="center">30.74</td>
|
|
</tr>
|
|
</tbody>
|
|
</table>
|
|
|
|
**Note**:
|
|
1. The evaluation metric is spBLEU, using the FLORES-200 tokenizer.
|
|
|
|
3. 'xx' represents the 52 source languages from the wmt24pp dataset.
|
|
|
|
## Usage Guide
|
|
|
|
For optimal inference performance, it is recommended to use frameworks such as `vllm`. The following provides a basic usage example using the Hugging Face `transformers` library.
|
|
|
|
**System Prompt:** For optimal automatic language inference, it is recommended to use the unified system prompt: `Translate this to {{to}} Language`. Replace `{{to}}` with the name of the target language. For instance, use `Translate this to Simplified Chinese Language` to translate into Chinese, or `Translate this to English Language` to translate into English. This method provides precise control over the translation direction and yields the most reliable results.
|
|
|
|
### Example
|
|
|
|
```python
|
|
from transformers import AutoModelForCausalLM, AutoTokenizer
|
|
|
|
model_name = "winninghealth/WiNGPT-Babel-2-AWQ"
|
|
|
|
model = AutoModelForCausalLM.from_pretrained(
|
|
model_name,
|
|
torch_dtype="auto",
|
|
device_map="auto"
|
|
)
|
|
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
|
|
|
# Example: Translation of text within a JSON object to Chinese
|
|
prompt_json = """{
|
|
"product_name": "High-Performance Laptop",
|
|
"features": ["Fast Processor", "Long Battery Life", "Lightweight Design"]
|
|
}"""
|
|
|
|
messages = [
|
|
{"role": "system", "content": "Translate this to Simplified Chinese Language"},
|
|
{"role": "user", "content": prompt_json} # Replace with the desired prompt
|
|
]
|
|
|
|
text = tokenizer.apply_chat_template(
|
|
messages,
|
|
tokenize=False,
|
|
add_generation_prompt=True
|
|
)
|
|
model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
|
|
|
|
generated_ids = model.generate(
|
|
**model_inputs,
|
|
max_new_tokens=4096,
|
|
temperature=0
|
|
)
|
|
|
|
generated_ids = [
|
|
output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
|
|
]
|
|
|
|
response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
|
|
```
|
|
|
|
For additional usage demos, you can refer to the original [WiNGPT-Babel](https://huggingface.co/winninghealth/WiNGPT-Babel#%F0%9F%8E%AC-%E7%A4%BA%E4%BE%8B).
|
|
|
|
## LICENSE
|
|
|
|
1. This project's license agreement is the Apache License 2.0
|
|
|
|
2. Please cite this project when using its model weights: https://huggingface.co/winninghealth/WiNGPT-Babel-2
|
|
|
|
3. Comply with [gemma-2-2b](https://huggingface.co/google/gemma-2-2b), [GemmaX2-28-2B-v0.1](https://huggingface.co/ModelSpace/GemmaX2-28-2B-v0.1), [immersive-translate](https://github.com/immersive-translate/immersive-translate), [VideoLingo](https://github.com/immersive-translate/immersive-translate) protocols and licenses, details on their website.
|
|
|
|
|
|
## Contact Us
|
|
|
|
- Apply for a token through the WiNGPT platform
|
|
- Or contact us at wair@winning.com.cn to request a free trial API_KEY |