README.md

---
base_model: google/gemma-2-2B
license: cc-by-nc-sa-4.0
language:
- de
- nl
- is
- es
- fr
- pt
- uk
- hi
- zh
- ru
- cs
- ko
- ja
- it
- en
- da
- pl
- hu
- sv
- 'no'
- ro
- fi
library_name: transformers
---

![Tower Plus Pareto](./Tower-plus-pareto.png)

# Model Description:

**Tower+ 2B** is build on top of Gemma 2 2B. The model goes through the Continuous Pretraining (CPT), Instruction Tuning (IT), Weighted Preference Optimization (WPO) and GRPO with verifiable rewards. During all stages we include parallel and multilingual data (covering 22 languages).

This approach makes Tower+ 2B one of the best multilingual LLMs under 3B parameters.

- **Developed by:** Unbabel
- **Model type:** A 2B parameter model fine-tuned on a mix of _translation-related tasks_ as well as  _general instruction-following_ datasets that include reasoning, code instructions, etc.
- **Languages:** German, Spanish, French, Italian, Korean, Dutch, Russian, English, Portuguese (Portugal), Portuguese (Brazilian), Spanish (Latin America), Chinese (Simplified), Chinese (Traditional), Czech, Ukrainian, Hindi, Icelandic, Japanese, Polish, Swedish, Hungarian, Romanian, Danish, Norwegian (Nynorsk), Norwegian (Bokmål), Finnish
- **License:** CC-BY-NC-4.0
- **Context Size:**: 8192 tokens

# Intended uses & limitations

Tower is intended for multilingual tasks and its specially strong on machine translation. 

Because Tower is also a strong multilingual model you can also use it for other multilingual tasks. 

Another usecase Tower works well is for creating multilingual synthethic data (for the languages it covers). You can do this either by translating instructions and the respective answers or by asking the model to create an instruction given a document as seed data.

# Usage:

When using the model, make sure your prompt is formated correctly! 

Also, we recommend using VLLM rather than Hugging Face.

### Using on VLLM:

```python
# pip install vllm
# Gemma by default only uses 4k context. You need to set the following variables:
# export VLLM_WORKER_MULTIPROC_METHOD=spawn
# export VLLM_ALLOW_LONG_MAX_MODEL_LEN=1

from vllm import LLM, SamplingParams

sampling_params = SamplingParams(
  best_of=1,
  temperature=0,
  max_tokens=8192,
)
llm = LLM(model="Unbabel/Tower-Plus-2B", tensor_parallel_size=1)
messages = [{"role": "user", "content": "Translate the following English source text to Portuguese (Portugal):\nEnglish: Hello world!\nPortuguese (Portugal): "}]
outputs = llm.chat(messages, sampling_params)
# Make sure your prompt_token_ids look like this
print (outputs[0].outputs[0].text)
# > Olá, mundo!
```

### Using on Transformers:

```python
# Install transformers from source - only needed for versions <= v4.34
# pip install git+https://github.com/huggingface/transformers.git
# pip install accelerate
import torch
from transformers import pipeline

pipe = pipeline("text-generation", model="Unbabel/Tower-Plus-2B", device_map="auto")
# We use the tokenizer’s chat template to format each message - see https://huggingface.co/docs/transformers/main/en/chat_templating
messages = [{"role": "user", "content": "Translate the following English source text to Portuguese (Portugal):\nEnglish: Hello world!\nPortuguese (Portugal): "}]
input_ids = pipe.tokenizer.apply_chat_template(messages, tokenize=True, add_generation_prompt=True)
outputs = pipe(messages, max_new_tokens=256, do_sample=False)
print(outputs[0]["generated_text"])
```

# Citation
If you use this model please cite our paper:
```
@misc{rei2025towerplus,
      title={Tower+: Bridging Generality and Translation Specialization in Multilingual LLMs}, 
      author={Ricardo Rei and Nuno M. Guerreiro and José Pombal and João Alves and Pedro Teixeirinha and Amin Farajian and André F. T. Martins},
      year={2025},
      eprint={2506.17080},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2506.17080}, 
}
```