This model is a fine-tuned version of meta-llama/Llama-3.2-1B, trained for bidirectional translation between English and Mandarin. It supports both:

🇬🇧 English → 🇨🇳 Mandarin
🇨🇳 Mandarin → 🇬🇧 English

Format

The model expects prompts in the following format:

### English:
The children were playing in the park.
### Mandarin:

### Mandarin:
孩子们在公园里玩耍。
### English:

Usage

python

from transformers import AutoModelForCausalLM, AutoTokenizer

model = AutoModelForCausalLM.from_pretrained("Sheikhaei/llama-3.2-1b-en-zh-translator", torch_dtype="auto", device_map="auto")
tokenizer = AutoTokenizer.from_pretrained("Sheikhaei/llama-3.2-1b-en-zh-translator")

prompt = """### English:
The children were playing in the park.
### Mandarin:
"""

inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
outputs = model.generate(**inputs, max_new_tokens=100, do_sample=False)
print(tokenizer.decode(outputs[0], skip_special_tokens=True))

Training Data

This model was fine-tuned on a custom English–Mandarin parallel dataset containing ~640,000 sentence pairs. The source data was collected from Tatoeba and then translated and expanded using the Gemma-3-12B model.

Evaluation

Direction	BLEU	COMET
English → Mandarin	0.47	0.87
Mandarin → English	0.44	0.89

License

Apache 2.0