This model is a fine-tuned sequence-to-sequence model based on Helsinki-NLP/opus-mt-zh-en for Chinese-to-English translation of Wuxia and Xianxia literature. It was developed as part of the Bachelor's Thesis (TFG):

"Enfoques de traducción automática con modelos de lenguaje en obras wuxia" (Degree in Data Science and Engineering, Universidade da Coruña)

Model Details

Base Model: Helsinki-NLP/opus-mt-zh-en
Training Method: Full architectural fine-tuning
Training Dataset: HSilvosa/hsc-wuxia-100k (100,000 stratified parallel sentence pairs)
Target Domain: Wuxia/Xianxia web novels

Intended Use

This model can be used directly for Chinese-to-English translation of literary text in the Wuxia domain.

Quick Start (Inference)

python

from transformers import MarianMTModel, MarianTokenizer

model_name = "HSilvosa/hsc-wuxia-marianmt-zh-en"
model = MarianMTModel.from_pretrained(model_name)
tokenizer = MarianTokenizer.from_pretrained(model_name)

text = "一念成沧海，一念化桑田。"
inputs = tokenizer(text, return_tensors="pt")
generated_tokens = model.generate(**inputs)
print(tokenizer.batch_decode(generated_tokens, skip_special_tokens=True)[0])

Evaluation Results

The model was evaluated on the Wuxia parallel test set, comparing the base model and the fine-tuned version:

Metric	Base Model (`opus-mt-zh-en`)	Fine-tuned Model (`hsc-wuxia-marianmt-zh-en`)	Delta (Improvement)
SacreBLEU	5.12	28.39	+23.27
chrF	23.56	50.02	+26.46
ROUGE-L	27.80	57.40	+29.60
METEOR	23.71	56.55	+32.84
COMET	59.06	77.57	+18.51

Training Hyperparameters

The model was fine-tuned using the following architectural training parameters:

Optimizer: AdamW
Learning Rate: 2e-5 (0.00002)
Weight Decay: 0.01
Epochs: Max 10 (with early stopping patience of 3 epochs based on validation loss)
Batch Size: 16 (per device/GPU)
Max Sequence Length: 128 tokens
Seed: 42

Project Repository

This model and the associated training/evaluation pipelines were developed in the following project repository:

GitHub Repository: HugoSilvosa/hsc-mtl-wuxia

hsilvosa/hsc-wuxia-marianmt-zh-en

HSC Wuxia MarianMT zh-en Model