billyskc/mbart-lora-paramed-merged

Model Card for billyskc/mbart-lora-zh-en-paramed

This is a LoRA-parameter-efficient fine-tuned version of facebook/mbart-large-50-many-to-many-mmt for English-to-Chinese biomedical translation tasks.

Model Details

Model Description

This model adapts mBART-50 for the biomedical domain using LoRA (Low-Rank Adaptation) to reduce fine-tuning memory and cost. The model was trained on curated biomedical English-to-Chinese parallel data.

• Developed by: Shengkang Chen
• Funded by [optional]: [More Information Needed]
• Shared by [optional]: [More Information Needed]
• Model type: Seq2Seq transformer (mBART)
• Language(s) (NLP): English (en_XX) → Chinese (zh_CN)
• License: Apache-2.0 (or match base model license)
• Finetuned from model [optional]: facebook/mbart-large-50-many-to-many-mmt

Model Sources [optional]

• Repository: https://huggingface.co/spaces/billyskc/mbart-lora-paramed
• Paper [optional]: [More Information Needed]
• Demo [optional]: https://huggingface.co/spaces/billyskc/mbart-lora-paramed

Uses

Direct Use

Biomedical domain translation (EN → ZH), particularly for medical research abstracts, clinical sentences, and drug descriptions.

Downstream Use [optional]

[More Information Needed]

Out-of-Scope Use

General domain translation (performance may degrade)

Chinese-to-English (ZH → EN)

Real-time or low-latency applications without testing

Bias, Risks, and Limitations

The model may reflect biases from the biomedical training data.

Mistranslations may cause harm in clinical or decision-critical use cases.

LoRA fine-tuning may limit generalization outside biomedical domain.

Recommendations

Always validate outputs with domain experts.

Avoid use in diagnostic or treatment recommendation systems without human oversight.

How to Get Started with the Model

from transformers import MBartForConditionalGeneration, MBart50Tokenizer

model = MBartForConditionalGeneration.from_pretrained("billyskc/mbart-lora-paramed-merged")
tokenizer = MBart50Tokenizer.from_pretrained("billyskc/mbart-lora-zh-en-tokenizer")

tokenizer.src_lang = "en_XX"
text = "Asciminib targets both native and mutated BCR-ABL1"
inputs = tokenizer(text, return_tensors="pt")
generated_tokens = model.generate(**inputs, forced_bos_token_id=tokenizer.lang_code_to_id["zh_CN"])
print(tokenizer.decode(generated_tokens[0], skip_special_tokens=True))

Training Details

Training Data

[More Information Needed]

Training Procedure

Preprocessing [optional]

[More Information Needed]

Training Hyperparameters

• Training regime: [More Information Needed]

Speeds, Sizes, Times [optional]

[More Information Needed]

Evaluation

Testing Data, Factors & Metrics

Testing Data

[More Information Needed]

Factors

[More Information Needed]

Metrics

[More Information Needed]

Results

[More Information Needed]

Summary

Model Examination [optional]

[More Information Needed]

Environmental Impact

Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019).

• Hardware Type: [More Information Needed]
• Hours used: [More Information Needed]
• Cloud Provider: [More Information Needed]
• Compute Region: [More Information Needed]
• Carbon Emitted: [More Information Needed]

Technical Specifications [optional]

Model Architecture and Objective

[More Information Needed]

Compute Infrastructure

[More Information Needed]

Hardware

[More Information Needed]

Software

[More Information Needed]

Citation [optional]

BibTeX:

[More Information Needed]

APA:

[More Information Needed]

Glossary [optional]

[More Information Needed]

More Information [optional]

[More Information Needed]

Model Card Authors [optional]

[More Information Needed]

Model Card Contact

[More Information Needed]