You can deploy this model and start earning money today!

lengocquangLAB/
phobert-large-job-title-match

safetensors

roberta

Views

No views yet

PhoBERT Large Fine-tuned for Job Title Matching

This model is a PhoBERT-Large fine-tuned with LoRA for sequence classification.
It is designed to predict if 2 job titles mention the same position.

Training Details

Base model: vinai/phobert-large
Task: Sequence Classification (2 labels: match / no match)
LoRA config: r=16, lora_alpha=32, target_modules=["query", "key", "value", "output.dense"], lora_dropout=0.05
Optimizer: AdamW, learning rate 2e-4
Batch size: 64 (with gradient accumulation 4)
Epochs: 8
Weight decay: 0.01
Mixed precision: fp16
Metrics: F1-score
Framework: HuggingFace Transformers + PEFT
Logging & tracking: WandB (report_to=["wandb"])

Usage

python

from transformers import AutoTokenizer, AutoModelForSequenceClassification

# Load tokenizer and model from Hugging Face Hub
tokenizer = AutoTokenizer.from_pretrained("lengocquangLAB/phobert-large-job-title-match")
model = AutoModelForSequenceClassification.from_pretrained("lengocquangLAB/phobert-large-job-title-match")

# Example input: Job title vs Job title
inputs = tokenizer("AI Engineer", "Machine Learning Engineer", return_tensors="pt")
outputs = model(**inputs)

# Prediction: 0 = no match, 1 = match
pred = outputs.logits.argmax(dim=-1)
print(pred)