phobert-absa-smartphone

Fine‑tuned from vinai/phobert-base on visolex/phobert-absa-smartphone for joint aspect detection + sentiment classification (shared heads).

Model Details

• Base Model: vinai/phobert-base
• Dataset: visolex/ViSFD
• Fine‑tuning framework: HuggingFace Transformers

Hyperparameters

• Batch size: 32
• Learning rate: 3e‑5
• Epochs: 100
• Max sequence length: 256
• Early stopping patience: 5

Usage

import torch
from transformers import AutoTokenizer, AutoModel

# Danh sách aspect và sentiment labels
aspect_labels = [
    "BATTERY", "CAMERA", "DESIGN", "FEATURES", "GENERAL",
    "PERFORMANCE", "PRICE", "SCREEN", "SERandACC", "STORAGE"
]
sentiment_labels = ["POSITIVE", "NEGATIVE", "NEUTRAL"]

# 1) Load tokenizer và model (phải về đúng class TransformerForABSA)
repo = "visolex/phobert-absa-smartphone"
tokenizer = AutoTokenizer.from_pretrained(repo, trust_remote_code=True)
model     = AutoModel.from_pretrained(repo, trust_remote_code=True)
model.eval()

def predict_absa_multi(
    text: str,
    aspect_labels: list[str],
    sentiment_labels: list[str],
    threshold: float = 0.5
) -> list[tuple[str,str]]:
    inputs = tokenizer(
        text,
        return_tensors="pt",
        padding=True,
        truncation=True,
        max_length=256
    )
    inputs.pop("token_type_ids", None)

    with torch.no_grad():
        out = model(**inputs)

    # out.logits có shape [1, A, S+1]
    logits = out.logits.squeeze(0)          # [A, S+1]
    probs  = torch.softmax(logits, dim=-1)  # [A, S+1]

    num_s  = len(sentiment_labels)
    none_id = probs.size(-1) - 1            # chỉ số của lớp "none"
    results = []

    for i, asp in enumerate(aspect_labels):
        prob_i = probs[i]
        pred_id = int(prob_i.argmax().item())

        if pred_id != none_id and pred_id < num_s:
            score = prob_i[pred_id].item()
            if score >= threshold:
                results.append((asp, sentiment_labels[pred_id].lower()))

    return results


text = "mới mua được một tuần pin bốn nghìn mà quá tệ cảm ứng hơi đơ nhận sim bị lỗi."
preds = predict_absa_multi(text, aspect_labels, sentiment_labels, threshold=0.2)
print(preds)
# ➔ [('BATTERY','negative'), ('PERFORMANCE','negative'), ...]