WangchanBERTa Base for Sentiment Analysis
This is a fine-tuned version of the WangchanBERTa model, trained for sentiment analysis in Thai language using simpletransformers.
Model Details
- Model Name: WangchanBERTa Base Sentiment Analysis
- Pretrained Base Model:
airesearch/wangchanberta-base-att-spm-uncased - Architecture: CamemBERT
- Language: Thai
- Task: Sentiment Classification
Training Configuration
- Training Dataset: (e.g., your dataset name or a public dataset if applicable)
- Number of Training Epochs: 6
- Train Batch Size: 16
- Eval Batch Size: 32
- Learning Rate: 2e-5
- Optimizer: AdamW
- Scheduler: Cosine
- Gradient Accumulation Steps: 2
- Seed: 42
- Training Framework:
simpletransformers - FP16: Disabled
Model Performance
Provide any performance metrics here, such as accuracy, F1-score, etc., depending on your dataset.
Usage
To use this model, you can load it as follows:
from transformers import AutoTokenizer, AutoModelForSequenceClassification
import torch
import torch.nn.functional as F
import numpy as np
from pythainlp.tokenize import word_tokenize
tokenizer = AutoTokenizer.from_pretrained("Pongsathorn/wangchanberta-base-sentiment")
model = AutoModelForSequenceClassification.from_pretrained("Pongsathorn/wangchanberta-base-sentiment")
id2label = {
0: "pos",
1: "neu",
2: "neg",
}
input_text = "พนักงานบริการดีมาก สัญญาณก็ดี แต่ร้านอยู่ที่ไหน อยากได้ข้อมูลเพิ่มเติม จะได้ประกาศบนเว็บถูก"
segmented_text = word_tokenize(input_text, engine="longest")
preprocessed_text = " ".join(segmented_text)
inputs = tokenizer(preprocessed_text, return_tensors="pt", padding=True, truncation=True)
with torch.no_grad():
outputs = model(**inputs)
logits = outputs.logits
probs = F.softmax(logits, dim=-1)
predicted_class = torch.argmax(probs, dim=-1).item()
predicted_label = id2label[predicted_class]
print("Predicted Label (ID):", predicted_class)
print("Predicted Label (Description):", predicted_label)
max_prob = np.max(probs.numpy())
print(f"Maximum Probability: {max_prob:.4f}")