nlp-prachathai67k-text-classification

This model is a fine-tuned version of meta-llama/Llama-3.2-1B on the prachathai67k dataset. It achieves the following results on the evaluation set:

• Loss: 0.1609
• Accuracy: 0.9349
• F1: 0.7418
• Precision: 0.8033
• Recall: 0.6890

Labels

id2label = {
    0:  'politics',
    1:  'human_rights',
    2:  'quality_of_life',
    3:  'international',
    4:  'social',
    5:  'environment',
    6:  'economics',
    7:  'culture',
    8:  'labor',
    9:  'national_security',
    10: 'ict',
    11: 'education'
}
label2id = {
    'politics':          0,
    'human_rights':      1,
    'quality_of_life':   2,
    'international':     3,
    'social':            4,
    'environment':       5,
    'economics':         6,
    'culture':           7,
    'labor':             8,
    'national_security': 9,
    'ict':               10,
    'education':         11
}

Usage

With pipeline API

from peft import PeftModel, PeftConfig
from transformers import AutoModelForSequenceClassification, AutoTokenizer, pipeline

model_id = "tonkaew131/nlp-prachathai67k-text-classification"
config = PeftConfig.from_pretrained(model_id)

tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForSequenceClassification.from_pretrained(
    config.base_model_name_or_path,
    problem_type="multi_label_classification",
    num_labels=12,
    id2label=id2label,
    label2id=label2id
)
model.resize_token_embeddings(len(tokenizer))

lora_model = PeftModel.from_pretrained(model, model_id)
classifier = pipeline(
    "text-classification",
    model=lora_model,
    tokenizer=tokenizer,
    top_k=None
)

text = "<news-content>"
results = classifier(text)
print(results)

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 3
• eval_batch_size: 3
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 12
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.1687	0.9998	4531	0.1698	0.9302	0.7251	0.7787	0.6784
0.1513	1.9997	9062	0.1609	0.9349	0.7418	0.8033	0.6890

Framework versions

• PEFT 0.13.0
• Transformers 4.44.2
• Pytorch 2.4.0
• Datasets 3.0.0
• Tokenizers 0.19.1