README.md

metadata

base_model: airesearch/wangchanberta-base-att-spm-uncased
tags:
  - generated_from_trainer
datasets:
  - universal_dependencies
metrics:
  - accuracy
  - recall
  - precision
  - f1
model-index:
  - name: wangchanberta-ud-thai-pud-upos
    results:
      - task:
          name: Token Classification
          type: token-classification
        dataset:
          name: universal_dependencies
          type: universal_dependencies
          config: th_pud
          split: test
          args: th_pud
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.9883334914161055
widget:
  - text: >-
      นักวิจัยกล่าวว่าการวิเคราะห์ดีเอ็นเอของเนื้องอกอาจช่วยอธิบายถึงสาเหตุที่แท้จริงของมะเร็งชนิดอื่นๆ
      ได้
    example_title: test_example_1
  - text: คือผมไม่ได้ชอบกดดันพวกคุณหรอกนะ แต่ชะตากรรมของสาธารณรัฐอยู่ในกำมือคุณ
    example_title: test_example_2
language:
  - th
library_name: transformers

wangchanberta-ud-thai-pud-upos

This model is a fine-tuned version of airesearch/wangchanberta-base-att-spm-uncased on the universal_dependencies dataset. It achieves the following results on the evaluation set:

• Loss: 0.0442
• Macro avg precision: 0.9221
• Macro avg recall: 0.9178
• Macro avg f1: 0.9199
• Weighted avg precision: 0.9883
• Weighted avg recall: 0.9883
• Weighted avg f1: 0.9883
• Accuracy: 0.9883

Model description

This model is train on thai UD Thai PUD corpus with Universal Part-of-speech (UPOS) tag to help with pos tagging in Thai language.

Example

from transformers import AutoModelForTokenClassification, AutoTokenizer, TokenClassificationPipeline

model = AutoModelForTokenClassification.from_pretrained("Pavarissy/wangchanberta-ud-thai-pud-upos")
tokenizer = AutoTokenizer.from_pretrained("Pavarissy/wangchanberta-ud-thai-pud-upos")

pipeline = TokenClassificationPipeline(model=model, tokenizer=tokenizer, grouped_entities=True)
outputs = pipeline("ประเทศไทย อยู่ใน ทวีป เอเชีย")
print(outputs)
# [{'entity_group': 'NOUN', 'score': 0.419697, 'word': '', 'start': 0, 'end': 1}, {'entity_group': 'PROPN', 'score': 0.8809489, 'word': 'ประเทศไทย', 'start': 0, 'end': 9}, {'entity_group': 'VERB', 'score': 0.7754166, 'word': 'อยู่ใน', 'start': 9, 'end': 16}, {'entity_group': 'NOUN', 'score': 0.9976932, 'word': 'ทวีป', 'start': 16, 'end': 21}, {'entity_group': 'PROPN', 'score': 0.97770107, 'word': 'เอเชีย', 'start': 21, 'end': 28}]

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Macro avg precision	Macro avg recall	Macro avg f1	Weighted avg precision	Weighted avg recall	Weighted avg f1	Accuracy
No log	1.0	125	0.5563	0.8103	0.7235	0.7552	0.8574	0.8522	0.8495	0.8522
No log	2.0	250	0.2316	0.8701	0.8460	0.8564	0.9320	0.9315	0.9310	0.9315
No log	3.0	375	0.1635	0.8903	0.8729	0.8809	0.9511	0.9511	0.9508	0.9511
0.5782	4.0	500	0.1112	0.9037	0.8964	0.8998	0.9687	0.9685	0.9685	0.9685
0.5782	5.0	625	0.0860	0.9110	0.9050	0.9079	0.9752	0.9752	0.9751	0.9752
0.5782	6.0	750	0.0675	0.9160	0.9103	0.9131	0.9815	0.9814	0.9814	0.9814
0.5782	7.0	875	0.0588	0.9189	0.9138	0.9163	0.9839	0.9839	0.9839	0.9839
0.1073	8.0	1000	0.0514	0.9214	0.9155	0.9184	0.9858	0.9858	0.9858	0.9858
0.1073	9.0	1125	0.0463	0.9225	0.9171	0.9197	0.9877	0.9876	0.9876	0.9876
0.1073	10.0	1250	0.0442	0.9221	0.9178	0.9199	0.9883	0.9883	0.9883	0.9883

Framework versions

• Transformers 4.34.1
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1