README.md

metadata

library_name: transformers
license: agpl-3.0
base_model: RonTon05/model_content_V2_test
tags:
  - generated_from_trainer
  - text-classification
  - multi-task-learning
model-index:
  - name: MTL_ATESG_Weighted
    results:
      - task:
          type: text-classification
          name: 'Task 1: Binary Classification'
        dataset:
          type: custom
          name: Validation Set
        metrics:
          - type: accuracy
            value: 0.9939
            name: Accuracy
          - type: f1
            value: 0.9892
            name: Macro F1
      - task:
          type: text-classification
          name: 'Task 2: 10-class Classification'
        dataset:
          type: custom
          name: Validation Set
        metrics:
          - type: accuracy
            value: 0.7587
            name: Accuracy
          - type: f1
            value: 0.7924
            name: Macro F1

MTL_ATESG_Weighted

This model is a fine-tuned version of RonTon05/model_content_V2_test on a custom multi-task dataset.

Training Results

TASK 1 — Binary Classification

• Accuracy: 99.39%
• Macro F1: 98.92%

Class	Precision	Recall	F1-Score	Support
0	0.9986	0.9940	0.9963	3654
1	0.9711	0.9933	0.9820	743
Macro Avg	0.9848	0.9936	0.9892	4397
Weighted Avg	0.9940	0.9939	0.9939	4397

TASK 2 — 10-class Classification

• Accuracy: 75.87%
• Macro F1: 79.24%

Class	Precision	Recall	F1-Score	Support
0	0.8114	0.6930	0.7475	329
1	0.9487	0.8605	0.9024	43
2	0.8743	0.9162	0.8947	167
3	0.9585	0.9373	0.9478	271
4	0.9400	0.8034	0.8664	117
5	0.6403	0.7860	0.7057	958
6	0.7981	0.7837	0.7908	1387
7	0.5900	0.5364	0.5619	110
8	0.8450	0.8074	0.8258	135
9	0.7299	0.6386	0.6812	880
Macro Avg	0.8136	0.7762	0.7924	4397
Weighted Avg	0.7653	0.7587	0.7592	4397

---

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 128
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 256
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 1035
• num_epochs: 50

Framework versions

• Transformers 5.10.2
• Pytorch 2.10.0+cu128
• Datasets 5.0.0
• Tokenizers 0.22.2