Pieces/time-classification-flan-t5-small-split-best

TIME-Module: Classification — flan-t5-small

Model Description

Temporal intent classification model trained on the split dataset. Classifies user queries into 6 temporal intent categories: Action Scheduling, Content Retrieval, Current Status, Future Information/Planning, Non-Temporal, and Temporal - General.

Training Details

• Base Model: google/flan-t5-small
• Architecture: T5ForSequenceClassification
• Dataset: Pieces/temporal-intent-classification-dataset-split (15,488 train / 1,721 val / 4,303 test)
• Training Steps: 17,500
• Learning Rate: 3e-4
• Batch Size: 64 × 2
• Mixed Precision: bf16
• Hardware: NVIDIA RTX 4090 (24 GB)

Results

Metric	Value
accuracy	95.91%
f1_weighted	~95.9%

Usage

from transformers import pipeline

pipe = pipeline('text-classification', model='Pieces/time-classification-flan-t5-small-split-best')
result = pipe('What meetings do I have tomorrow?')

Part of the TIME-Module Project

This model is part of the TIME (Temporal Intent, Mapping, and Extraction) module, a suite of models for understanding and processing temporal information in natural language.

Related models:

• Pieces/time-classification-flan-t5-small-split-best — Intent classification
• Pieces/time-mapping-flan-t5-small-quality-best — Span prediction (best)
• Pieces/time-mapping-t5gemma-270m-best — Span prediction (T5Gemma)

Citation

@software{time_module,
  title={TIME-Module: Temporal Intent, Mapping, and Extraction},
  author={Pieces},
  year={2026},
  url={https://huggingface.co/Pieces}
}