CoolFace
Modelpublic

SimoneAstarita/october-project-first-sweep-20251007-150229-t01

sourceHugging Faceapache-2.0updated 1y agoView on Hugging Face
0likes
Model Card

october-project-first-sweep-20251007-150229-t01

Multilingual XLM-T (EN/IT/ES) binary classifier Task: LGBTQ+ reclamation vs non-reclamation on social media text.

Trial timestamp (UTC): 2025-10-07 15:02:29

Configuration (trial hyperparameters)

HyperparameterValue
LR3e-05
EPOCHS3
MAX_LENGTH256
USE_BIOTrue
USELANGTOKENTrue
GATED_BIOTrue
FOCAL_LOSSTrue
FOCAL_GAMMA1.5
USE_SAMPLERTrue
R_DROPTrue
RKLALPHA1.0

Dev set results

MetricValue
trial1
cfg{'LR': 3e-05, 'EPOCHS': 3.0, 'MAXLENGTH': 256.0, 'USEBIO': 1.0, 'USELANGTOKEN': 1.0, 'GATEDBIO': 1.0, 'FOCALLOSS': 1.0, 'FOCALGAMMA': 1.5, 'USESAMPLER': 1.0, 'RDROP': 1.0, 'RKL_ALPHA': 1.0}
f1macrodev0.8083653435766112
bestthresholddev0.75
precisionmacrodev0.8363588717362302
recallmacrodev0.7865056818181818

Data

  • —Train/Dev: private dataset (merged train_en.csv, train_it.csv, train_es.csv with 15% stratified Dev).
  • —The dataset id is intentionally not disclosed in metadata.

Usage

python
from transformers import AutoTokenizer, AutoModelForSequenceClassification
tok = AutoTokenizer.from_pretrained("SimoneAstarita/october-project-first-sweep-20251007-150229-t01")
model = AutoModelForSequenceClassification.from_pretrained("SimoneAstarita/october-project-first-sweep-20251007-150229-t01")