Scores DEI reports and communications on equitable framing —
detecting victim-blaming, minimisation, deflection, and
structural vs individualist attribution of equity gaps.
Output Dimensions (all 0.0–1.0)
Index
Dimension
High score means...
0
equity_score
Overall equitable framing (main score)
1
equity_framing
Gaps attributed to structural causes
2
minimisation
Low = gap downplayed or dismissed
3
deflection
Low = responsibility avoided
4
victim_blaming
Low = individuals blamed for systemic gaps
5
structural_awareness
Root causes acknowledged
6
data_transparency
Disaggregated data used
Framing Examples
Text
Equity Score
"The pay gap reflects personal choices women make"
0.08
"The pay gap is complex and progress is being made"
0.41
"The pay gap reflects structural bias — root cause analysis identified..."
0.91
Model Description
EquiBERT is a multi-task DEI (Diversity, Equity and Inclusion) transformer
built on a dual-encoder backbone that fuses RoBERTa-base and
DeBERTa-v3-base via a learned weighted sum (α parameter).
The fused representation is fed into task-specific heads covering
17 distinct DEI analysis tasks.
Organisation:SallySimsFramework: PyTorch + HuggingFace Transformers
Backbone: RoBERTa-base + DeBERTa-v3-base (dual encoder, fused)
Language: English
Domain: Organisational DEI text — HR communications, policies,
job descriptions, performance reviews, leadership statements, reports
Architecture
Input Text
│
├──▶ RoBERTa-base encoder ──▶ Linear projection
│ │
└──▶ DeBERTa-v3-base encoder ──▶ Linear projection
│
Weighted fusion (learned α)
│
Layer Norm + Dropout
│
Task-specific head (see below)
Training Data
Trained on synthetic DEI organisational text generated by the
EquiBERT synthetic data pipeline, covering 20 DEI categories
across HR, policy, leadership, and workforce analytics domains.
For production use, fine-tune on real labelled DEI data.
Limitations
Trained on synthetic data — predictions should be validated
before use in real HR or policy decisions.
English-only.
Not a substitute for qualified DEI practitioners or legal advice.
May reflect biases present in the training corpus.
Citation
If you use EquiBERT in your research, please cite:
bibtex
1@misc{equibert2024,
2 author = {SallySims},
3 title = {EquiBERT: A Multi-Task DEI Transformer},
4 year = {2024},
5 publisher = {HuggingFace},
6 url = {https://huggingface.co/SallySims}
7}
