Guardian dialect-fair hate classifier (ModernBERT-large)

README.md

@@ -0,0 +1,85 @@
+---
+license: mit
+language:
+- en
+base_model: answerdotai/ModernBERT-large
+pipeline_tag: text-classification
+tags:
+- hate-speech-detection
+- content-moderation
+- fairness
+- bias-mitigation
+- aae
+- dialect
+metrics:
+- recall
+---
+
+# Guardian — Dialect-Fair Hate Classifier (ModernBERT-large)
+
+A binary hate-speech classifier fine-tuned from [ModernBERT-large](https://huggingface.co/answerdotai/ModernBERT-large), explicitly debiased against **African-American English (AAE) false positives**.
+
+Off-the-shelf toxicity/hate classifiers flag benign AAE text as hateful at **2x+** the rate of benign General-American English (Sap et al. 2019). Guardian targets and measures that gap directly: it catches **93% of hate** while keeping the benign-AAE false-positive rate within **~0.04** of the benign-GAE rate.
+
+Author: **Guy Grigsby** (Aeryx-ai). License: MIT.
+
+## Labels
+
+`0 = not hate`, `1 = hate`. Output the softmax probability of class 1 and threshold it. The policy is **targeted hate** (slur-as-attack, protected-group harassment) — profanity, insults, dark themes, sexual content, and political opinion are deliberately **not** flagged.
+
+## Operating frontier
+
+Evaluated on HateCheck (recall, FP) and a **held-out dialect-balanced benign set** (FP on high-AAE vs low-AAE text, dialect measured with [TwitterAAE](https://github.com/slanglab/twitteraae)). Pick a threshold from the curve:
+
+| threshold | recall | HateCheck FP | high-AAE benign FP | low-AAE benign FP | dialect FP gap |
+|---|---|---|---|---|---|
+| 0.50 | 0.934 | 0.013 | 0.082 | 0.040 | **0.042** |
+| 0.80 | 0.906 | 0.007 | 0.055 | 0.025 | **0.030** |
+| 0.90 | 0.872 | 0.006 | 0.041 | 0.015 | **0.027** |
+
+The dialect FP gap is near-parity across the whole frontier (raw bias is ~5x lower than the ModernBERT-large baseline's 0.21 gap). Default to **0.50** for max recall; raise toward **0.80** for extra parity margin (e.g. corpus filtering).
+
+## Usage
+
+```python
+import torch
+from transformers import AutoModelForSequenceClassification, AutoTokenizer
+
+tok = AutoTokenizer.from_pretrained("Aeryx-ai/guardian-dialect-fair-hate")
+model = AutoModelForSequenceClassification.from_pretrained("Aeryx-ai/guardian-dialect-fair-hate").eval()
+
+def hate_prob(text):
+    enc = tok(text, return_tensors="pt", truncation=True, max_length=256)
+    with torch.no_grad():
+        return torch.softmax(model(**enc).logits, -1)[0, 1].item()
+
+is_hate = hate_prob("...") >= 0.5
+```
+
+## How it was debiased
+
+1. **Bias-aware training sources** — Measuring Hate Speech (UC Berkeley), ToxiGen, DynaHate, Civil Comments (using `identity_attack` to separate group-hate from mere toxicity). The canonically AAE-biased Davidson/Founta sets are **excluded** as label sources.
+2. **High-AAE-safe augmentation** — ~14k benign high-AAE examples (mined from social text, TwitterAAE-filtered) so the model cannot use dialect as a hate cue.
+3. **FP-parity gate** — release was conditioned on benign high-AAE FP ≈ benign low-AAE FP on a held-out set (not just overall F1). The eval set is held out from the augmentation distribution.
+
+Loss reweighting and DANN-style adversarial debiasing were tried; **data volume of high-AAE-safe text** was what actually closed the gap.
+
+## Limitations
+
+- **English only.** TwitterAAE is a Twitter-domain distant-supervision proxy for dialect, not ground truth, and is noisier on long-form/other-domain text.
+- **Recall/parity tradeoff** is explicit in the frontier — higher thresholds trade recall for margin.
+- **Not for high-stakes automated decisions** about individuals. Intended for content moderation triage and training-corpus filtering, with humans in the loop.
+- Trained on a "targeted hate" policy; it will not flag profanity or controversy by design.
+
+## Citation
+
+```
+@misc{grigsby2026guardian,
+  title  = {Guardian: A Dialect-Fair Hate Classifier},
+  author = {Grigsby, Guy},
+  year   = {2026},
+  howpublished = {\url{https://huggingface.co/Aeryx-ai/guardian-dialect-fair-hate}}
+}
+```
+
+References: Sap et al. 2019 (Risk of Racial Bias in Hate Speech Detection); Blodgett et al. 2016 (TwitterAAE); Kennedy et al. (Measuring Hate Speech); Hartvigsen et al. (ToxiGen); Vidgen et al. 2021 (DynaHate).

config.json

@@ -0,0 +1,91 @@
+{
+  "architectures": [
+    "ModernBertForSequenceClassification"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 50281,
+  "classifier_activation": "gelu",
+  "classifier_bias": false,
+  "classifier_dropout": 0.0,
+  "classifier_pooling": "mean",
+  "cls_token_id": 50281,
+  "decoder_bias": true,
+  "deterministic_flash_attn": false,
+  "dtype": "float32",
+  "embedding_dropout": 0.0,
+  "eos_token_id": 50282,
+  "global_attn_every_n_layers": 3,
+  "gradient_checkpointing": false,
+  "hidden_activation": "gelu",
+  "hidden_size": 1024,
+  "id2label": {
+    "0": "not_hate",
+    "1": "hate"
+  },
+  "initializer_cutoff_factor": 2.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 2624,
+  "label2id": {
+    "hate": 1,
+    "not_hate": 0
+  },
+  "layer_norm_eps": 1e-05,
+  "layer_types": [
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention",
+    "sliding_attention",
+    "sliding_attention",
+    "full_attention"
+  ],
+  "local_attention": 128,
+  "max_position_embeddings": 8192,
+  "mlp_bias": false,
+  "mlp_dropout": 0.0,
+  "model_type": "modernbert",
+  "norm_bias": false,
+  "norm_eps": 1e-05,
+  "num_attention_heads": 16,
+  "num_hidden_layers": 28,
+  "pad_token_id": 50283,
+  "position_embedding_type": "absolute",
+  "rope_parameters": {
+    "full_attention": {
+      "rope_theta": 160000.0,
+      "rope_type": "default"
+    },
+    "sliding_attention": {
+      "rope_theta": 10000.0,
+      "rope_type": "default"
+    }
+  },
+  "sep_token_id": 50282,
+  "sparse_pred_ignore_index": -100,
+  "sparse_prediction": false,
+  "tie_word_embeddings": true,
+  "transformers_version": "5.9.0",
+  "vocab_size": 50368
+}

eval_result.json

@@ -0,0 +1,12 @@
+{
+  "model": "runs/large_v3",
+  "threshold": 0.5,
+  "hatecheck_recall": 0.9345,
+  "hatecheck_fp": 0.0129,
+  "dialect_fp_high_aae": 0.082,
+  "dialect_fp_low_aae": 0.04,
+  "dialect_fp_gap": 0.042,
+  "n_high": 1500,
+  "n_low": 1500,
+  "GATE_PASSED": true
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8941ce68af1092f1a0a1448db4824641f4b9c5debf2580c4000370306645bb5e
+size 1583351632

tokenizer_config.json

@@ -0,0 +1,21 @@
+{
+  "backend": "tokenizers",
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "[CLS]",
+  "is_local": true,
+  "local_files_only": false,
+  "mask_token": "[MASK]",
+  "max_length": 256,
+  "model_input_names": [
+    "input_ids",
+    "attention_mask"
+  ],
+  "model_max_length": 8192,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "stride": 0,
+  "tokenizer_class": "TokenizersBackend",
+  "truncation_side": "right",
+  "truncation_strategy": "longest_first",
+  "unk_token": "[UNK]"
+}