train_ner_pii_newlabel.py

# /// script
# dependencies = [
#   "transformers>=4.48.0",
#   "datasets>=2.20.0",
#   "evaluate>=0.4.0",
#   "seqeval>=1.2.2",
#   "trackio",
#   "numpy<2.0",
#   "accelerate>=0.34.0",
# ]
# ///
"""
ModernBERT PII NER — remapped to 11 company policy labels.

Trains from answerdotai/ModernBERT-base with a new 23-label classification
head.  Fixes the entity-scan alignment bug: instead of reading char_labels
only at real_s (the first non-space position), we now scan the entire token
span [real_s, tok_e) for the first entity character.  This ensures entities
that start after punctuation (e.g. "(Home" or ":John") are correctly labeled
rather than silently dropped.

Run with:  uv run train_ner_pii.py
"""

import os
import json
import numpy as np
import trackio
from huggingface_hub import login
from datasets import load_dataset
from transformers import (
    AutoTokenizer,
    AutoModelForTokenClassification,
    TrainingArguments,
    Trainer,
    DataCollatorForTokenClassification,
    EarlyStoppingCallback,
)
import evaluate

# Authenticate using the HF_TOKEN secret injected by HF Jobs
_token = os.environ.get("HF_TOKEN")
if _token:
    login(token=_token)
    print("Logged in to Hugging Face Hub.")
else:
    print("WARNING: HF_TOKEN not found in environment — hub push will fail.")

# ── Training label map: 56 source types → 17 training categories ─────────────
# At inference, LABEL_MAP_INFER collapses these to 11 policy categories.
LABEL_MAP_TRAIN = {
    # PER — names only; PREFIX removed (standalone Mr./Dr. caused boundary FPs)
    "FIRSTNAME":           "PER",
    "MIDDLENAME":          "PER",
    "LASTNAME":            "PER",
    "PREFIX":              "O",
    # ORG
    "COMPANYNAME":         "ORG",
    # New labels
    "CUSTOMER_NAME":       "CUSTOMER_NAME",
    "PROJECT_NAME":        "PROJECT_NAME",
    # EMAIL
    "EMAIL":               "EMAIL",
    # PHONE
    "PHONENUMBER":         "PHONE",
    # ADDRESS
    "BUILDINGNUMBER":      "ADDRESS",
    "STREET":              "ADDRESS",
    "SECONDARYADDRESS":    "ADDRESS",
    "CITY":                "ADDRESS",
    "COUNTY":              "ADDRESS",
    "STATE":               "ADDRESS",
    "ZIPCODE":             "ADDRESS",
    # GOV_ID
    "SSN":                 "GOV_ID",
    # FINANCIAL_ID
    "CREDITCARDNUMBER":    "FINANCIAL_ID",
    "CREDITCARDCVV":       "FINANCIAL_ID",
    "IBAN":                "FINANCIAL_ID",
    "BIC":                 "FINANCIAL_ID",
    "BITCOINADDRESS":      "FINANCIAL_ID",
    "ETHEREUMADDRESS":     "FINANCIAL_ID",
    "LITECOINADDRESS":     "FINANCIAL_ID",
    "MASKEDNUMBER":        "FINANCIAL_ID",
    # ACCOUNT_ID — ACCOUNTNAME removed (too ambiguous)
    "ACCOUNTNAME":         "O",
    "ACCOUNTNUMBER":       "ACCOUNT_ID",
    "USERNAME":            "ACCOUNT_ID",
    # DEVICE_ID
    "IP":                  "DEVICE_ID",
    "IPV4":                "DEVICE_ID",
    "IPV6":                "DEVICE_ID",
    "MAC":                 "DEVICE_ID",
    "PHONEIMEI":           "DEVICE_ID",
    "USERAGENT":           "DEVICE_ID",
    "VEHICLEVIN":          "DEVICE_ID",
    "VEHICLEVRM":          "DEVICE_ID",
    # DATE_OF_BIRTH
    "DOB":                 "DATE_OF_BIRTH",
    # Training-only categories (model learns them; suppressed at inference)
    "AMOUNT":              "AMOUNT",
    "DATE":                "DATE",
    "NEARBYGPSCOORDINATE": "NEARBYGPSCOORDINATE",
    "PASSWORD":            "PASSWORD",
    "PIN":                 "PIN",
    "TIME":                "TIME",
    "URL":                 "URL",
    # Explicitly O
    "AGE":                 "O",
    "CURRENCY":            "O",
    "CURRENCYCODE":        "O",
    "CURRENCYNAME":        "O",
    "CURRENCYSYMBOL":      "O",
    "EYECOLOR":            "O",
    "GENDER":              "O",
    "SEX":                 "O",
    "HEIGHT":              "O",
    "JOBAREA":             "O",
    "JOBTITLE":            "O",
    "JOBTYPE":             "O",
    "ORDINALDIRECTION":    "O",
}

TRAIN_LABELS = [
    "ACCOUNT_ID", "ADDRESS", "AMOUNT", "CUSTOMER_NAME", "DATE", "DATE_OF_BIRTH",
    "DEVICE_ID", "EMAIL", "FINANCIAL_ID", "GOV_ID", "NEARBYGPSCOORDINATE",
    "ORG", "PASSWORD", "PER", "PHONE", "PIN", "PROJECT_NAME", "TIME", "URL",
]

label_list = (
    ["O"]
    + sorted(f"B-{l}" for l in TRAIN_LABELS)
    + sorted(f"I-{l}" for l in TRAIN_LABELS)
)
id2label = {i: l for i, l in enumerate(label_list)}
label2id = {l: i for i, l in id2label.items()}

# ── Config ────────────────────────────────────────────────────────────────────
MODEL_NAME   = "answerdotai/ModernBERT-base"        # train from base
DATASET_NAME = "jefftherover/pii-masking-200k-newlabel"
HUB_MODEL_ID = "jefftherover/modernbert-pii-mapped-v5"
OUTPUT_DIR   = "modernbert-pii-mapped-v5"
MAX_LENGTH   = 512

print(f"Labels ({len(label_list)}): {label_list}")

# ── Dataset ───────────────────────────────────────────────────────────────────
print("Loading dataset...")
en = load_dataset(DATASET_NAME, split="train")
print(f"Rows: {len(en)}")

splits   = en.train_test_split(test_size=0.1, seed=42)
train_ds = splits["train"]
eval_ds  = splits["test"]
print(f"Train: {len(train_ds)}  Eval: {len(eval_ds)}")

# ── Tokenizer ─────────────────────────────────────────────────────────────────
tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)

# ── Tokenisation + label alignment ────────────────────────────────────────────
# Identical to v6 logic, with LABEL_MAP applied inside make_char_labels.

def make_char_labels(text, raw):
    spans = json.loads(raw) if isinstance(raw, str) else raw
    cl = ["O"] * len(text)
    for span in spans:
        s, e, src_lbl = int(span[0]), int(span[1]), span[2]
        tgt_lbl = LABEL_MAP_TRAIN.get(src_lbl)
        if tgt_lbl is None:
            continue
        for i in range(s, min(e, len(text))):
            cl[i] = f"B-{tgt_lbl}" if i == s else f"I-{tgt_lbl}"
    return cl

def tokenize_and_align(examples):
    enc = tokenizer(
        examples["source_text"],
        truncation=True,
        max_length=MAX_LENGTH,
        return_offsets_mapping=True,
    )
    all_labels = []
    for idx in range(len(examples["source_text"])):
        text    = examples["source_text"][idx]
        cl      = make_char_labels(text, examples["span_labels"][idx])
        offsets = enc["offset_mapping"][idx]
        labels, prev_end = [], None
        for tok_s, tok_e in offsets:
            if tok_s == tok_e:
                labels.append(-100)
                prev_end = None
                continue
            # Space-offset fix: ModernBERT absorbs leading space into token offset
            real_s = tok_s
            while real_s < tok_e and text[real_s] == " ":
                real_s += 1
            is_word_start = prev_end is None or real_s > tok_s
            # Scan the full token span for the first entity character.
            # Fixes the case where an entity begins after punctuation with no
            # preceding space (e.g. "(Home Loan Account") — previously real_s
            # landed on "(" (O) and the entity was silently dropped.
            lbl = "O"
            for c in range(real_s, min(tok_e, len(cl))):
                if cl[c] != "O":
                    lbl = cl[c]
                    break
            if lbl == "O":
                labels.append(label2id["O"] if is_word_start else -100)
            else:
                labels.append(label2id.get(lbl, label2id["O"]))
            prev_end = tok_e
        all_labels.append(labels)
    enc.pop("offset_mapping")
    enc["labels"] = all_labels
    return enc

print("Tokenising datasets...")
cols      = train_ds.column_names
train_tok = train_ds.map(tokenize_and_align, batched=True, remove_columns=cols)
eval_tok  = eval_ds.map(tokenize_and_align,  batched=True, remove_columns=cols)

# ── Metrics ───────────────────────────────────────────────────────────────────
seqeval = evaluate.load("seqeval")

def compute_metrics(p):
    logits, labels = p
    preds = np.argmax(logits, axis=2)
    true_preds  = [[id2label[pp] for pp, ll in zip(pr, la) if ll != -100]
                   for pr, la in zip(preds,  labels)]
    true_labels = [[id2label[ll] for pp, ll in zip(pr, la) if ll != -100]
                   for pr, la in zip(preds,  labels)]
    res = seqeval.compute(predictions=true_preds, references=true_labels)
    return {
        "precision": res["overall_precision"],
        "recall":    res["overall_recall"],
        "f1":        res["overall_f1"],
        "accuracy":  res["overall_accuracy"],
    }

# ── Model ─────────────────────────────────────────────────────────────────────
# v5: full fine-tune — body + head trained end-to-end on the 49k PII dataset.
# This produces a PII-adapted body that becomes the frozen base for v6+.
#
# v6+ LoRA plan (re-enable this block, change MODEL_NAME to the v5 hub id):
#   from peft import LoraConfig, get_peft_model, TaskType  # add peft to deps too
#   lora_config = LoraConfig(
#       task_type=TaskType.TOKEN_CLS,
#       r=16, lora_alpha=32, lora_dropout=0.1,
#       target_modules=["Wqkv", "out_proj", "Wi", "Wo"],
#       modules_to_save=["classifier"],
#       bias="none",
#   )
#   model = get_peft_model(model, lora_config)
#   model.print_trainable_parameters()
print(f"Loading model (ModernBERT-base + new {len(label_list)}-label head)...")
model = AutoModelForTokenClassification.from_pretrained(
    MODEL_NAME,
    num_labels=len(label_list),
    id2label=id2label,
    label2id=label2id,
)

# ── Trackio ───────────────────────────────────────────────────────────────────
trackio.init(project="modernbert-pii-mapped", name="modernbert-pii-mapped-v5")

# ── Training args ─────────────────────────────────────────────────────────────
args = TrainingArguments(
    output_dir=OUTPUT_DIR,
    num_train_epochs=5,
    per_device_train_batch_size=16,
    per_device_eval_batch_size=32,
    gradient_accumulation_steps=2,          # effective batch = 32
    learning_rate=5e-5,
    weight_decay=0.01,
    warmup_ratio=0.1,
    lr_scheduler_type="cosine_with_restarts",
    eval_strategy="steps",
    eval_steps=500,
    save_strategy="steps",
    save_steps=500,
    save_total_limit=3,
    load_best_model_at_end=True,
    metric_for_best_model="f1",
    greater_is_better=True,
    push_to_hub=True,
    hub_model_id=HUB_MODEL_ID,
    hub_private_repo=False,
    hub_token=os.environ.get("HF_TOKEN"),
    hub_strategy="every_save",
    report_to="trackio",
    run_name="modernbert-pii-mapped-v5",
    fp16=True,
    logging_steps=100,
    dataloader_num_workers=2,
)

# ── Train ─────────────────────────────────────────────────────────────────────
trainer = Trainer(
    model=model,
    args=args,
    train_dataset=train_tok,
    eval_dataset=eval_tok,
    data_collator=DataCollatorForTokenClassification(tokenizer),
    compute_metrics=compute_metrics,
    callbacks=[EarlyStoppingCallback(early_stopping_patience=3)],
)

print("Starting training...")
trainer.train()
trainer.push_to_hub()
print(f"Done! Model pushed to: https://huggingface.co/{HUB_MODEL_ID}")