temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc3

temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc3

Third QA release candidate for Irish core PII detection with OpenMed mLiteClinical.

This repository should be evaluated against:

• —current public release: temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v1
• —previous RC: temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc2
• —this repository: temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc3

This RC is the first version in the IrishCorePII line that is stronger than temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc2 across the main suites used in this workspace for release gating:

• —Irish core multilabel suite
• —Irish PPSN and phone edge suite
• —exact numeric QA suite
• —exact gap-feedback QA suite
• —user PPSN regression suite
• —multilingual PPSN suite
• —strict exact-span boundary suite

This repository also includes a stronger bundled ONNX q8 artifact for CPU inference in onnx/model_quantized.onnx.

Included Variants

Coverage

• —PPSN
• —account_number
• —bank_routing_number
• —credit_debit_card
• —PASSPORT_NUMBER
• —postcode
• —phone_number
• —email
• —first_name
• —last_name
• —swift_bic

Recommended Inference

Full checkpoint:

python3 inference_mask.py   --model temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc3   --ppsn-min-score 0.55   --other-min-score 0.40   --text "My sort code is 90-00-17 for AIB."   --json

Fast CPU path with the bundled ONNX q8 artifact:

python3 inference_mask_onnx.py   --model temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc3   --ppsn-min-score 0.60   --other-min-score 0.45   --text "Please provide your passport: NN5123456."   --json

What Improved Versus temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc2

Full checkpoint:

• —core suite F1: 0.9554 -> 0.9806
• —edge suite F1: 0.9500 -> 1.0000
• —exact numeric QA F1: 0.8966 -> 0.9333
• —exact gap-feedback QA F1: 0.8696 -> 0.9167
• —user PPSN regression F1: 0.8571 -> 1.0000
• —multilingual PPSN F1: 0.8038 -> 0.9333

Bundled ONNX q8:

• —core suite F1: 0.9677 -> 0.9677
• —edge suite F1: 0.9500 -> 1.0000
• —exact numeric QA F1: 0.8667 -> 0.9333
• —exact gap-feedback QA F1: 0.8696 -> 0.9167
• —user PPSN regression F1: 0.8571 -> 1.0000
• —multilingual PPSN F1: 0.8077 -> 0.9333

Quality Snapshot

Key label changes on the Irish core suite for the full checkpoint:

• —PASSPORT_NUMBER: 0.8000 -> 1.0000
• —POSTCODE: 0.7500 -> 1.0000
• —PPSN: 0.8571 -> 0.9231
• —PHONE_NUMBER on the numeric QA suite: 0.7500 -> 0.8889

Benchmark Tables

Broader CPU benchmarks:

Exact QA suites:

CPU throughput:

Quantized Artifact

The bundled quantized artifact is:

• —onnx/model_quantized.onnx

For this release line, the promoted q8 recipe remains the standard dynamic int8 ONNX export with per-channel quantization. Two alternative q8 recipes were tested locally and not promoted because they reduced multilingual PPSN quality.

Known Limits

This is still a raw-model-only release candidate. QA should still test these carefully:

• —Passport PA 1234567 was used to board the flight.
• —Usaideadh pas PA 1234567 chun dul ar bord an eitilt.
• —My PPSN is 1234567T and my sort code is 90-00-17.
• —the full checkpoint can still trim the final digit in some short English compact-phone contexts such as Call me on 0851234567 tomorrow.
• —multilingual PPSN quality is stronger than temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc2, but still below temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v1 on the multilingual PPSN suite

Included Files

• —full transformers checkpoint in the repo root
• —dynamic int8 ONNX artifact in onnx/model_quantized.onnx
• —inference_mask.py
• —inference_mask_onnx.py
• —qa_config.json
• —training_sources.json
• —benchmark summaries in eval/

License And Attribution

• —release license: Apache-2.0
• —base model: OpenMed/OpenMed-PII-mLiteClinical-Base-135M-v1
• —upstream attributed data: joelniklaus/mapa, gretelai/synthetic_pii_finance_multilingual
• —synthetic Irish training and replay data created in this workspace

See NOTICE for attribution details.

<!-- portfolio-comparison:start -->

Portfolio Comparison

Updated: 2026-03-16.

Use this section for the fastest public comparison across the temsa PII masking portfolio.

• —The first core table only includes public checkpoints that ship both comparable q8 accuracy and q8 CPU throughput.
• —The first PPSN table only includes public artifacts that ship comparable PPSN accuracy and CPU throughput.
• —Missing cells in the archive tables mean the older release did not ship that metric in its public bundle.
• —DiffMask rows use the reconciled clean_single_pass harness that matches the deployed runtime.
• —GlobalPointer rows use the public raw-only span-matrix release bundle and its packaged q8 ONNX artifact.
• —The same content is shipped as PORTFOLIO_COMPARISON.md inside each public model repo.

Irish Core PII: Comparable Public Checkpoints

Irish Core PII: Other Public Checkpoints

Finance-boundary q8 F1 is 1.0000 for OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc6, OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc7, OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc8, and all public IrishCore-DiffMask releases from rc1 to rc6. OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc5 ships 0.8750 on that public q8 suite.

PPSN-Only: Comparable Public Artifacts

PPSN-Only: Historical Public Checkpoints

If you need the strongest current raw-only Irish core model, start with IrishCore-GlobalPointer-135M-v1-rc4. If you need the fastest CPU-first raw-only line, compare it against IrishCore-DiffMask-135M-v1-rc6. If you need a PPSN-only artifact, compare the canonical fp32, fp16, and q8 variants of OpenMed-mLiteClinical-IrishPPSN-135M-v1 directly in the table above. <!-- portfolio-comparison:end -->