CoolFace
Modelpublic

temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc5

sourceHugging Faceapache-2.0updated 6mo agoView on Hugging Face
0likes28downloads
Model Card

OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc5

Token-classification checkpoint for Irish core PII in English and Irish Gaelic.

Included Variants

  • —Full transformers checkpoint in the repo root
  • —Unquantized ONNX export in onnx/model.onnx
  • —Dynamic q8 ONNX artifact in onnx/model_quantized.onnx
  • —inference_mask.py for the full checkpoint
  • —inference_mask_onnx.py for the ONNX q8 artifact
  • —benchmark files in eval/

Coverage

  • —PPSN
  • —ACCOUNT_NUMBER
  • —BANK_ROUTING_NUMBER
  • —CREDIT_DEBIT_CARD
  • —PASSPORT_NUMBER
  • —POSTCODE
  • —PHONE_NUMBER
  • —EMAIL
  • —FIRST_NAME
  • —LAST_NAME
  • —SWIFT_BIC

What Changed From rc4

rc5 keeps the same fine-tuned checkpoint weights as temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc4, but changes the shipped inference stack:

  • —recommended PPSN threshold lowered from 0.71 to 0.55
  • —recommended decoder is now the Irish core label-aware repair decoder for both full and q8 inference
  • —bundled q8 artifact is rebuilt from a preprocessed ONNX export before dynamic int8 quantization

This is the right change because the new QA misses in Gaelic weak-context PPSN text were calibration/inference failures, not weight-quality failures.

Recommended Inference

Full checkpoint:

bash
uv run python inference_mask.py \
  --model temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc5 \
  --ppsn-min-score 0.55 \
  --other-min-score 0.50 \
  --text "Duradh liom mo uimhir 1234567T a sholatar agus me ag denamh iarratais." \
  --json

Dynamic q8 ONNX:

bash
uv run python inference_mask_onnx.py \
  --model temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc5 \
  --onnx-file onnx/model_quantized.onnx \
  --ppsn-min-score 0.55 \
  --other-min-score 0.50 \
  --text "Is e mo upsp na 1234567tw agus teastaionn uaim eolas faoi liuntas curamora." \
  --json

The bundled pyproject.toml is intended for uv. Use uv run so onnxruntime is available for the q8 script.

Key Benchmarks

Fix For The Reported Gaelic PPSN Regression

VariantQA Gaelic weak-context PPSN F1
rc4 full published defaults0.0000
rc4 q8 published defaults0.6667
rc5 full1.0000
rc5 q81.0000

Base OpenMed vs rc5

SuiteBase OpenMedrc5 fullrc5 ONNX q8
Irish core manual0.61190.97370.9669
Irish PPSN/phone edge0.07690.97440.9744
Remaining gapsn/a1.00000.8889
Phone/passport/financen/a0.96000.9362
Finance boundary repairn/a0.91430.8750
Multilingual PPSN0.00000.93330.9333
User PPSN regressionsn/a1.00001.0000
Irish PPSN overlapn/a1.00001.0000

Core Label Breakdown

LabelBase OpenMedrc5 fullrc5 ONNX q8
PPSN0.00000.92310.9231
PHONE_NUMBER0.00000.95650.9565
POSTCODE0.00001.00000.8571
PASSPORT_NUMBER0.00001.00001.0000
ACCOUNT_NUMBER0.40000.85710.8571
BANKROUTINGNUMBER0.00001.00001.0000
EMAIL1.00001.00001.0000
FIRST_NAME0.89471.00001.0000
LAST_NAME0.88891.00001.0000

Dynamic q8 Artifact

Artifact paths:

  • —unquantized: onnx/model.onnx
  • —quantized: onnx/model_quantized.onnx

Quantization recipe used in this repo:

  • —ONNX pre-processing before quantization
  • —ONNX Runtime dynamic int8
  • —qint8
  • —per_channel=true
  • —op_types=MatMul,Gemm,Attention

This q8 path keeps the same F1 as the best prior q8 recipe on the sampled comparison suites while improving CPU throughput on the manual Irish-core suites.

CPU Throughput

SuiteBase OpenMedrc5 fullrc5 ONNX q8
Irish core manual15.796.7034.43
Irish PPSN/phone edge16.6016.5036.56
Multilingual PPSN121.08125.30289.49

Limits

  • —The full checkpoint is still stronger than q8 on the finance-boundary suite.
  • —The q8 artifact is still weaker than the full checkpoint on the strict remaining-gap suite.
  • —Grouped credit/debit-card boundary cases remain the main shared weakness and should still be QA tested.

License And Attribution

  • —Release license: Apache-2.0
  • —Base model: OpenMed/OpenMed-PII-mLiteClinical-Base-135M-v1
  • —See NOTICE and training_sources.json for attribution and release 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

RepoStackFull Core F1Q8 Core F1Q8 Multilingual PPSN F1Q8 Core ex/s
`temsa/IrishCore-GlobalPointer-ContextPII-4L-122M-v1-rc4`4-layer GlobalPointer distilled fast student1.00001.00000.9333299.0
`temsa/IrishCore-GlobalPointer-ContextPII-4L-122M-v1-rc3`4-layer GlobalPointer distilled fast student1.00001.00000.9333317.9
`temsa/IrishCore-GlobalPointer-ContextPII-4L-122M-v1-rc2`4-layer GlobalPointer distilled fast student1.00001.00000.9333292.5
`temsa/IrishCore-GlobalPointer-ContextPII-4L-122M-v1-rc1`4-layer GlobalPointer distilled fast student1.00001.00000.9333337.3
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc27`GlobalPointer raw-only + context labels1.00001.00000.9333270.0
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc25`GlobalPointer raw-only + context labels1.00001.00000.9333212.1
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc24`GlobalPointer raw-only + context labels1.00001.00000.9333278.9
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc23`GlobalPointer raw-only + context labels1.00001.00000.9333237.6
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc22`GlobalPointer raw-only + context labels1.00001.00000.9333106.8
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc21`GlobalPointer raw-only + context labels1.00001.00000.9333150.8
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc20`GlobalPointer raw-only + context labels1.00001.00000.9333181.9
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc19`GlobalPointer raw-only + context labels1.00001.00000.933373.1
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc18`GlobalPointer raw-only + context labels1.00001.00000.9333126.2
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc17`GlobalPointer raw-only + context labels1.00001.00000.9333125.5
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc16`GlobalPointer raw-only + context labels1.00001.00000.9333125.5
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc15`GlobalPointer raw-only + context labels1.00001.00000.9333125.5
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc14`GlobalPointer raw-only + context labels1.00001.00000.9333119.2
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc13`GlobalPointer raw-only + context labels1.00001.00000.9333126.1
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc12`GlobalPointer raw-only + context labels1.00001.00000.933373.6
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc11`GlobalPointer raw-only + context labels1.00001.00000.933394.1
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc10`GlobalPointer raw-only + context labels1.00001.00000.9333125.8
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc9`GlobalPointer raw-only + context labels1.00001.00000.9333119.8
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc8`GlobalPointer raw-only + context labels1.00001.00000.9333128.9
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc7`GlobalPointer raw-only + context labels1.00001.00000.933389.0
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc6`GlobalPointer raw-only + context labels1.00001.00000.933389.0
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc5`GlobalPointer raw-only + context labels1.00001.00000.933384.5
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc4`GlobalPointer raw-only + context labels0.99350.99350.933361.5
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc3`GlobalPointer raw-only + context labels0.99350.99350.933361.5
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc2`GlobalPointer raw-only + context labels0.99350.99350.922261.5
`temsa/IrishCore-GlobalPointer-ContextPII-135M-v1-rc1`GlobalPointer raw-only + context labels0.99350.99350.922261.5
`temsa/IrishCore-GlobalPointer-135M-v1-rc4`GlobalPointer raw-only span-matrix1.00001.00000.9333221.6
`temsa/IrishCore-GlobalPointer-135M-v1-rc3`GlobalPointer raw-only span-matrix1.00001.00000.9213204.9
`temsa/IrishCore-GlobalPointer-135M-v1-rc2`GlobalPointer raw-only span-matrix0.99340.99340.9326231.2
`temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc8`Raw-only token-span0.97370.97370.917646.1
`temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc7`Hybrid classifier + generated scanner spec1.00000.99341.000030.0
`temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc6`Hybrid classifier + repair decoders1.00000.99341.000029.5
`temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc5`Hybrid classifier + repair decoders0.97370.96690.933334.4
`temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc4`Hybrid classifier + repair decoders0.98700.97400.9600114.2
`temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc3`Hybrid classifier + repair decoders0.98060.96770.933344.9
`temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc2`Hybrid classifier + repair decoders0.95540.96150.7887119.1
`temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v1`Hybrid classifier baseline0.95300.93330.9882103.3
`temsa/IrishCore-DiffMask-135M-v1-rc6`DiffMask token-span, scanner-free0.98010.97330.9274130.3
`temsa/IrishCore-DiffMask-135M-v1-rc5`DiffMask token-span, scanner-free0.97330.97330.9379249.2
`temsa/IrishCore-DiffMask-135M-v1-rc4`DiffMask token-span, scanner-free0.97330.97330.937129.5
`temsa/IrishCore-DiffMask-135M-v1-rc3`DiffMask token-span, scanner-free0.96640.96640.959130.0
`temsa/IrishCore-DiffMask-135M-v1-rc2`DiffMask token-span, scanner-free0.96640.96640.9212247.1
`temsa/IrishCore-DiffMask-135M-v1-rc1`DiffMask token-span, scanner-free0.98010.99340.9412251.2

Irish Core PII: Other Public Checkpoints

RepoStackFull Core F1Q8 Core F1Q8 Multilingual PPSN F1Notes
`temsa/OpenMed-mLiteClinical-IrishCorePII-135M-v2-rc1`Hybrid classifier prototype0.9487——Predates the public q8 artifact.

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

RepoArtifactIrish Large F1Multilingual PPSN F1User Raw F1QA v8 F1CPU ex/s
`temsa/OpenMed-mLiteClinical-IrishPPSN-135M-v1`fp32 canonical checkpoint0.89790.97040.80000.738557.4
`temsa/OpenMed-mLiteClinical-IrishPPSN-135M-v1-fp16`fp16 CPU/GPU artifact—0.97040.80000.738545.8
`temsa/OpenMed-mLiteClinical-IrishPPSN-135M-v1-q8`dynamic int8 CPU artifact—0.9040——132.1

PPSN-Only: Historical Public Checkpoints

RepoMain Published MetricsNotes
`temsa/OpenMed-PPSN-mLiteClinical-v1`same as canonical fp32 repo: multilingual 0.9704, user raw 0.8000Legacy alias; prefer temsa/OpenMed-mLiteClinical-IrishPPSN-135M-v1.
`temsa/OpenMed-PPSN-v6-raw-rc2`irishregv5 0.8750; userraw 0.8000; qav8 0.7385Raw PPSN-only research checkpoint; no packaged multilingual CPU benchmark row.
`temsa/OpenMed-PPSN-v5_1`irishlargev2 raw 0.9285; qa_v6 hybrid strict 1.0000Hybrid PPSN-only checkpoint; predates the canonical multilingual suite packaging.
`temsa/OpenMed-PPSN-v5`irishregv5 raw 0.8235; irishregv5 hybrid strict 1.0000Hybrid PPSN-only checkpoint; predates the canonical multilingual suite packaging.
`temsa/OpenMed-PPSN-v4`synthetic non-PPSN drift check onlyPredates the current PPSN eval suite; no packaged apples-to-apples multilingual CPU row.

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 -->