redflag-detection-V4.2 API & Inference Endpoint

V4.2 vs V4.1

V4.1 (V20d) included 6 "extension" rules (Row 60-65) and 7 symptoms not in the official 59-rule spec (testicular_torsion, pyelonephritis, PID, angle-closure glaucoma, cauda equina, PPH). These were removed in V4.2 to comply with Medical Advisory scope.

Closed symptom set: 96 → 89 items (dropped 7 orphaned)
Rule engine: 65 → 59 rules
Prompt v11 (was v10)
Training data: sanitized (dropped 81 samples of orphan-only cases + 160 orphan symptom mentions from labels) + 31 negative reinforcement samples
Fix: is_male schema hallucination remains 0

Performance (val 269, strict)

Table with columns: V4.0 (V20c), V4.1 (V20d), V4.2 (V20g)
	V4.0 (V20c)	V4.1 (V20d)	V4.2 (V20g)
Exact Match	89.2%	87.0%	89.6%
Precision	0.918	0.901	0.921
Recall	0.749	0.786	0.779
F1	0.825	0.839	0.844

Clinical-fair (SUSPECTED = TP for recall)

Precision: 0.771
Recall: 0.865
F1: 0.815

Notes

Orphan symptom residual: 7/269 model outputs mention removed symptoms. Engine ignores them (0 flag impact). Downstream systems can safely strip them or ignore.
V4.2 covers exactly what's in the official spec. Coverage gaps for testicular torsion, angle-closure glaucoma, cauda equina, PPH require Medical Advisory spec expansion (see git history commit b61a4e9/cd5b628 for the extension logic that was removed).

Usage

bash
python3 -m vllm.entrypoints.openai.api_server \
    --model ningpy/redflag-detection-V4.2 \
    --served-model-name redflag \
    --dtype float16

Pair with the 59-rule V20 engine (redflag_scripts/v20/rule_engine_v20.py).

Training

3766 samples (V20f 3735 + patch20e_neg 31)
LoRA r=32 alpha=64, 7 target projections
3 epochs @ 2e-5, seq len 1900, 8×V100 DDP

V4.2 vs V4.1

Closed symptom set: 96 → 89 items (dropped 7 orphaned)
Rule engine: 65 → 59 rules
Prompt v11 (was v10)
Training data: sanitized (dropped 81 samples of orphan-only cases + 160 orphan symptom mentions from labels) + 31 negative reinforcement samples
Fix: is_male schema hallucination remains 0

Performance (val 269, strict)

Table with columns: V4.0 (V20c), V4.1 (V20d), V4.2 (V20g)
	V4.0 (V20c)	V4.1 (V20d)	V4.2 (V20g)
Exact Match	89.2%	87.0%	89.6%
Precision	0.918	0.901	0.921
Recall	0.749	0.786	0.779
F1	0.825	0.839	0.844

Clinical-fair (SUSPECTED = TP for recall)

Precision: 0.771
Recall: 0.865
F1: 0.815

Notes

Orphan symptom residual: 7/269 model outputs mention removed symptoms. Engine ignores them (0 flag impact). Downstream systems can safely strip them or ignore.
V4.2 covers exactly what's in the official spec. Coverage gaps for testicular torsion, angle-closure glaucoma, cauda equina, PPH require Medical Advisory spec expansion (see git history commit b61a4e9/cd5b628 for the extension logic that was removed).

Usage

bash
python3 -m vllm.entrypoints.openai.api_server \
    --model ningpy/redflag-detection-V4.2 \
    --served-model-name redflag \
    --dtype float16

Pair with the 59-rule V20 engine (redflag_scripts/v20/rule_engine_v20.py).

Training

3766 samples (V20f 3735 + patch20e_neg 31)
LoRA r=32 alpha=64, 7 target projections
3 epochs @ 2e-5, seq len 1900, 8×V100 DDP

redflag-detection-V4.2

README

V4.2 vs V4.1

Performance (val 269, strict)

Clinical-fair (SUSPECTED = TP for recall)

Notes

Usage

Training

Explore FriendliAI today

README

V4.2 vs V4.1

Performance (val 269, strict)

Clinical-fair (SUSPECTED = TP for recall)

Notes

Usage

Training