add ONNX export (fp16, 94MB)

add ONNX export (fp32)

Add README

Add evaluation metrics

Add model config

HRM-Text prompt injection detector — F1=0.9886

Training in progress, step 14145

Training in progress, step 14000

Training in progress, step 13000

Training in progress, step 12000

Training in progress, step 11000

Training in progress, step 10000

Training in progress, step 9000

Training in progress, step 8000

Training in progress, step 7000

Training in progress, step 6000

Training in progress, step 5000

Training in progress, step 4000

Training in progress, step 3000

Training in progress, step 2000

Training in progress, step 1000

v2 training script: paper-aligned recipe (bf16, lr=2.2e-4, wd=0.1, constant LR, beta2=0.95)

Remove misplaced eval script

Add proper model card

Training in progress, step 16000

Training in progress, step 12000

Training in progress, step 8000

add DistilBERT eval/fine-tune script for bordair comparison

Training in progress, step 4000

fix: match eval and save strategy (both steps/4000)

add hub push on every save + save every 4000 steps for crash recovery

upload training script for HF Jobs benchmark

initial commit