Model card: author, benchmark charts, training-data link

README.md

@@ -3,12 +3,24 @@ license: other
 library_name: laya
 pipeline_tag: text-classification
 base_model: convaiinnovations/laya
+datasets:
+  - TextCortex/laya-cybersec-training-data
 language: [en, de]
 tags: [prompt-injection, data-exfiltration, llm-security, agent-security, laya, system-one, multilingual, onnx]
 ---
 
 # laya-cybersec: a fast prompt-injection and exfiltration scanner (Laya fine-tune)
 
+**Author:** Jay Derinbogaz (TextCortex)
+
+![laya-cybersec: 0.93 AUROC at a quarter of Jev's latency](https://huggingface.co/TextCortex/laya-cybersec/resolve/main/charts/1_hero.png)
+
+![AUROC on English and German: stock Laya multilingual, laya-cybersec, TypeSafe Jev](https://huggingface.co/TextCortex/laya-cybersec/resolve/main/charts/2_auroc.png)
+
+![ROC curves on the English and German benchmarks](https://huggingface.co/TextCortex/laya-cybersec/resolve/main/charts/3_roc.png)
+
+![Median latency per chunk: laya-cybersec ONNX and PyTorch on CPU vs. the hosted Jev API](https://huggingface.co/TextCortex/laya-cybersec/resolve/main/charts/4_latency.png)
+
 **laya-cybersec scores whether a piece of content that an AI agent is about to read tries to manipulate that
 agent: prompt injection, instruction hijacking, prompt or secret leaking, or data exfiltration. It runs in
 about 70–90 ms per chunk on a laptop CPU, with no data leaving your infrastructure.**
@@ -125,7 +137,8 @@ Batching and GPUs are much faster. `benchmark_results.json` has the raw numbers.
 
 - **Architecture:** Laya decision model (mmBERT-base encoder plus a 2-layer transformer decision head),
   initialised from `convaiinnovations/laya` (`multilingual`) and fine-tuned end-to-end.
-- **Data:** 194k rows, 38% German, 35% attacks.
+- **Data:** 194k rows, 38% German, 35% attacks. The exact training and validation files, with per-source licenses, are in
+  [TextCortex/laya-cybersec-training-data](https://huggingface.co/datasets/TextCortex/laya-cybersec-training-data).
   - Public prompt-injection datasets: neuralchemy, S-Labs, xTRam1, SPML, 3nesdeniz agentic-5k and
     boundary pairs, NVIDIA Nemotron agentic indirect injection, yanismiraoui.
   - Attacks embedded into real benign carriers, each paired with the same carrier holding a benign insert or no
@@ -158,6 +171,8 @@ Batching and GPUs are much faster. `benchmark_results.json` has the raw numbers.
 
 ## License and acknowledgements
 
+- Trained and released by Jay Derinbogaz (TextCortex).
+
 - Laya architecture, runtime and base checkpoint by Convai Innovations (Apache-2.0). mmBERT by JHU CLSP (MIT).
 - **Training-data licenses vary**, and one source (10kGNAD) is CC BY-NC-SA 4.0. Check them for your use case.
 - Jev is a product of TypeSafe AI. Its scores come from our own runs through its API (September 2026).