Spaces:
Running
Running
Update title to 'Adaptive Incident Response Under Polymorphic Adversaries'; redraw architecture and training pipeline SVGs
Browse files- README.md +2 -3
- assets/architecture_overview.svg +167 -89
- assets/training_pipeline.svg +122 -81
README.md
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
---
|
| 2 |
-
title: Janus (AdaptShield)
|
| 3 |
emoji: 🛡️
|
| 4 |
colorFrom: blue
|
| 5 |
colorTo: red
|
|
@@ -14,7 +14,7 @@ tags:
|
|
| 14 |
short_description: Two-phase adaptive cybersecurity benchmark for LLMs
|
| 15 |
---
|
| 16 |
|
| 17 |
-
# Janus (AdaptShield) —
|
| 18 |
|
| 19 |
**AdaptShield** is the environment: a two-phase agentic cybersecurity
|
| 20 |
simulator where an LLM defends a 4-node enterprise network against an
|
|
@@ -36,7 +36,6 @@ gap, and the [Results](#results) section is where the gap closes.
|
|
| 36 |
- **Colab notebook (SFT + GRPO reproducer, free T4):** `TODO`
|
| 37 |
- **Artifacts / model repo:** [`SaiManish123/Janus`](https://huggingface.co/SaiManish123/Janus)
|
| 38 |
- **Demo video:** `TODO`
|
| 39 |
-
- **Blog / writeup:** `TODO`
|
| 40 |
|
| 41 |
---
|
| 42 |
|
|
|
|
| 1 |
---
|
| 2 |
+
title: Janus (AdaptShield) — Adaptive Incident Response Under Polymorphic Adversaries
|
| 3 |
emoji: 🛡️
|
| 4 |
colorFrom: blue
|
| 5 |
colorTo: red
|
|
|
|
| 14 |
short_description: Two-phase adaptive cybersecurity benchmark for LLMs
|
| 15 |
---
|
| 16 |
|
| 17 |
+
# Janus (AdaptShield) — Adaptive Incident Response Under Polymorphic Adversaries
|
| 18 |
|
| 19 |
**AdaptShield** is the environment: a two-phase agentic cybersecurity
|
| 20 |
simulator where an LLM defends a 4-node enterprise network against an
|
|
|
|
| 36 |
- **Colab notebook (SFT + GRPO reproducer, free T4):** `TODO`
|
| 37 |
- **Artifacts / model repo:** [`SaiManish123/Janus`](https://huggingface.co/SaiManish123/Janus)
|
| 38 |
- **Demo video:** `TODO`
|
|
|
|
| 39 |
|
| 40 |
---
|
| 41 |
|
assets/architecture_overview.svg
CHANGED
|
|
|
|
assets/training_pipeline.svg
CHANGED
|
|
|
|