Add APT classification model and configuration

README.md

@@ -0,0 +1,19 @@
+# CySecBERT APT Classifier
+
+Advanced Persistent Threat group classification using CySecBERT.
+
+## Model Details
+- Architecture: CySecBERTMaxPerformance  
+- Classes: 12 APT groups
+- F1-Score: 95.74%
+- Training time: 12.6 minutes
+- Model size: 0.41GB
+
+## Usage
+```python
+from transformers import AutoTokenizer
+import torch
+
+# Load model
+model = torch.load("best_cysecbert_max_performance.pt")
+tokenizer = AutoTokenizer.from_pretrained("markusbayer/CySecBERT")

best_cysecbert_max_performance.pt

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c32ef320c2133d03982b6bb7606685f27c17d91009d6d0559f4250cc221bbc09
+size 440146673

config.json

@@ -0,0 +1,9 @@
+{
+  "model_type": "cysecbert-apt",
+  "architectures": ["CySecBERTMaxPerformance"],
+  "num_classes": 12,
+  "hidden_size": 768,
+  "max_position_embeddings": 512,
+  "model_class": "APTClassifier",
+  "framework": "pytorch"
+}