Downloads · 30 days
0
wong-hau-pepelu/synshade-insider-threat-detector
synshade-insider-threat-detector is a text classification model from wong-hau-pepelu. Use it when you need a label for a piece of text. It is set up for tensorflow.
SynShade is a fine-tuned BERT model designed to detect insider threat recruitment activity in dark web communications. The model analyzes Telegram chatter to distinguish between general advertising content and active…
Downloads · 30 days
0
Access
Public
Updated May 18, 2026
Repo size
134 MB
Likes
2
Public
Click a slice to open those files.
.h5134 MB · 100%
From the Hugging Face model README
SynShade is a fine-tuned BERT model designed to detect insider threat recruitment activity in dark web communications. The model analyzes Telegram chatter to distinguish between general advertising content and active insider recruitment attempts.
Model Type: Text Classification (Binary)
Base Model: Small BERT (small_bert/bert_en_uncased_L-4_H-256_A-4/2)
Framework: TensorFlow
Language: English
0: Advertising/General Content1: Insider Recruitment ActivityBertTokenizer from bert-base-uncasedsmall_bert/bert_en_uncased_L-4_H-256_A-4/2 from TensorFlow Hubsmall_bert checkpoints were pretrained on the
same vocabulary as bert-base-uncased. Token IDs and special tokens
([CLS], [SEP], [PAD]) are therefore aligned, and the Hugging Face
tokenizer can be used directly with the TF Hub encoder without a
vocabulary mismatch.Vocabulary compatibility was verified directly against the checkpoint by
comparing special-token IDs and sample token mappings between the Hugging
Face bert-base-uncased tokenizer and the TF Hub small_bert vocabulary:
token HF TFHub
[PAD] 0 0
[UNK] 100 100
[CLS] 101 101
[SEP] 102 102
[MASK] 103 103
recruiting HF=[14357] TFHub=[14357] OK
insider HF=[25297] TFHub=[25297] OK
telegram HF=[23921] TFHub=[23921] OK
threat HF=[5081] TFHub=[5081] OK
exfiltration HF=[4654, 28674] TFHub=[4654, 28674] OK
forum HF=[7057] TFHub=[7057] OK
vocab size match: True | specials match: True | samples match: True
SAFE TO ASSERT: True
The model was trained on a custom dataset of Telegram communications collected from dark web sources. The dataset contains labeled examples of:
Dataset Size:
Note: Due to the sensitive nature of the data, the training dataset is not publicly available.
bert-base-uncased)@software{synshade2026,
title={SynShade: Insider Threat Detection in Dark Web Communications},
author={[Wong Hau Pepelu]},
year={2026},
url={https://huggingface.co/wong-hau-pepelu/synshade-insider-threat-detector}
}
For questions or issues, please open an issue on GitHub or contact [[email protected]].
This model is provided for cybersecurity research and defensive purposes only. The authors and distributors are not responsible for any misuse of this tool. Users are responsible for ensuring their use complies with all applicable laws and regulations.