AI & ML interests
None yet
Recent Activity
repliedto tegridydev's post about 23 hours ago Open-MalSec v0.1 – Open-Source Cybersecurity Dataset
Evening! 🫡
📂 Just uploaded an early-stage open-source cybersecurity dataset focused on phishing, scams, and malware-related text samples.
This is the base version (v0.1)—a few structured sample files. Full dataset builds will come over the next few weeks.
🔗 Dataset link:
https://huggingface.co/datasets/tegridydev/open-malsec
🔍 What’s in v0.1?
A few structured scam examples (text-based)
Covers DeFi, crypto, phishing, and social engineering
Initial labelling format for scam classification
⚠️ This is not a full dataset yet (samples are currently available). Just establishing the structure + getting feedback.
📂 Current Schema & Labelling Approach
"instruction" → Task prompt (e.g., "Evaluate this message for scams")
"input" → Source & message details (e.g., Telegram post, Tweet)
"output" → Scam classification & risk indicators
🗂️ Current v0.1 Sample Categories
Crypto Scams → Meme token pump & dumps, fake DeFi projects
Phishing → Suspicious finance/social media messages
Social Engineering → Manipulative messages exploiting trust
🔜 Next Steps
- Expanding datasets with more phishing & malware examples
- Refining schema & annotation quality
- Open to feedback, contributions, and suggestions
If this is something you might find useful, bookmark/follow/like the dataset repo <3
💬 Thoughts, feedback, and ideas are always welcome! Drop a comment or DMs are open 🤙 repliedto merve's post about 2 months ago Don't sleep on new AI at Meta Vision-Language release! 🔥
https://huggingface.co/collections/facebook/perception-encoder-67f977c9a65ca5895a7f6ba1
https://huggingface.co/collections/facebook/perception-lm-67f9783f171948c383ee7498
Meta dropped swiss army knives for vision with A2.0 license 👏
> image/video encoders for vision language modelling and spatial understanding (object detection etc) 👏
> The vision LM outperforms InternVL3 and Qwen2.5VL 👏
> They also release gigantic video and image datasets
The authors attempt to come up with single versatile vision encoder to align on diverse set of tasks.
They trained Perception Encoder (PE) Core: a new state-of-the-art family of vision encoders that can be aligned for both vision-language and spatial tasks. For zero-shot image tasks, it outperforms latest sota SigLIP2 👏
> Among fine-tuned ones, first one is PE-Spatial. It's a model to detect bounding boxes, segmentation, depth estimation and it outperforms all other models 😮
> Second one is PLM, Perception Language Model, where they combine PE-Core with Qwen2.5 LM 7B. it outperforms all other models (including InternVL3 which was trained with Qwen2.5LM too!)
The authors release the following checkpoints in sizes base, large and giant:
> 3 PE-Core checkpoints (224, 336, 448)
> 2 PE-Lang checkpoints (L, G)
> One PE-Spatial (G, 448)
> 3 PLM (1B, 3B, 8B)
> Datasets
Authors release following datasets 📑
> PE Video: Gigantic video datasete of 1M videos with 120k expert annotations ⏯️
> PLM-Video and PLM-Image: Human and auto-annotated image and video datasets on region-based tasks
> PLM-VideoBench: New video benchmark on MCQA Organizations
None yet
