This repository contains a deliberately crafted 'pickle bomb' inside a converted BERT-tiny model. It serves as a security demonstration showing how superficial scanning may miss malicious payloads in PyTorch model files. The underlying weights originate from the official Google BERT repository.
In the Other AI space, Bert Tiny Torch Picklebomb takes a focused approach. It focuses on demonstrating a security vulnerability where a PyTorch model can contain malicious pickle payloads. It is built as an open-source project for security researchers and ML engineers. Bert Tiny Torch Picklebomb is open source under the Apache-2.0 license. Bert Tiny Torch Picklebomb is available on the web and API.
It is developed by drhyrum, and it first shipped in 2018. The GitHub repository has been archived.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Same category — not a similarity match