guess who's back? back again
this time it's not cyber, not algorithmic, it's was a data/ml hackathon on finance
the annual winning streak since 2022 has been secured earlier than expected
replacing the linear MLP of the transformer encoder with a custom layer that has fewer than n² parameters (around n log n), among other modifications.
so far, it works very well with ViTs, but I still haven’t tried it with LLMs
#eupot dot eu update
- dynamic profile photo (updated each time you reconnect)
- fixed the country counter (I don't know how to count)
DAMN, there are people, hi everyone 👋
excited to announce brainROTx: a state-of-the-art LLM for very advanced encryption.
featuring three models (4M, 14M, and 31M parameters) for ROT26, ROT52, and ROT104 operations.
(ROT208 model currently in test by our red team, maybe too dangerous to release)
repo in replies