Pinned post
GLADIA Research Lab
Verified account This account is verified. Learn more@GladiaLab
Based in Rome, GLADIA is a team of computer scientists, physicists, engineers and mathematicians venturing beyond the boundaries of machine intelligence
28 Following 5K Followers 84 Posts
Latest posts 5
the paper behind the speedrun’s library is out 📄
Perseus: A Full-Stack System for Homomorphic Model Inference
tl;dr: public code, python api, and 3.96× faster encrypted Llama-3-8B decoding than the previous closed-source SOTA.
arxiv.org/abs/2611.04488
Join the speedrun on GPT-2 👇
Perseus: A Full-Stack System for Homomorphic Model Inference
tl;dr: public code, python api, and 3.96× faster encrypted Llama-3-8B decoding than the previous closed-source SOTA.
arxiv.org/abs/2611.04488
Join the speedrun on GPT-2 👇
arxiv.org Perseus: A Full-Stack System for Homomorphic Model Inference Large Language Models (LLMs) have made it exceptionally convenient for users to send their private data to model providers, which process them in plaintext. Fully Homomorphic Encryption (FHE)... wanna help close a 4,655× gap?
Join the Encrypted GPT-2 Speedrun 🏁
The current best stands at 3.031 s/tok.
The goal is breaking the 1.0 s/tok barrier!
🧵(1/4)View quoted post on X
wanna help close a 4,655× gap?
Join the Encrypted GPT-2 Speedrun 🏁
The current best stands at 3.031 s/tok.
The goal is breaking the 1.0 s/tok barrier!
🧵(1/4)
Join the Encrypted GPT-2 Speedrun 🏁
The current best stands at 3.031 s/tok.
The goal is breaking the 1.0 s/tok barrier!
🧵(1/4)
Original or AI reconstruction? 🎧 You’ll see every switch. Can you hear one?
Meet SAGE, our 105M music autoencoder:
⚡ As fast as Stable Audio Open’s VAE
🎧 On par with SAME-L in listening tests, at 1/8 the size
🏆 SOTA on 23/25 fidelity scores + 19/19 semantic tasks
(1/6)
Meet SAGE, our 105M music autoencoder:
⚡ As fast as Stable Audio Open’s VAE
🎧 On par with SAME-L in listening tests, at 1/8 the size
🏆 SOTA on 23/25 fidelity scores + 19/19 semantic tasks
(1/6)
0:50
Language Models are Injective and Hence Invertible (ICLR 2026), aka “pringle paper", is now a public graph on @paradigmainc’s Flywheel
In the paper, we show that LMs can be inverted and, contrary to common belief, do not discard information about their inputs at inference time.
In the paper, we show that LMs can be inverted and, contrary to common belief, do not discard information about their inputs at inference time.
LLMs are injective and invertible.
In our new paper, we show that different prompts always map to different embeddings, and this property can be used to recover input tokens from individual embeddings in latent space.
(1/6)View quoted post on X