Twitxy Saved

Lexsi Labs

Verified account This account is verified. Learn more

@Lexsi_labs

Lexsi Labs is dedicated to building the foundations for Safe Superintelligence for the future
London, Paris and Mumbai lexsi.ai Joined November 2025
24 Following 335 Followers 61 Posts

Latest posts 5

Lexsi Labs Verified account @Lexsi_labs
1,535 team registrations across 75+ countries. Less than 48 hours left.

The AI Safety and Alignment Hackathon is underway, and the response has been remarkable.

Problem statements teams are taking on:

🔴 Red Teaming, Safety & Alignment: 478 teams

🤖 Agentic Fine-tuning: 313 teams

📚 Fine-tuning & Data Curation: 238 teams

Participation from top universities: IIT Madras, UCL, IIIT Bangalore, IIIT Allahabad, Purdue, Brown, IIT Roorkee, IIT Kharagpur, IIT Bombay, BITS Pilani, UC Berkeley, Yale, Georgia Tech

And from industry and research: Microsoft, IBM, Salesforce, Accenture, Infosys, TCS, Lloyds Banking Group, Fetch.ai, Cohere Labs, Wadhwani AI, SPAR

⏳ Registrations close in less than 48 hours Nov 5 (AoE). $30,000+ in prizes and credits, plus hiring opportunities.

Register: hackathon.lexsi.ai

From @Lexsi_labs with @Cohere_Labs and @huggingface.

#AISafety #Alignment #LLM #Hackathon #PostTraining
0 replies 1 reposts 13 likes 715 views View on X
Lexsi Labs Verified account @Lexsi_labs
One of the biggest AI safety and Alignment hackathon has started and only 3 days left to register.

Registration closes Nov 5 (AoE).

$30,000+ in prizes and credits, plus hiring opportunities, in collaboration with
Hugging Face and Cohere Labs

Register: hackathon.lexsi.ai/r/x

Also, we will be hosting a bunch of events and hackathons round the year.

Join our discord server : discord.com/invite/dtEDQ2Z…

#AISafetyandAlignment #LexsiLabs #CohereLabs #Huggingface
1 replies 1 reposts 22 likes 1.2K views View on X
Lexsi Labs Verified account @Lexsi_labs
Today we are releasing our first version of MoE class model - OrionMoE-Small-v1.0 from Lexsi Labs.

It is the first MoE tabular foundation model to stand alongside frontier dense models, competing with models up to twice its size.

Tables are the most varied data enterprises run on, yet most foundation models still push every table through the same computation. We think the next step for tabular AI is models that adapt how they compute, not only how large they grow.

This is version 1. We will scale the OrionMoE family from here.

Early access: orionmoe.lexsi.ai

Technical report: arxiv.org/abs/2610.20418
Blog: lexsi.ai/resources/arti…

Big kudos to Mohamed Bouadi, PhD, Nassim Bouarour, @AdityaTann18002, @SVinayKumar12 from our Paris lab.

Its experts sit at the core of the backbone, and each table chooses its own, routed by a learned summary of the dataset.

On TabArena it ranks 12th of 38 methods, above every tuned tree and neural baseline. On every suite, it rates above larger models.

#LexsiLabs #TabularFoundationalModels #OrionMoE #TFMs #EntepriseModels
3 replies 11 reposts 45 likes 2.2K views View on X