Building advanced AI. Advancing research. Prioritizing safety. Jhelum Labs is an independent AI research organization building advanced language models and reliable, production-ready systems.
PyTorch 
Python 
HuggingFace 
GitHub 
Vercel 
NumPy 
Nvidia 
Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda Learn more
W-1.1 Architecture
Decoder-only Transformer with RoPE, RMSNorm, SwiGLU and tied embeddings — 123M parameters trained from scratch.
Project Instruct
Instruction-tuning on 300K Q&A pairs. W-1.1 learns to follow instructions across diverse task types.
Next generation
353M parameters, grouped-query attention, 50K vocab, 4,096-token context, 8B token training target.
Model safety
Every release includes explicit experimental-use warnings, benchmark transparency and SHA-256 weight verification.
PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda PyTorch Python HuggingFace GitHub Vercel NumPy Nvidia Anaconda Latest research
View all →Vitasta: A Recurrent-Core Hybrid Architecture
Vitasta is a language-model architecture designed to address six standing limitations of full self-attention Transformer...
Verified Steps, Not Faster Guesses
A program for mathematical rigor and deliberate reasoning in language models. We propose forcing mathematical reasoning ...
W-1.2 Architecture Specification
W-1.2 is Jhelum Labs' planned second-generation experimental language model, scaling the laboratory's W-1.1 work from ap...
Project Instruct
W-1.1 Project Instruct converted the pretrained W-1.1-125M causal language model into an instruction-following checkpoin...
Final Report
This report consolidates the recorded W-1.1 experiment lineage, final dataset preparation, custom architecture, H100 pre...