Purpose
A measuring instrument, not a chatbot.
BarunLM is an English-centric base completion model. It is designed for architecture research, controlled evaluation, and local experimentation rather than instruction-following chat.
Open model
BarunLM 35M
A compact decoder-only base model for efficient language research and reproducible architecture experiments.
Open on Hugging FaceLanguage model
BarunLM is a 35M-parameter decoder-only base model, built to study how architecture, data quality and evaluation interact under hard compute limits. Twelve layers repeat three local-attention blocks and one global one. It is a measuring instrument, not a chatbot.
35.07Mparameters
5.7Btraining tokens
2,048context length, 12 layers
Design
Local attention, then global context.
The twelve-layer network repeats a three-local, one-global attention pattern. This creates a useful laboratory for studying efficiency and information flow.
Release standard
Evidence travels with the artifact.
Weights, configuration, evaluation details, limitations, and usage guidance are treated as one release surface.