ORGANISATION

LUCAN 3B

LUCAN 3B is an approximately 3-billion-parameter scientific AI model from Shodh AI for processing numerical reactor data. This release includes the model weights, dedicated tokenizers and matching adapters for bioreactor and continuous-flow reactor applications, along with loading instructions and an introductory guide.

About Model

LUCAN 3B processes structured scientific data through a shared neural model and domain-specific components. This release includes two application domains: bioreactor and continuous-flow reactor.

Each domain has a tokenizer that converts numerical field data into the model’s input representation, and an adapter that connects that representation to the shared model. A decoder converts the resulting representation back into numerical fields.

The package contains:

LUCAN 3B model weights.
A nine-channel bioreactor tokenizer and matching adapter.
A twenty-five-channel continuous-flow reactor tokenizer and matching adapter.
Model-loading code, usage instructions, file-integrity checks and an introductory PDF.

Basic technical details:

Model size: approximately 3 billion parameters.
Internal representation width: 3,072.
Encoded document dimensions: 4,096 tokens × 256 features.
Weight formats: Safetensors for the main model and PyTorch checkpoint files for tokenizer and adapter components.