MLdata Module

HCC hosts multiple datasets and models for machine learning and AI research on Swan.

These are accessible through environment variables provided by the "mldata" module, which needs to be loaded before the variables are available for use.

For how to load module, please check Module Commands.

Note

The mldata module is maintained and updated by HCC staff on a per-request basis. Datasets and models are not automatically updated. Please email hcc-support@unl.edu with any questions or issues with the module.

Loading the "mldata" module will pre-set many environment variables, but most likely you will only need a subset of them. Environment variables can be used in your command or script by prefixing $ to the name.

The major environment variables are:

  • $MLDATA - Main database directory
  • $IMAGENET - ImageNet database (most recent version)
  • $IMAGENET_WINTER21 - ImageNet database Winter 21 (currently same as most recent)
  • $TCGA - TCGA database (most recent version)
  • $TCGA_MAY2022 - TCGA database retrieved May 2022 (currently same as most recent)
  • $CAMELYON16 - CAMELYON 16 database
  • $CAMELYON17 - CAMELYON 17 database
  • $TCIA - TCIA database (most recent version)
  • $TCIA_JUNE2023 - TCIA database retrieved June 2023 (currently same as most recent)
  • $TCIA_PANCREAS_CT - TCIA PANCREAS_CT Collection (most recent version)
  • $TCIA_PANCREATIC_CT_CBCT_SEG - TCIA PANCREATIC_CT_CBCT_SEG Collection (most recent version)
  • $OLLAMA_MODELS - Ollama models retrieved September 2025
  • $ESM_MODELS - ESM models retrieved October 2025
  • $EVO2_MODELS - EVO2 models retrieved September 2026

An example of using the pre-downloaded models with Ollama is shown below. Ollama will use the $OLLAMA_MODELS variable for the location of where to read and write models. The mldata module changes this variable to be the system wide models that are pre downloaded.

Ollama Example
#!/bin/bash
#SBATCH --job-name=ollama_gpu
#SBATCH --output=ollama_gpu.%j.out
#SBATCH --error=ollama_gpu.%j.err
#SBATCH --time=01:00:00
#SBATCH --nodes=1
#SBATCH --ntasks-per-node=1
#SBATCH --partition=gpu
#SBATCH --gres=gpu:1
#SBATCH --mem=10GB

module purge
module load mldata/1.0
module load ollama-gpu/0.15

# ~1 minute runtime
# Run Ollama server
ollama serve &
sleep 10
ollama pull gemma3:1b
ollama run gemma3:1b "What is geometry?"

A full list and details about the available models and datasets are available by running the module help mldata command on Swan.