SOFT : Ollama
------

Site du soft: https://github.com/ollama/ollama
-------------

LICENSE:
--------

MIT

See software documentation for more informations.

Location: /usr/local/bioinfo/src/Ollama
---------

Load binaries and environment:
------------------------------

-> Version 0.20.7

module load bioinfo/Ollama/0.20.7
export OLLAMA_MODELS=/path/to/your/working/directory

Add "sleep 10" after "ollama serve" to gives the server time to start before executing other commands:
ollama serve &
sleep 10

Default port on calcul node: 11434 if used: export OLLAMA_HOST="127.0.0.1:<NEW PORT>"

To run on GPU:
#Need GPU. To submit on gpuq, see FAQ: https://bioinfo.genotoul.fr/index.php/faq/job_submission_faq/ -> How to use GPU node

Example directory for use on cluster:
-------------------------------------

CPU example (TEST EN COURS)
/usr/local/bioinfo/src/Ollama/example_on_cluster

To submit: sbatch test_Ollama-v0.20.7.sh

GPU example:

/usr/local/bioinfo/src/Ollama/example_on_cluster/Ollama-v0.20.7_GPU 

To submit: sbatch test_Ollama-v0.20.7_GPU.sh

See software documentation and our FAQ (https://vm-genoword.toulouse.inrae.fr/FAQ) for more informations.