[Beowulf] HPC-GPT: Integrating Large Language Model for High-Performance Computing

[Jonathan Engwall]

Hello Beowulf,
In section 3.2 of the article below you will read example instructions for
interacting with GPT.
Communicating with GPT is difficult. Although the sample constraints seem
obsessive, in the process of proofing a novel, I experienced a similar
barrier. I sought advice on the use of commas. Sounds simple, perhaps, but
it is not. GPT might regurgitate the previous answer or rewrite your sample
entirely.
You must strictly define the results you seek.
Besides that, when I installed Ollama I did not have enough disk space in
/usr. The LLM inflates almost instantly to many gigabytes.
Add a tera-byte to /usr beforehand. I was forced to reinstall my OS over
the weekend. First, /usr blew up and then /var exploded.

https://dl.acm.org/doi/fullHtml/10.1145/3624062.3624172
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://beowulf.org/pipermail/beowulf/attachments/20241229/3e0deb01/attachment.htm>