[WIP] Support AmSC IRI API (deprecated)

[EZoni]

Overview

Coming soon.

To do

• Update conda environment
• Add new GUI components
• Implement API connection
• Merge first [WIP] Harden training script: credentials validation, error handling, security fixes #432
• Merge first ML Model Download Progress & Logo #434

Notes

Here's a list of related repository, issues, tutorials, etc.:

• https://gitlab.com/amsc2/infrastructure-and-services/amsc-interfaces/amsc-python-client
• https://gitlab.com/amsc2/infrastructure-and-services/amsc-interfaces/amsc-python-client/-/work_items/3
• https://gitlab.com/amsc2/infrastructure-and-services/amsc-interfaces/amsc-python-client/-/work_items/7
• https://github.com/amsc-interfaces/amsc-client-tutorial/blob/main/notebooks/nersc_facility_tutorial.ipynb
• https://gitlab.com/amsc2/ai-services/intelligent-interfaces/agent-skills/iri-api/-/blob/main/SKILL.md (?)

[EZoni]

I post here a minimal working example that I would like to use to test things first and I will update it as needed. It is based on similar examples found in https://gitlab.com/amsc2/infrastructure-and-services/amsc-interfaces/amsc-python-client.

from amsc_client import Client


# Create client instance
client = Client(token="<my_token>")

# Check facility resources
nersc = client.facility("nersc")
for r in nersc.resources():
    print(r.name, r.status)

# Submit a job
perlmutter = nersc.resource("compute")
run_dir = "/pscratch/sd/<first-letter-of-username>/<username>/runs/iri"  # FIXME
job = perlmutter.submit(
    executable="/bin/echo",
    arguments=["Hello World!"],
    directory=run_dir,
    name="hello",
    queue="debug",
    account="<my_project>",
    duration=300,
    nodes=1,
)
job.wait()

# Read job output via the filesystem API
home = nersc.resource("homes")
task = home.fs.head(f"{run_dir}/hello.stdout", lines=10)
task.wait()
print(task.result)

See tracking issue at https://gitlab.com/amsc2/infrastructure-and-services/amsc-interfaces/amsc-python-client/-/work_items/7.

[EZoni]

Temporary installation from local source to apply the patch pushed to https://gitlab.com/edoardo.zoni/amsc-python-client/-/tree/add_upload.

[EZoni]

Note that the syntax "-e /home/edoardo/src/amsc-python-client" did not work for me.

[EZoni]

I confirm that as of 5cbee8d the ML training through AmSC IRI API seems to be working correctly both with and without containerization.

[EZoni]

This file may be redundant and could be removed.