OpenEuroLLM / OpenEuroLLM/Taskboard
Detail and oversee the compute plan
Nobody has claimed this yet.
- Dominant language
- No language data
- Stars
- 3
- Forks
- 0
- PR merge metrics
- No merged PRs in 30d
Description
Where does the job run and for how long? Are there different phases that require more planning like mid-training, annealing, context-lengthening, etc)
How many nodes/GPUs
Planned throughput in terms of tokens/s/GPU
Planned total compute time in terms of HPC budget
For Jupiter in CPU hours
For Leonardo in x hours
For MareNostrum 5 in CPU hours
For LUMI in x hours
Planned required compute wall time (for example 40 days if the job can run indefinitely)
Actual job wall time
How long does each job run for?
How many tokens does each job process?
How many checkpoints does each job process (see storage plan)
(Auto-restart next job?)
What is the QoS
Which partition
Any special partition, reservation or increased priority?
Maximum walltime
Maximum node count
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
No files, tests, or entry points are identified; start by turning the compute-planning questions into an agreed requirements checklist. Done should mean the scope and representation of planned and actual resources, runtimes, throughput, checkpoints, QoS, partitions, and limits are defined well enough to implement.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- hpc
- Domain
- hpc
- Issue type
- Feature
- Difficulty
- 5/5
- Estimated time
- Over a week
- Activity status
- Quiet
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100