Manage GPU compute jobs on the Qizhi platform via a kubectl-style CLI.
npx skills add https://github.com/duypham2801/ThS_LLM --skill qzcli-duypham2801
Please help me install this Agent Skill. Skill: qzcli Source: https://github.com/duypham2801/ThS_LLM/tree/main/.claude/skills/qzcli Command: npx skills add https://github.com/duypham2801/ThS_LLM --skill qzcli-duypham2801
qzcli eliminates the hassle of manually searching, authenticating, and tracking GPU training jobs on the Qizhi (启智) platform by providing a single kubectl-style command interface.
Use Case: When launching distributed training for your experiment, use qzcli to log in, refresh resource caches, inspect available nodes, submit training jobs, and monitor/stop them if needed.
Run qzcli login to authenticate, then qzcli create -i to interactively select a workspace, compute group, and job spec for submission.
High-intent search queries and answers about installing and using this skill.
You can manage GPU jobs on the Qizhi platform using a kubectl-style CLI to handle login, resource discovery, job submission with Docker images, and lifecycle actions. It supports interactive and non-interactive modes for distributed training workflows.
Yes, you can check node availability for distributed training jobs by refreshing cached resources. This allows you to discover available compute resources, including low-priority nodes, before submitting your Docker image jobs.
To submit a compute job interactively, use the create command with the interactive flag. This guides you through selecting a workspace, compute group, and job spec to launch your Docker image training job on available nodes.
The best way to monitor and stop active HPC scheduling jobs is using dedicated CLI commands to list jobs, check their status, watch ongoing progress, and reliably stop jobs when needed without navigating the web interface.
No, you do not need Kubernetes to use this kubectl-style CLI for HPC scheduling. It operates directly on the Qizhi platform, resolving workspace and compute-group identifiers through cached resource management rather than requiring a Kubernetes environment.
GPU job management commands support configuration via CLI arguments, stdin password input, environment variables, or a local environment file. This allows flexible authentication setups for both interactive and automated job submissions.
mhintz1980Authenticate users and manage sessions with the Supabase Auth API.
zhuyingqinManage GPU compute jobs on the Qizhi platform via kubectl-style qzcli commands.
felixminBuild, push, and safely swap LRZ Enroot container images on the cluster.
randDiscover ML and AI skills during model training and inference workflows.
