generated: '2026-08-02'
method: searched
source: https://github.com/tensorwavecloud/ScalarLM#cli-commands
docs: https://www.scalarlm.com/cli/
name: scalarlm
summary: 'TensorWave publishes no CLI for its GPU cloud control plane (that platform is driven
  over SSH, Slurm sbatch/squeue and kubectl). The first-party CLI is ScalarLM''s, which comes in
  two forms: a repository-local orchestration script that builds and runs a deployment, and a
  pip-installed monitoring client that talks to a running deployment.'
surfaces:
- name: ./scalarlm
  kind: repository script
  install:
  - method: git
    command: git clone git@github.com:tensorwavecloud/scalarlm.git
  scope: Builds, starts and tests a ScalarLM deployment from a checkout.
  commands:
  - command: up
    description: Start the ScalarLM server (development server on localhost:8000 with an
      OpenAI-compatible API).
  - command: benchmark
    description: Run performance benchmarks.
  - command: llm-logs
    description: View LLM logs.
  - command: llm-ls
    description: List available models.
  - command: llm-plot
    description: Plot training metrics.
  - command: llm-squeue
    description: View training queue status.
  - command: test
    description: Run tests.
  - command: build-image
    description: Build the Docker image.
- name: scalarlm
  kind: pip-installed monitoring CLI
  install:
  - method: pip
    command: pip install scalarlm
  package: packages/tensorwave-packages.yml#scalarlm
  configuration:
  - env: SCALARLM_API_URL
    description: Base URL of the ScalarLM deployment to talk to.
    example: export SCALARLM_API_URL="http://<your-deployment-ip>"
  usage: scalarlm [-h] {logs,plot,ls,squeue} ...
  commands:
  - command: logs
    description: Stream training/service logs from the deployment.
    backing_operations:
    - getTrainingLogs
    - getServiceLogs
  - command: plot
    description: Plot training metrics.
  - command: ls
    description: List models deployed in the instance.
    backing_operations:
    - listDeployedModels
  - command: squeue
    description: Show the Slurm queue.
    backing_operations:
    - getMegatronSqueue
    - getSlurmSqueue
notes: 'backing_operations are operationIds in openapi/tensorwave-scalarlm-openapi.yml, matched
  by command semantics against the CLI subcommands published in the ScalarLM README. The CLI
  source was not read line by line, so these bindings are semantic rather than verified call
  sites.'