Miscellaneous Commands

These helpers surface project metadata and quick diagnostics that complement the major chunk/infer sub-commands.

badc version

Print the installed package version (badc.__version__).

Usage:

badc version

Typical output:

─────────────────── Bird Acoustic Data Compiler ───────────────────
BADC version: 0.1.0

badc gpus

Report GPUs visible via nvidia-smi. The command is a thin wrapper around badc.gpu.detect_gpus so its output mirrors what the inference scheduler will see.

Usage:

badc gpus

Sample output:

Detected GPUs:
 - #0: NVIDIA A100-SXM4-80GB (81251 MiB)
 - #1: NVIDIA A100-SXM4-80GB (81251 MiB)

Troubleshooting:

  • If the list is empty, confirm nvidia-smi exists on PATH and the container/VM exposes GPUs.

  • On CPU-only nodes, use badc infer run --cpu-workers N to request additional CPU threads (BADC still runs at least one CPU worker automatically when GPUs are unavailable).

Help excerpt

$ badc gpus --help
Usage: badc gpus [OPTIONS]
  Display GPU inventory as reported by `nvidia-smi`.
Options:
  --help  Show this message and exit.

badc telemetry

Tail the most recent telemetry entries (defaults to data/telemetry/infer/log.jsonl).

Usage:

badc telemetry [--log PATH]

Option reference

Option / Argument

Description

Default

--log PATH

Telemetry JSONL to read (supports arbitrary pipelines).

data/telemetry/infer/log.jsonl

Help excerpt

$ badc telemetry --help
Usage: badc telemetry [OPTIONS]
  Print the most recent telemetry records for quick inspection.
Options:
  --log PATH  Telemetry log path.  [default: data/telemetry/infer/log.jsonl]
  --help      Show this message and exit.

Each line prints status, chunk id, GPU index, timestamp, and runtime seconds. Example:

Telemetry records (3):
[success] GNWT-290_chunk_1 (GPU 0) 2025-12-06T18:22:11 runtime=12.4

Use this command when monitoring SLURM jobs or local batches to confirm progress without opening the JSONL file manually. Point --log at alternate telemetry files (e.g., chunking or aggregation pipelines) as those land. See Aggregate detections and monitor telemetry for a telemetry + aggregation walkthrough.