Batch Inference

Summarize a batch of documents

Run offline batch summarization over a set of documents (reports, transcripts, articles), producing one summary per document in a single job.

Run this jobGet help running this

Estimated $0.005–$0.03 per document, depending on length · Scales with document count and length

What inputs you need

  • Document files or JSONL of text
  • Summary length/style instructions
  • Model choice

What Badgr returns

  • JSONL of document → summary pairs
  • Per-document run log
  • Single receipt on completion

Recommended GPU routes

A100 40GBA100 80GBL40S 48GB

Estimated cost: $0.005–$0.03 per document, depending on length · Estimated runtime: Scales with document count and length

Example command

badgr run "python batch_summarize.py --input docs.jsonl --style bullet" \
  --gpu A100_40GB \
  --max-cost 15 \
  --max-runtime 90

Common setup failures Badgr avoids

Long documents truncated silently — job checks document length against the model's context window and routes long documents to a higher-VRAM GPU.

Job stops partway through a large document set — progress is checkpointed per document so a restart resumes instead of reprocessing.

Ready to run this?

Launch from the dashboard, CLI, or Compute API. Max-cost protection included.

Run this jobBack to Batch Inference