Batch inference

Partition a large input, infer concurrently, and merge traceable predictions.

load model→partition input→infer shard Ainfer shard Binfer shard N→merge predictions→publish

Batch inference

Activity responsibilities

Activity	Responsibility
load-model	Execute the load model stage and publish its declared outputs for downstream activities.
partition-input	Execute the partition input stage and publish its declared outputs for downstream activities.
infer-shard-a	Execute the infer shard A stage and publish its declared outputs for downstream activities.
infer-shard-b	Execute the infer shard B stage and publish its declared outputs for downstream activities.
infer-shard-n	Execute the infer shard N stage and publish its declared outputs for downstream activities.
merge-predictions	Execute the merge predictions stage and publish its declared outputs for downstream activities.
publish	Execute the publish stage and publish its declared outputs for downstream activities.

Inputs

• Approved model
• Input dataset
• Inference configuration

Outputs

• Prediction shards
• Merged predictions
• Completeness report

Provenance to preserve

• Shard boundaries
• Model digest
• Per-shard logs

Runnable example

This is the complete checked-in bundle for this pattern. Download the environment, scope, topology, workflow, input, container recipe, runner, and validation contract from this page before executing it.

Read setup instructions ↓Environment YAML ↓Execution scope YAML ↓Network topology YAML ↓Workflow YAML ↓Input data ↓Container recipe ↓Run the workflow ↓Validate outputs ↓Expected output manifest ↓

Verified local execution

These captures and the output manifest were produced by the fixture's local Docker run and validator. They are published with the same bundle as the runnable files.

Workflow evidence

Execution evidence

Output evidence

Open verified output manifest ↓

Execution considerations

Place shards according to memory, accelerator availability, and data locality. The merge stage must detect missing shards.

AkôFlow boundary: the engine schedules, deploys, executes, transfers data, and records evidence. The ML or agent framework remains an implementation choice inside each activity.