poprox_recommender.evaluation.generate.worker

Functions

cluster_recommend(pipeline, profiles, outs, ...)
dynamic_remote(task_or_actor)	Dynamically configure the resource requirements of a task or actor based on CUDA availability and parallelism configuration.
generate_profile_recs(dataset, outs, pipeline)
recommend_batch(pipeline, batch, writers)	Batch-recommend function, to be used as a Ray worker task.
recommend_for_profile(pipeline, request)	Generate recommendations for a single request, returning the pipeline state.
serial_recommend(pipeline, profiles, outs, pb)

poprox_recommender.evaluation.generate.worker.recommend_for_profile(pipeline, request)

Generate recommendations for a single request, returning the pipeline state.

Parameters:

• pipeline (str)

• request (RecommendationRequestV1)

Return type:

PipelineState

poprox_recommender.evaluation.generate.worker.recommend_batch(pipeline, batch, writers)

Batch-recommend function, to be used as a Ray worker task.

Parameters:

• batch (list[RecommendationRequestV1])

• writers (list[RecommendationWriter])

poprox_recommender.evaluation.generate.worker.dynamic_remote(task_or_actor)

Dynamically configure the resource requirements of a task or actor based on CUDA availability and parallelism configuration.