#multi-agent

'A hands-on tutorial showing how lightweight Qwen2.5-0.5B-Instruct agents manage ingestion, quality, and infrastructure optimization in multi-agent data pipelines.'

'TUMIX ensembles diverse tool-using agents that share notes and use an LLM judge for early stopping, improving accuracy on hard reasoning benchmarks while cutting inference cost.'