Tag
The author discusses challenges in testing LLM-based agent pipelines where outputs vary with similar prompts, advocating for property-based checks over exact matches to handle non-deterministic outputs.
This paper models hallucination propagation in multi-agent LLM pipelines as a Markov process, showing that errors become less detectable across stages and proposing early verification to reduce survival rates.
Datatrove, the data-processing library behind FineWeb datasets, released version 0.10.0 with new features like JobsPipelineExecutor for running pipelines on Hugging Face Jobs and support for HF storage buckets.
The article investigates whether recurring LLM workloads can be replaced by automatically synthesized deterministic pipelines of typed ML/NLP operators, and asks for feedback on feasibility and approaches.
This paper presents adaptive pipelines for legal retrieval, entailment, and judgment prediction tasks in the COLIEE 2026 competition, using multi-stage retrieval, reranking, and LLM-based reasoning.
The article explains why F# is a great choice for scripting and automation tasks, highlighting its helpful type system and natural pipeline operator that make code predictable and reliable.