-
Jensen-Shannon divergence for meaningful clustering
Silhouette score validates geometry, not meaning. Using Jensen-Shannon divergence to measure feature distribution divergence bridges the gap between mathematical separation and interpretability.
-
Hybrid intent classification: the rationale for production-grade shallow-model-first architectures
Production chatbots route most requests through fast shallow classifiers, escalating to large language models only on low-confidence queries. This hybrid architecture mitigates the latency and cost overheads of monolithic LLM solutions, achieving significant speed gains while preserving high classification accuracy.
-
Few-shot prompt ordering: the impact of example position
Investigating positional bias in few-shot prompting. While 'Lost in the Middle' suggests boundary importance, the specific ordering of examples remains an important factor for performance stability.
-
Temporal for LLM pipelines: durable execution starter pack
LLM agents often crash, losing state and expensive API work. Temporal provides durable execution for LLM pipelines: automatic state recovery, configurable retries, and long-running orchestration at the cost of determinism constraints and ops overhead.
-
GraphRAG: beyond vector search for connecting the dots
Vector search finds similar text while GraphRAG finds connected facts. A look at the trade-offs, high indexing costs, and lighter-weight alternatives.