[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"glossary-chain-of-density::en":3,"gloss-cluster-chain-of-density::en":23,"gloss-next-chain-of-density::en":9},{"slug":4,"category":5,"name":6,"definition":7,"meta_desc":8,"faq":9,"schema_markup":9,"related":10},"chain-of-density","prompt-eng","Chain of Density (CoD)","Chain of Density (CoD) is a summarization prompting technique that produces a dense, information-rich summary through several drafting passes at a fixed length. Introduced by Adams et al. (2023), it works like this: the model writes an initial, fairly sparse summary, then repeatedly rewrites it — each round identifying important entities missing from the previous version and weaving them in without making the summary longer. To fit new facts into the same word budget, it fuses clauses, drops filler, and compresses, yielding summaries that pack far more into each sentence. It matters because naive 'summarize this' prompts tend to be vague and front-load the obvious, while CoD forces coverage of specifics. For builders doing digests, release notes, or document previews, it's a recipe for summaries that feel substantive rather than generic. Caveats: pushed too far, density hurts readability and reads as a cramped list of entities — so cap the number of rounds (the original used five) and pick the draft that best balances coverage against fluency.","Chain of Density rewrites a summary several times at a fixed length, adding missing entities each pass — dense information instead of padded generalities.",null,[11,14,17,20],{"slug":12,"name":13},"chain-of-thought-prompting","Chain-of-Thought Prompting",{"slug":15,"name":16},"prompt-chaining","Prompt Chaining",{"slug":18,"name":19},"self-refine","Self-Refine",{"slug":21,"name":22},"summarization","Summarization",[24,28,31,33,36,39,42,45,48,51,54,57],{"slug":25,"category":5,"name":26,"updated_at":27},"analogical-prompting","Analogical Prompting","2026-08-24T02:46:37+00:00",{"slug":29,"category":5,"name":30,"updated_at":27},"automatic-prompt-optimization","Automatic Prompt Optimization",{"slug":12,"category":5,"name":13,"updated_at":32},"2026-08-24T02:46:36+00:00",{"slug":34,"category":5,"name":35,"updated_at":27},"chain-of-verification","Chain-of-Verification",{"slug":37,"category":5,"name":38,"updated_at":32},"chunking","Chunking",{"slug":40,"category":5,"name":41,"updated_at":32},"constrained-decoding","Constrained Decoding",{"slug":43,"category":5,"name":44,"updated_at":32},"context-stuffing","Context Stuffing",{"slug":46,"category":5,"name":47,"updated_at":32},"delimiter","Delimiter",{"slug":49,"category":5,"name":50,"updated_at":27},"directional-stimulus-prompting","Directional Stimulus Prompting",{"slug":52,"category":5,"name":53,"updated_at":27},"emotion-prompting","Emotion Prompting",{"slug":55,"category":5,"name":56,"updated_at":32},"few-shot-prompting","Few-Shot Prompting",{"slug":58,"category":5,"name":59,"updated_at":27},"generated-knowledge-prompting","Generated Knowledge Prompting"]