test time compute allocation

1 articles · 5 co-occurring · 0 contradictions · 0 briefs

RLMs are a concrete implementation of allocating computation at inference time based on learned importance, similar to reasoning-time scaling discussed in recent LLM research.

Related concepts

tool use external function calling 1 tool integration patterns 1 rag 1 prompt engineering 1 context window management 1

Evidence chain (1 articles, showing 1)

@raw_works: Reasoning models were the first clear proof that language model capability ca... example_of

RLMs are a concrete implementation of allocating computation at inference time based on learned importance, similar to reasoning-time scaling discussed in recent LLM research.

query this concept

$ db.articles("test-time-compute-allocation")

$ db.cooccurrence("test-time-compute-allocation")

$ db.contradictions("test-time-compute-allocation")