Pith. sign in

REVIEW 9 cited by

Are More LLM Calls All You Need? Towards Scaling Laws of Compound Inference Systems

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2403.02419 v2 pith:3WVIZI3Z submitted 2024-03-04 cs.LG cs.AIcs.CLcs.SYeess.SY

classification cs.LGcs.AIcs.CLcs.SYeess.SY
keywords callsperformancenumbercompoundsystemsscalingvotefilter-vote
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Many recent state-of-the-art results in language tasks were achieved using compound systems that perform multiple Language Model (LM) calls and aggregate their responses. However, there is little understanding of how the number of LM calls - e.g., when asking the LM to answer each question multiple times and taking a majority vote - affects such a compound system's performance. In this paper, we initiate the study of scaling properties of compound inference systems. We analyze, theoretically and empirically, how the number of LM calls affects the performance of Vote and Filter-Vote, two of the simplest compound system designs, which aggregate LM responses via majority voting, optionally applying LM filters. We find, surprisingly, that across multiple language tasks, the performance of both Vote and Filter-Vote can first increase but then decrease as a function of the number of LM calls. Our theoretical results suggest that this non-monotonicity is due to the diversity of query difficulties within a task: more LM calls lead to higher performance on "easy" queries, but lower performance on "hard" queries, and non-monotone behavior can emerge when a task contains both types of queries. This insight then allows us to compute, from a small number of samples, the number of LM calls that maximizes system performance, and define an analytical scaling model for both systems. Experiments show that our scaling model can accurately predict the performance of Vote and Filter-Vote systems and thus find the optimal number of LM calls to make.

Discussion (0). Sign in to comment.

Forward citations

Cited by 9 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Safety Under Scaffolding: How Evaluation Conditions Shape Measured Safety

    cs.SE 2026-03 conditional novelty 7.0 of 10

    Map-reduce scaffolding degrades measured safety mainly by stripping multiple-choice options (40–89% of the loss is format conversion); scaffold architecture explains only 0.4% of variance and composite safety scores h...

  2. DeepLook: Deeper Thinking with Lookahead

    cs.AI 2026-06 conditional novelty 6.0 of 10

    A training-free decoder that branches only at low-confidence reasoning segments and ranks continuations by average lookahead confidence reports equal or better accuracy at roughly 1/8 of a confidence-filtered baseline...

  3. Murakkab: Resource-Efficient Agentic Workflow Orchestration in Cloud Platforms

    cs.MA 2025-08 conditional novelty 6.0 of 10

    Murakkab uses declarative workflow specs and a profile-guided MILP optimizer to reduce GPU, energy, and cost for agentic workflow serving while meeting percentile-defined SLOs.

  4. BEST-Route: Adaptive LLM Routing with Test-Time Optimal Compute

    cs.LG 2025-06 conditional novelty 6.0 of 10

    A routing system that chooses both the model and the number of samples per query to meet a quality threshold, yielding up to 60% cost savings.

  5. TongSearch-QR: Reinforced Query Reasoning for Retrieval

    cs.IR 2025-06 conditional novelty 6.0 of 10

    TongSearch-QR trains 1.5B and 7B models with GRPO and a frozen-embedding reward to rewrite queries, reaching 27.9 nDCG@10 on BRIGHT with BM25, above GPT-4o's 26.5.

  6. Adaptive Graph Pruning for Multi-Agent Communication

    cs.CL 2025-06 conditional novelty 6.0 of 10

    AGP trains a graph neural network to jointly decide which agents to keep and how strongly they should communicate, and reports state-of-the-art average accuracy across six LLM benchmarks with large token savings.

  7. OSC: Cognitive Orchestration through Dynamic Knowledge Alignment in Multi-Agent LLM Collaboration

    cs.AI 2025-09 reject novelty 5.0 of 10

    OSC uses learned Collaborator Knowledge Models and RL-trained communication policies to make LLM agents communicate adaptively, claiming gains on AlpacaEval 2.0 and MT-Bench.

  8. OpenCodeReasoning-II: A Simple Test Time Scaling Approach via Self-Critique

    cs.CL 2025-07 conditional novelty 5.0 of 10

    A 2.5M-example code reasoning dataset with critique traces enables Qwen2.5-based models to surpass prior open-weight distilled models on LiveCodeBench via test-time self-critique selection.

  9. ReasonBridge: Efficient Reasoning Transfer from Closed to Open-Source Language Models

    cs.AI 2025-06 conditional novelty 4.0 of 10

    ReasonBridge distills 1,000 curated reasoning traces from a closed model into open-source models via hierarchical adapters and guided inference, claiming up to 23% benchmark gains.

Pith tools