SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2408.03314 · 2024

Scaling LLM Test-Time Compute Optimally can be More Effective than Scaling Model Parameters

Aviral Kumar, Jaehoon Lee, Kelvin Xu, Charlie Snell ♦

arXiv · PDF · Open in the Atlas

Code that ran

6 of 9 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 5 have no contract verdict; 1 failure of our instrument; 3 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2408.03314.json · badge: /api/ran/2408.03314.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
huggingface/search-and-learn community (classifier label)
classifier basis not recorded on the link
6 ran · 3 no recorded run
FunctionStatusWhere it lives
aggregate_scores Ran
ran on a synthesized input
huggingface/search-and-learn/src/sal/utils/score.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from hmarkc/VG-Search, is pointer only (licence: Apache-2.0) · get_code("e2fa8778aaa14477")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from hmarkc/VG-Search, and Apache-2.0, as GitHub detects it on that repository, is in the permissive set, but a licence decision keeps the body a pointer: its licence file names an outside party, so it may cover vendored code rather than the repository owner's files (recorded as pointer:licence_names_third_party_corporation). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

build_conv Ran
ran on a synthesized input
huggingface/search-and-learn/src/sal/search/utils.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from hmarkc/VG-Search, is pointer only (licence: Apache-2.0) · get_code("fcb58cf3620ba3fa")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from hmarkc/VG-Search, and Apache-2.0, as GitHub detects it on that repository, is in the permissive set, but a licence decision keeps the body a pointer: its licence file names an outside party, so it may cover vendored code rather than the repository owner's files (recorded as pointer:licence_names_third_party_corporation). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

extract_multi_choice_answer Ran
ran; our contract draft was wrong, not the code
huggingface/search-and-learn/src/sal/utils/qwen_math_parser.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from QwenLM/Qwen2.5-Math, is pointer only (licence: NONE) · get_code("ed2f4689128f9885")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from QwenLM/Qwen2.5-Math, and GitHub detected no licence in that repository when it was checked. The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

last Ran
ran on a synthesized input
huggingface/search-and-learn/src/sal/search/utils.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from hmarkc/VG-Search, is pointer only (licence: Apache-2.0) · get_code("0a2729199ea72cd2")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from hmarkc/VG-Search, and Apache-2.0, as GitHub detects it on that repository, is in the permissive set, but a licence decision keeps the body a pointer: its licence file names an outside party, so it may cover vendored code rather than the repository owner's files (recorded as pointer:licence_names_third_party_corporation). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

list_mean Ran
ran on a synthesized input
huggingface/search-and-learn/src/sal/search/utils.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from hmarkc/VG-Search, is pointer only (licence: Apache-2.0) · get_code("e1583186b85bad63")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from hmarkc/VG-Search, and Apache-2.0, as GitHub detects it on that repository, is in the permissive set, but a licence decision keeps the body a pointer: its licence file names an outside party, so it may cover vendored code rather than the repository owner's files (recorded as pointer:licence_names_third_party_corporation). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

subsample_completions Ran
ran on a synthesized input
huggingface/search-and-learn/src/sal/utils/math.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from hmarkc/VG-Search, is pointer only (licence: Apache-2.0) · get_code("940f44544351a853")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from hmarkc/VG-Search, and Apache-2.0, as GitHub detects it on that repository, is in the permissive set, but a licence decision keeps the body a pointer: its licence file names an outside party, so it may cover vendored code rather than the repository owner's files (recorded as pointer:licence_names_third_party_corporation). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. A run is recorded.

batched_math_shepherd_inference Not yet run
unverified (harvested, no recorded run)
huggingface/search-and-learn/src/sal/models/reward_models.py
code served (permissive licence) · get_code("0f5e24b2cb4f615b")
How it ran

no invocation recorded.

extract_completion_answers Not yet run
unverified (harvested, no recorded run)
huggingface/search-and-learn/src/sal/utils/math.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from hmarkc/VG-Search, is pointer only (licence: Apache-2.0) · get_code("3c4ade317b7b432b")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from hmarkc/VG-Search, and Apache-2.0, as GitHub detects it on that repository, is in the permissive set, but a licence decision keeps the body a pointer: its licence file names an outside party, so it may cover vendored code rather than the repository owner's files (recorded as pointer:licence_names_third_party_corporation). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

get_dataset_revisions Not yet run
unverified (harvested, no recorded run)
huggingface/search-and-learn/src/sal/utils/hub.py
pointer only (licence: Apache-2.0); the stored copy get_code serves, from hmarkc/VG-Search, is pointer only (licence: Apache-2.0) · get_code("07e55a87bf057096")
How it ran

The body of this sample is withheld: this paper's link allows the body inline (Apache-2.0), but get_code serves the stored copy, first harvested from hmarkc/VG-Search, and Apache-2.0, as GitHub detects it on that repository, is in the permissive set, but a licence decision keeps the body a pointer: its licence file names an outside party, so it may cover vendored code rather than the repository owner's files (recorded as pointer:licence_names_third_party_corporation). The setup block that called it is withheld with it, because it is written against that body. source_path and origin say where the code lives, to read under the repository's own terms. No invocation recorded.

Also present in other papers' code

Of 9 samples lifted from this paper's linked repositories, 9 carry a function identity and 0 do not. Of the 9 distinct functions with an identity, 8 are also present in other papers whose code we hold (1 with at least one paper dated earlier, 0 the same month, 8 later) and 1 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

FunctionAlso present inEarliest-datedOthers named
extract_multi_choice_answer 12 other papers
1 dated earlier, 10 later, 1 undated
2405.14365 (2024-05)
JiuZhang3.0: Efficiently Improving Mathematical Reasoning by Training Small Data Synthesis Models
2410.07985 (2024-10)
Omni-MATH: A Universal Olympiad Level Mathematic Benchmark For Large Language Models
2502.06773 (2025-02)
On the Emergence of Thinking in LLMs I: Searching for the Right Intuition
and 9 more not named here
aggregate_scores 1 other paper
1 later
2505.11730 (2025-05)
Rethinking Optimal Verification Granularity for Compute-Efficient Test-Time Scaling
build_conv 1 other paper
1 later
2505.11730 (2025-05)
Rethinking Optimal Verification Granularity for Compute-Efficient Test-Time Scaling
extract_completion_answers 1 other paper
1 later
2505.11730 (2025-05)
Rethinking Optimal Verification Granularity for Compute-Efficient Test-Time Scaling
get_dataset_revisions 1 other paper
1 later
2505.11730 (2025-05)
Rethinking Optimal Verification Granularity for Compute-Efficient Test-Time Scaling
last 1 other paper
1 later
2505.11730 (2025-05)
Rethinking Optimal Verification Granularity for Compute-Efficient Test-Time Scaling
list_mean 1 other paper
1 later
2505.11730 (2025-05)
Rethinking Optimal Verification Granularity for Compute-Efficient Test-Time Scaling
subsample_completions 1 other paper
1 later
2505.11730 (2025-05)
Rethinking Optimal Verification Granularity for Compute-Efficient Test-Time Scaling

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Work done during an internship at Google DeepMind Enabling LLMs to improve their outputs by using more test-time computation is a critical step towards building generally self-improving agents that can operate on open-ended natural language. In this paper, we study the scaling of inference-time computation in LLMs, with a focus on answering the question: if an LLM is allowed to use a fixed but non-trivial amount of inference-time compute, how much can it improve its performance on a challenging prompt? Answering this question has implications not only on the achievable performance of LLMs, but also on the future of LLM pretraining and how one should tradeoff inference-time and pre-training compute. Despite its importance, little research attempted to understand the scaling behaviors of various test-time inference methods. Moreover, current work largely provides negative results for a number of these strategies. In this work, we analyze two primary mechanisms to scale test-time computation: (1) searching against dense, process-based verifier reward models; and (2) updating the model's distribution over a response adaptively, given the prompt at test time. We find that in both cases, the effectiveness of different approaches to scaling test-time compute critically varies depending on the difficulty of the prompt. This observation motivates applying a "compute-optimal" scaling strategy, which acts to most effectively allocate test-time compute adaptively per prompt. Using this compute-optimal strategy, we can improve the efficiency of test-time compute scaling by more than 4× compared to a best-of-N baseline. Additionally, in a FLOPs-matched evaluation, we find that on problems where a smaller base model attains somewhat non-trivial success rates, test-time compute can be used to outperform a 14× larger model.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2408.03314")
get_code_for_paper("2408.03314")
have("2408.03314")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2408.03314.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2408.03314.svg)](https://syntology.ai/paper/2408.03314)

Connect an agent — have() is free.