SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2410.07076 · ICLR · 2025

MOOSE-Chem: Large Language Models for Rediscovering Unseen Chemistry Scientific Hypotheses

Wanli Ouyang, Erik Cambria, Tong Xie, Soujanya Poria, Zonglin Yang, Wanhao Liu, Ben Gao, Yuqiang Li, Dongzhan Zhou

arXiv · PDF · Open in the Atlas

Code that ran

4 of 8 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; 1 honoured the contract we drafted, 0 violated it and 0 have no contract verdict; 3 failures of our instrument; 4 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/2410.07076.json · badge: /api/ran/2410.07076.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
ZonglinY/MOOSE-Chem canonical (classifier label)
classifier basis: census
4 ran · 4 no recorded run
FunctionStatusWhere it lives
analysis_list_of_list_of_scores Ran
ran, honoured the contract we drafted
ZonglinY/MOOSE-Chem/Analysis/analysis.py
code served (permissive licence) · get_code("418d4d1894f4dce8")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned tuple. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import numpy as np

# The function takes a list of list of scores
# It converts it to a numpy array and performs operations along different axes
# We need to create a list of lists with numeric values

# Create a simple 2D list of scores (4 samples, 3 score dimensions)
list_of_list_of_scores = [
    [0.1, 0.2, 0.3],
    [0.4, 0.5, 0.6],
    [0.7, 0.8, 0.9],
    [0.2, 0.3, 0.4]
]

ARGS = {
    "list_of_list_of_scores": list_of_list_of_scores
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
exchange_order_in_list Ran
ran; the contract check could not run (RAISES, DEP_MISSING or TIMEOUT)
ZonglinY/MOOSE-Chem/Method/utils.py
code served (permissive licence) · get_code("a46d31273c1cda64")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 2, dtype family int; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Analyzing the function:
# - It takes input_list as parameter
# - It iterates through input_list
# - Each element can be either a list (with exactly 2 elements) or a string
# - If it's a list, it reverses it
# - If it's a string, the entire input_list must have exactly 2 elements, and it reverses input_list

# Test case 1: input_list with lists
input_list_with_lists = [[1, 2], [3, 4], [5, 6]]

# This works: each element is a 2-element list, gets reversed individually

ARGS = {
    "input_list": input_list_with_lists
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 2, dtype family int — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
instruction_prompts Ran
ran; our contract draft was wrong, not the code
ZonglinY/MOOSE-Chem/Method/utils.py
code served (permissive licence) · get_code("f48e1cf015ec00b1")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch

torch.manual_seed(0)

ARGS = {
    'module_name': 'first_round_inspiration_screening',
    'more_info': None
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
recover_raw_background Ran
ran; our contract draft was wrong, not the code
ZonglinY/MOOSE-Chem/Method/utils.py
code served (permissive licence) · get_code("bf19b64cc289243b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Analyzing the function:
# - background_strict_raw: iterable of strings that may have whitespace
# - background_normal: iterable of strings (used when nan_indicator is True)
# - background_strict_raw_nan_indicator: iterable of booleans

# The function iterates through indices, checks if nan_indicator[i] is True,
# and either uses background_normal[i] or background_strict_raw[i] (both stripped)
# It also asserts that len(cur_survey) > 10 when using background_strict_raw

background_strict_raw = [
    "  this is a long survey string  ",
    "  another long survey text here  "
]

background_normal = [
    "  fallback string one  ",
    "  fallback string two  "
]

background_strict_raw_nan_indicator = [
    False,
    True
]

ARGS = {
    "background_strict_raw": background_strict_raw,
    "background_normal": background_normal,
    "background_strict_raw_nan_indicator": background_strict_raw_nan_indicator
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
compare_score_between_gold_insp_and_others Not yet run
unverified (harvested, no recorded run)
ZonglinY/MOOSE-Chem/Analysis/analysis.py
code served (permissive licence) · get_code("6a6bec93afdd68e6")
How it ran

no invocation recorded.

compare_score_between_inter_recom_and_self_explore Not yet run
unverified (harvested, no recorded run)
ZonglinY/MOOSE-Chem/Analysis/analysis.py
code served (permissive licence) · get_code("939c369e4ba796b5")
How it ran

no invocation recorded.

load_title_abstract Not yet run
unverified (harvested, no recorded run)
ZonglinY/MOOSE-Chem/Preprocessing/construct_custom_inspiration_corpus.py
code served (permissive licence) · get_code("78723528e7ed4b7d")
How it ran

no invocation recorded.

setup_logger Not yet run
unverified (harvested, no recorded run)
ZonglinY/MOOSE-Chem/Method/logging_utils.py
code served (permissive licence) · get_code("ff5aa05e7bb75093")
How it ran

no invocation recorded.

Also present in other papers' code

Of 8 samples lifted from this paper's linked repositories, 8 carry a function identity and 0 do not. Of the 8 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 8 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.

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

Scientific discovery plays a pivotal role in advancing human society, and recent progress in large language models (LLMs) suggests their potential to accelerate this process. However, it remains unclear whether LLMs can autonomously generate novel and valid hypotheses in chemistry. In this work, we investigate whether LLMs can discover high-quality chemistry hypotheses given only a research background-comprising a question and/or a survey-without restriction on the domain of the question. We begin with the observation that hypothesis discovery is a seemingly intractable task. To address this, we propose a formal mathematical decomposition grounded in a fundamental assumption: that most chemistry hypotheses can be composed from a research background and a set of inspirations. This decomposition leads to three practical subtasks-retrieving inspirations, composing hypotheses with inspirations, and ranking hypotheses-which together constitute a sufficient set of subtasks for the overall scientific discovery task. We further develop an agentic LLM framework, MOOSE-Chem, that is a direct implementation of this mathematical decomposition. To evaluate this framework, we construct a benchmark of 51 high-impact chemistry papers published and online after January 2024, each manually annotated by PhD chemists with background, inspirations, and hypothesis. The framework is able to rediscover many hypotheses with high similarity to the groundtruth, successfully capturing the core innovations-while ensuring no data contamination since it uses an LLM with knowledge cutoff date prior to 2024. Finally, based on LLM's surprisingly high accuracy on inspiration retrieval, a task with inherently out-of-distribution nature, we propose a bold assumption: that LLMs may already encode latent scientific knowledge associations not yet recognized by humans. 1

For agents

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

get_harvested_code_for_paper("2410.07076")
get_code_for_paper("2410.07076")
have("2410.07076")

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

curl https://syntology.ai/api/ran/2410.07076.json

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

[![Syntology run record](https://syntology.ai/api/ran/2410.07076.svg)](https://syntology.ai/paper/2410.07076)

Connect an agent — have() is free.