SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2602.03554 · ICML · 2026

When Single Answer Is Not Enough: Rethinking Single-Step Retrosynthesis Benchmarks for LLMs

Roman Schutski, Zulfat Miftahutdinov, Maksim Kuznetsov, Alex Aliper, Rim Shayakhmetov, Alex Zhavoronkov, Bogdan Zagribelnyy, Ivan Ilin, Nikita Bondarev, Mathieu Reymond, Thomas Macdougall, Mikolaj Mizera, and 1 more

arXiv · PDF · Open in the Atlas

Code that ran

1 of 5 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 0 have no contract verdict; 1 failure of our instrument; 4 no recorded run), measured 2026-11-22; 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/2602.03554.json · badge: /api/ran/2602.03554.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
pandegroup/reaction_prediction_seq2seq canonical (classifier label)
classifier basis: census
1 ran · 4 no recorded run
FunctionStatusWhere it lives
templatemethod Ran
ran; our contract draft was wrong, not the code
pandegroup/reaction_prediction_seq2seq/seq2seq/graph_utils.py
code served (permissive licence) · get_code("c73e2ed8b047237f")
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 function. 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 tensorflow as tf

def templatemethod(name_):
    """This decorator wraps a method with `tf.make_template`. For example,

  @templatemethod
  def my_method():
    # Create variables
  """

    def template_decorator(func):
        """Inner decorator function"""

        def func_wrapper(*args, **kwargs):
            """Inner wrapper function"""
            templated_func = tf.make_template(name_, func)
            return templated_func(*args, **kwargs)
        return func_wrapper
    return template_decorator


# Test the decorator by applying it to a simple function
@templatemethod("test_template")
def test_func(x):
    return x + 1


# Create setup arguments
ARGS = {
    "name_": "my_template_func"
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
tensorflow — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type function — a function output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a function output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not 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
att_sum_bahdanau Not yet run
unverified (harvested, no recorded run)
pandegroup/reaction_prediction_seq2seq/seq2seq/decoders/attention.py
code served (permissive licence) · get_code("181e72bda062f028")
How it ran

no invocation recorded.

att_sum_dot Not yet run
unverified (harvested, no recorded run)
pandegroup/reaction_prediction_seq2seq/seq2seq/decoders/attention.py
code served (permissive licence) · get_code("af4e61cae8ea2e6d")
How it ran

no invocation recorded.

cross_entropy_sequence_loss Not yet run
unverified (harvested, no recorded run)
pandegroup/reaction_prediction_seq2seq/seq2seq/losses.py
code served (permissive licence) · get_code("358434439625669b")
How it ran

no invocation recorded.

get_dict_from_collection Not yet run
unverified (harvested, no recorded run)
pandegroup/reaction_prediction_seq2seq/seq2seq/graph_utils.py
code served (permissive licence) · get_code("66b405ecdb68e352")
How it ran

no invocation recorded.

Also present in other papers' code

Of 5 samples lifted from this paper's linked repositories, 5 carry a function identity and 0 do not. Of the 5 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 0 later) and 2 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
cross_entropy_sequence_loss 1 other paper
1 dated earlier
1703.03906 (2017-03)
Massive Exploration of Neural Machine Translation Architectures
get_dict_from_collection 1 other paper
1 dated earlier
1703.03906 (2017-03)
Massive Exploration of Neural Machine Translation Architectures
templatemethod 1 other paper
1 dated earlier
1703.03906 (2017-03)
Massive Exploration of Neural Machine Translation Architectures

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

Recent progress has expanded the use of large language models (LLMs) in drug discovery, including synthesis planning. However, objective evaluation of retrosynthesis performance remains limited. Existing benchmarks and metrics typically rely on published synthetic procedures and Top-K accuracy based on single groundtruth, which does not capture the open-ended nature of real-world synthesis planning. We propose a new benchmarking framework for singlestep retrosynthesis that evaluates both generalpurpose and chemistry-specialized LLMs using ChemCensor, a novel metric for chemical plausibility. By emphasizing plausibility over exact match, this approach better aligns with human synthesis planning practices. We also introduce CREED, a novel dataset comprising millions of ChemCensor-validated reaction records for LLM training, and use it to train a model that improves over the LLM baselines under this benchmark.

For agents

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

get_harvested_code_for_paper("2602.03554")
get_code_for_paper("2602.03554")
have("2602.03554")

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

curl https://syntology.ai/api/ran/2602.03554.json

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

[![Syntology run record](https://syntology.ai/api/ran/2602.03554.svg)](https://syntology.ai/paper/2602.03554)

Connect an agent — have() is free.