FlowingFlowing
← Back to all development cases
Development case · blind model judgment pending independent human review; not a formal Benchmark conclusion.
AutocompleteEvidence version preferred30 / 162 · 7b1f0f7d85f4bc3a

From Words to Molecules: A Survey of Large Language Models in Chemistry

Quantitative biology · 2402.01439v1

FLOWING EVIDENCE BENCHMARK

How do we tell whether Evidence helps?

At the same writing position, with the same model and task, how does supplying retrieved paper passages change the first output? We compare matched versions and retain ties, unusable outputs and incomplete reviews.

SAME DRAFT · EVIDENCE ON OR OFF

01 · FIXED WRITING POSITION

MANUSCRIPT

Same writing position ▌

Matched autocomplete at the same draft position

02 · TWO MATCHED INPUTS

SHARED BY BOTH

Manuscript context, model, task and prompt

A · With Evidence

Retrieved paper passages supplied

B · Without Evidence

No retrieved passages supplied

LLM

Same model and version

A → first output

B → first output

Blind judge agent

First outputs are anonymized as X and Y

Continuation review: accuracy, fit to the writing task and usability
Returns: X preferred / tie / Y preferred / both unusable

Order check: X / Y → Y / X

HOW DOES BLIND JUDGING WORK?

① Anonymize both outputs
The judge sees the same draft and both first outputs without knowing which received Evidence.

② Compare and swap order
The judge applies task-specific criteria in X/Y and then Y/X order.

③ Review disagreements
A third pass resolves disagreements. Incomplete reviews remain in the denominator.

How are autocomplete positions stratified?

Before seeing generation outcomes, we check whether a retrieved passage contains a specific proposition that directly supports the next writing move. Those positions appear in the left opportunity group; the rest are ordinary positions on the right. We select a balanced sample from admitted papers in each field. The 50/50 split is experimental, not a measure of how often either type occurs in writing.

AUTOCOMPLETE · 110

Biology, statistics and astrophysics

55 positions on each side; statistics uses the ten-paper rerun.

AUTOCOMPLETE · 52

Psychology and climate science

26 positions on each side; psychology includes nine papers and climate science four.

How are the table percentages calculated?

Across five fields, 51 of 81 left-column positions preferred the Evidence version. Ties, pairs where both versions were unusable and incomplete reviews remain in the denominator.

51Evidence version preferred
÷
81All positions in this group
=
63%Evidence preference in this group

Source contribution is a separate review: 16 of 22 Evidence wins entered into source review directly used retrieved papers; another 29 wins await review.

These are development-stage model judgments pending independent human review. They are not formal Benchmark conclusions and do not, on their own, establish causality.

Manuscript writing position

Text excerpt · not a PDF page

Research manuscript · excerpt

From Words to Molecules: A Survey of Large Language Models in Chemistry

1 Introduction

…based on the domain and modality of their input data, and discuss approaches for integrating these inputs for LLMs. Furthermore, this paper delves into the pretraining objectives with adaptations to chemical LLMs. After that, we explore the diverse applications of LLMs in chemistry, including novel paradigms for their application in chemistry tasks. Finally, we identify promising research directions, including further integration with chemical knowledge, advancements in continual learning, and improvements in model interpretability, paving the way for groundbreaking developments in the field.

Humans understand and describe their environment using natural language, which reflects the complexity of human thought. The emergence of Large Language Models (LLMs) marks a significant advancement in artificial intelligence, showcasing remarkable abilities in various domains. These models excel at understanding and generating complex text, making them crucial for tasks that demand deep textual analysis and creation.

↓ Next paragraph was generated here

The highlighted text is near the writing position. This is a layout of the evaluation input, not a PDF screenshot.

A matched comparison at the same writing positionJudgment: Evidence version preferred

Both arms use the same model, task and manuscript context. Only the retrieved passages differ. The first recorded output from each arm is shown below.

Evidence

Same model and position · retrieved passages supplied · first output

Preferred by blind review

Unlike earlier task-specific models, LLMs acquire these capabilities through pretraining on large, diverse corpora, which allows them to generalize across writing styles, languages, and domains rather than remaining confined to a single well-defined task. This breadth of pretraining is what makes them attractive as a foundation for chemistry, where inputs range from natural-language descriptions to molecular representations.

Passages supplied to the Evidence version

Harnessing the Power of LLMs in Practice: A Survey on ChatGPT and Beyond

Handling such real-world scenarios requires coping with ambiguity, understanding context, and handling noisy input. Compared to fine-tuned models, LLMs are better equipped for this because they have been trained on diverse data sets that encompass various writing styles, languages, and domains. Addi…

Read full passage excerpt

Handling such real-world scenarios requires coping with ambiguity, understanding context, and handling noisy input. Compared to fine-tuned models, LLMs are better equipped for this because they have been trained on diverse data sets that encompass various writing styles, languages, and domains. Additionally, LLMs demonstrate a strong ability to generate open-domain responses, making them well-suited for these scenarios. Fine-tuned models, on the other hand, are often tailored to specific, well-defined tasks and may struggle to adapt to new or unexpected user requests.

No Evidence

Same model and position · no retrieved passages · first output

Not preferred in this comparison

The intersection of chemistry and natural language processing has become an increasingly active area of research, as chemical knowledge is expressed not only through structured representations such as molecular graphs and SMILES strings but also through the vast body of scientific text in the literature. This dual nature of chemical information motivates the adaptation of general-purpose language models to chemical domains.

Blind model judgment

The blinded model review preferred the Evidence version. Its original rationale appears below.

Read the original model judgment

Output A is fully grounded: it draws directly from the source's comparison of LLMs versus fine-tuned models (diverse data sets, various writing styles/languages/domains, open-domain vs. specific well-defined tasks) and connects this to the draft's chemistry context without adding unsupported claims. Output B introduces an unsupported claim: it states that 'chemical knowledge is expressed not only through structured representations such as molecular graphs and SMILES strings' — but neither the draft nor the source mentions molecular graphs, SMILES strings, or structured representations of chemical knowledge. The source only discusses text-based LLM capabilities, and the draft's prior paragraph mentions only 'domain and modality of their input data' without specifying what those modalities are. The claim about SMILES strings and molecular graphs is a material factual claim about chemical information representation that lacks closed-book support.

The passages are input material; their presence does not mean an output used them or that they support every claim. Draft and source passages are excerpted; no simulated scores or PDF appear here.