Compare registrations with papers.

Note: this app is in beta. It has not yet been extensively tested.

Choose A Parser

This determines what method RegCheck uses to read your paper.

If the paper is already typeset to academic journal format, GROBID will perform best.
For other document types, DPT-2 will perform best.

Choose an LLM

This determines which LLM is used for comparing text.

ChatGPT is generally the most accurate (but can be slower with higher reasoning effort).
Llama 3.3b is typically fastest. DeepSeek can perform well but may be slower.
Exact model versions are configured on the server.

Select Comparison Dimensions

	Dimension	Definition

Append previous outputs?

Should RegCheck include earlier dimension responses when prompting the model for later dimensions?

Selecting "Yes" will generally lead to better performance.
Selecting "No" keeps outputs independent, which might be desirable for research purposes.

Does the paper have multiple experiments?

Giving this information helps RegCheck to know which study specifically to look at.

Is your study registered on ClinicalTrials.gov?

Tell us where to look for your registration.

If you select "Yes", we will fetch the registration directly from ClinicalTrials.gov. Otherwise, please upload your preregistration file.

Upload files

Preregistration:

Choose file No file chosen

Supported file types:
.pdf, .docx, and .txt

Paper:

Choose file No file chosen