Dataset Viewer
Auto-converted to Parquet Duplicate
paper
large_stringclasses
5 values
date
int64
20.2M
20.2M
old
large_stringclasses
1 value
new
large_stringclasses
2 values
correct
bool
1 class
repro
bool
2 classes
PMC3274599
20,151,125
GO:0035195
GO:0035279
true
false
PMC4113387
20,170,717
GO:0035195
GO:0035195
true
true
PMC4737258
20,161,116
GO:0035195
GO:0035279
true
false
PMC3467063
20,230,808
GO:0035195
GO:0035195
true
true
PMC6854602
20,230,712
GO:0035195
GO:0035195
true
true

GOFlowLLM development and evaluation datasets

These are the datasets used to develop and evaluate GOFlowLLM. You will find the splits used in the relevant subdirectory. There are two types of dataset here, the three on which we have some expert assessment (Development, Historical & Modern) and the one with just the PMCID, RNA ID, GO term and target, with no expert assessment.

Development

Five articles used to develop the flowchart prompts, alongside expert guidance

Historical

10 articles curated before 2022, when the curation guidelines were different.

Modern

5 articles curated after 2022, using the modern curation guidelines.

All

This is the full set of 157 open access papers for which we have used existing curation to benchmark GOFlowLLM

Downloads last month
3