OGBN-ARXIV

anchor

Findings

• IC-996GPT-3.5 achieves 73.5% zero-shot accuracy on OGBN-ARXIV 40-class node classification and 73.56% on TAPe-ARXIV23, a dataset of papers published after its knowledge cutoff [eval]
• IC-997LLaMA-2-13B-Chat achieves 44.23% zero-shot accuracy on OGBN-ARXIV, substantially below GPT-3.5's 73.5% [eval]
• IC-998GPT-3.5's zero-shot accuracy on OGBN-ARXIV depends on the position of the title relative to the abstract in the prompt: 0.720 when abstract precedes title, 0.695 when title precedes abstract [eval]