Table 3 CORAL accuracy across various training dataset sizes. Performance consistently increases with more training data but remains promising even with three orders of magnitude less training data
From: CORAL: COde RepresentAtion learning with weakly-supervised transformers for analyzing data analysis
Model | Number of Cells | |||
|---|---|---|---|---|
1k | 10k | 100k | 1M | |
CORAL | 61.9 | 62.7 | 63.6 | 72.2 |
CORAL (No Masked Attention) | 53.6 | 57.0 | 59.7 | 70.4 |
BERT (AST, No Finetuning) | 41.0 | 52.4 | 63.2 | 67.9 |