Average token usage per workflow node, based on the latest completed run for each paper.
| Node | Papers | Avg Input | Avg Output | Avg Total |
|---|---|---|---|---|
| Code_Availability_Check | 321 | 1,702 (±1,879) | 98 (±41) | 1,800 (±1,893) |
| Code_Embedding | 13 | 33,541 (±33,058) | 237 (±203) | 33,778 (±33,190) |
| Code_Repository_Analysis | 13 | 71,908 (±15,742) | 1,783 (±593) | 73,691 (±16,022) |
| Dataset_Documentation_Check | 2 | 13,372 (±1,694) | 1,714 (±120) | 15,085 (±1,813) |
| Final_Aggregation | 15 | 865 (±80) | 509 (±49) | 1,374 (±60) |
| Paper_Type_Classification | 15 | 4,898 (±485) | 264 (±33) | 5,162 (±501) |
| Reproducibility_Checklist | 15 | 27,611 (±2,804) | 2,643 (±684) | 30,254 (±2,680) |
| Section_Embeddings | 15 | 4,702 (±412) | 0 | 4,702 (±412) |