claim
pending-review
claim:logarithm-transformation-is-simpler-and-more-effective-than-learnable-loss-transformationLogarithm transformation is simpler and more effective than learnable loss transformation
lin-2023-dual-balancing.mdFrontmatter (10 fields)
{
"doc": "lin-2023-dual-balancing.md",
"context": "Compared to IMTL-L: parameter-free, no extra computational cost, achieves same theoretical goal",
"category": "ai",
"norm_label": "Logarithm transformation is simpler and more effective than learnable loss transformation",
"graphify_id": "claim_log_transformation_superiority",
"source_file": "lin-2023-dual-balancing.md",
"imported_from": "/Users/antonborzov/Documents/Research.nosync/papers/extract_typed_out/lin-2023-dual-balancing/graph.json",
"extracted_type": "claim",
"source_location": "§3.1",
"graphify_file_type": "claim"
}Outgoing (0)
None.
Incoming (1)
Supported by (1)
Mentions (1)
- papers-typed
lin-2023-dual-balancing.md