ritik12 commited on
Commit
4205985
·
verified ·
1 Parent(s): 811083c

Add eval.yaml: register WaterfallBench as a Hub benchmark (Harbor)

Browse files
Files changed (1) hide show
  1. eval.yaml +22 -0
eval.yaml ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ name: WaterfallBench
2
+ description: >
3
+ WaterfallBench tests whether a model can execute an asset-backed securities waterfall to the cent from the deal
4
+ documents alone. Each of the 20 tasks is a real auto-loan or lease securitization at its first distribution date:
5
+ the final prospectus, the closing agreements and that month's servicer report with every waterfall result removed
6
+ (7-11 SEC EDGAR filings, 500K-740K tokens). The model returns the servicing fee, each note class's interest,
7
+ principal and ending balance, the reserve balance and the residual in integer cents, graded exactly against what
8
+ the trustee paid. Tasks run as Harbor tasks; answer keys are private. Scores are percentages (0-100).
9
+ evaluation_framework: harbor
10
+
11
+ leaderboard:
12
+ base_models_only: false
13
+
14
+ tasks:
15
+ # Primary metric: share of tasks where every graded field is exact to the cent.
16
+ - id: fully_resolved
17
+ config: default
18
+ split: test
19
+ # Share of non-trivial graded fields exactly right.
20
+ - id: field_accuracy
21
+ config: default
22
+ split: test