Datasets:

Formats:
json
Size:
< 1K
ArXiv:
License:
HaoyuDong commited on
Commit
77975dc
·
verified ·
1 Parent(s): 08e546a

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +12 -7
README.md CHANGED
@@ -57,12 +57,17 @@ This process yields **172 enterprise-grade single- and multi-task workflows** wi
57
 
58
  ## Dataset Structure
59
 
60
- The instruction-tuning corpus is released in **JSONL** format and follows the standard **LLaMA Factory Alpaca** schema.
61
- Each line is one **workflow-centric example**:
62
-
63
- ```json
64
  {
65
- "instruction": "Task instruction for a specific finance & accounting workflow (e.g., 'Reconcile the cash balance and explain the variance using the provided spreadsheet tabs.').",
66
- "input": _to be added_,
67
- "output": _to be added_
 
 
 
 
 
 
 
 
68
  }
 
 
57
 
58
  ## Dataset Structure
59
 
 
 
 
 
60
  {
61
+ "id": "<workflow identifier>",
62
+ "instruction_en": "<English task instruction for a finance & accounting workflow>",
63
+ "source_files": ["<input file name>", "..."],
64
+ "source_files_urls": ["<input file download URL>", "..."],
65
+ "reference_outputs": {
66
+ "files": ["<reference output file name>", "..."],
67
+ "text": "<textual reference output>"
68
+ },
69
+ "reference_file_urls": ["<reference output file download URL>", "..."],
70
+ "task_type": "<task category (e.g., reporting, modeling, cleaning)>",
71
+ "business_type": "<business domain (e.g., budgeting, trading, auditing)>"
72
  }
73
+