Fix Dataset Viewer: remove all non-JSONL JSON files
ismail.tasdelen
8521039
HF worker scans all JSON files in repo and tries to cast them to dataset
schema, causing CastError. Renamed all non-JSONL JSON files to .txt:
- manifest.json → manifest.txt
- statistics/summary.json → statistics/summary.txt
- evaluation/rubrics/rubric_v1.json → evaluation/rubrics/rubric_v1.txt
- validation/validation_summary.json → validation/validation_summary.txt
Also removed dataset_info from README.md and added .hfignore.
Add dataset viewer note to appsecbench.py docstring
ismail.tasdelen
323f2c6
Remove dataset_info block entirely - HF auto-infers features from JSONL
ismail.tasdelen
06c53c2
Previous attempt kept partial dataset_info which caused CastError due to
column name mismatches. Removing it completely lets Hugging Face infer
all features (including list/dict types) directly from JSONL files.
Fixes: CastError - column names don't match
Fix Dataset Viewer: remove dict/list dtype features from dataset_info
ismail.tasdelen
99440a7
scoring_criteria, tags, references (list dtype), evaluation_rubric, metadata (dict dtype) kaldırıldı.
Hugging Face bunları JSONL dosyalarından otomatik olarak Sequence/Struct olarak çıkarır.
Bu değişiklik Dataset Viewer hatasını düzelttiyor.
Fixes: TypeError: list_() takes at least 1 positional argument (0 given)
