-
fix typos
-
fix typos
-
fix typos and escaped characters
-
fix hash
-
split with a diff hash
-
using hash based dataset split
-
Remove old 'train' split, keeping only train_sft/train_rl/test
-
Upload dataset
-
Upload dataset
-
Upload dataset
-
Upload dataset
-
fix: remove trailing quotes from group_linking_terms
-
feat: rebalance all data with stratified split by difficulty (bins: 0-2, 2-3, 3-4, 4-4.5, 4.5-5) into train_sft (10%), train_rl (80%), and test (10%)
-
feat: reseed training data and split into train_sft (10%) and train_rl (90%)
-
fix: restore original train/test split
-
Removed trailing quotes from group_themes
-
Clean dataset: Remove 13 URL puzzles, fix 2 backtick typos, convert puzzle_id to string
-
Upload dataset
-
Upload dataset
-
Upload dataset
-
Upload dataset
