refactor(data_pipeline): optimize data generation pipeline; add multiple preset configurations for data generation

This commit is contained in:
chenyongkang 2025-08-26 18:40:21 +08:00
parent 1a8477c8d8
commit 8d493b35a0
2160 changed files with 69199 additions and 154 deletions

View file

@ -0,0 +1,47 @@
[
{
"n_min": 5,
"n_max": 10,
"k_min": 1,
"k_max": 5,
"subset_size_min": 1,
"subset_size_max": 3,
"max_retries": 1000
},
{
"n_min": 15,
"n_max": 20,
"k_min": 10,
"k_max": 15,
"subset_size_min": 3,
"subset_size_max": 5,
"max_retries": 1000
},
{
"n_min": 5,
"n_max": 20,
"k_min": 3,
"k_max": 10,
"subset_size_min": 1,
"subset_size_max": 5,
"max_retries": 1000
},
{
"n_min": 5,
"n_max": 10,
"k_min": 1,
"k_max": 3,
"subset_size_min": 1,
"subset_size_max": 2,
"max_retries": 1000
},
{
"n_min": 10,
"n_max": 15,
"k_min": 5,
"k_max": 10,
"subset_size_min": 2,
"subset_size_max": 4,
"max_retries": 1000
}
]