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 +1,37 @@
[
{
"min_m": 1,
"max_m": 5,
"invalid_prob": 0.3
},
{
"min_m": 1,
"max_m": 3,
"invalid_prob": 0.3
},
{
"min_m": 1,
"max_m": 7,
"invalid_prob": 0.3
},
{
"min_m": 4,
"max_m": 6,
"invalid_prob": 0.4
},
{
"min_m": 1,
"max_m": 3,
"invalid_prob": 0.4
},
{
"min_m": 1,
"max_m": 3,
"invalid_prob": 0.2
},
{
"min_m": 1,
"max_m": 10,
"invalid_prob": 0.5
}
]