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 +1,38 @@
[
{
"max_s_length": 8,
"min_s_length": 4,
"max_queries": 3,
"min_queries": 1
},
{
"max_s_length": 7,
"min_s_length": 3,
"max_queries": 3,
"min_queries": 1
},
{
"max_s_length": 10,
"min_s_length": 5,
"max_queries": 5,
"min_queries": 1
},
{
"max_s_length": 6,
"min_s_length": 3,
"max_queries": 5,
"min_queries": 2
},
{
"max_s_length": 15,
"min_s_length": 8,
"max_queries": 7,
"min_queries": 2
},
{
"max_s_length": 20,
"min_s_length": 12,
"max_queries": 8,
"min_queries": 3
}
]