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 +1,32 @@
[
{
"intensional_ratio": 0.3,
"extensional_ratio": 0.5,
"lexical_ratio": 0.2
},
{
"intensional_ratio": 0.45,
"extensional_ratio": 0.35,
"lexical_ratio": 0.2
},
{
"intensional_ratio": 0.3,
"extensional_ratio": 0.4,
"lexical_ratio": 0.3
},
{
"intensional_ratio": 0.5,
"extensional_ratio": 0.3,
"lexical_ratio": 0.2
},
{
"intensional_ratio": 0.25,
"extensional_ratio": 0.5,
"lexical_ratio": 0.25
},
{
"intensional_ratio": 0.35,
"extensional_ratio": 0.35,
"lexical_ratio": 0.3
}
]