同步代码更新 (2026-06-16)

- 移除 generate_diverse_answers.py(多样化回答已废止)
- 更新 generate_fresh_qa.py(目标50万条新问答)
- 更新 generate_multiturn.py(6~12轮长对话,目标50万)
- 更新 regenerate_all_high_temp.py(独立保存,不追加训练集)
- 更新 .gitignore(忽略新输出目录)
- 所有API Key改为环境变量读取
This commit is contained in:
shuiyu1123 committed 2026-06-16 11:48:30 +00:00
1 parent eb1d62f85c
commit 91db488ad4
5 files changed
+21 -291

No files matched your search

+2 -2
View File
@@ -171,8 +171,8 @@ def main():
print(f"去重指纹池: {len(existing_fps)} 个")
# ====== 参数 ======
TARGET = 100000 # 目标 10 万条全新问答
BATCH_SIZE = 10 # 每批次生成 10 条
TARGET = 500000 # 目标 50 万条全新问答
BATCH_SIZE = 15 # 每批次生成 15 条(更多问题,效率更高)
BATCHES_PER_SAVE = 20 # 每 20 批次保存一次
MAX_WORKERS = 5 # Astron 5并发