issues Search Results · language:Dune language:Python linked:pr language:Java language:Java language:JavaScript linked:pr
Filter by
5.8M results
Context
- Issue #874 landing repair sweep 결과, weak chord-tone landing risk 6 - 0
- objective outside-soloing risk count: 5
- outside-soloing residual risk after landing repair: 2
- landing repair ...
ohhalim
Description
The prompts/ directory at the project root is a design-document archive containing two Markdown feature-plan files:
- prompts/feat-animal-and-livestock-awareness.md
- prompts/feat-colony-foundation-and-milestones.md ...
sshcrack
배경
#182 버그 수정 후에도 vision 필드가 null로 반환되는 원인 추가 분석.
원인
Spring Boot의 spring.jpa.open-in-view 기본값이 true(OSIV 활성화).
OSIV가 활성화되면 하나의 HTTP 요청 전체 라이프사이클 동안 EntityManager가 공유됨. endSession()에서 처음 로드한 StudySession ...
bug
Silverblond
Surfaced by running the #840 activation CLI (python -m src.mlops.prediction_synthesizer_deploy --verify) on prod after
deploying main (ad8c9b99).
Gap 1 — MLModelRegistryRepository.get_models_for_target ...
enunezvn
问题描述
CatBoost early stopping 的 eval_set 设置不当：
- 模型一/二：eval_set=(X_train_factors[-6:], y_train[-6:]) — 验证集是训练集的子集，早停失去 验证 意义
- 模型三融合层：eval_set=(fusion_train[-3:], y_train[seq_len:][-3:]) — 验证集仅 3 ...
bug
lilacseeking
Action: research Target: Re-evaluate next priorities given current code state — find a new angle
Created by kaizen delivery.
kaizen:running
SevFle
问题 A: LSTM 训练未记录收敛 epoch
train_lstm_model() 实现了 early stopping (patience=30)，但训练循环结束后未记录最终 epoch 数。日志中完全看不到每个 LSTM 在多少轮后停止，无法确认是否真正收敛。
建议：在 break 前添加 logger.debug 记录 epoch 数
问题 B: 未使用的变量
1. main() ...
enhancement
问题描述
避雷器 sheet 中的台风(分)列：36 个月中 34 个月为 0，仅 2 个月（2023年7月、2024年8月）为 1。
影响
94.4% 为零值的特征对 CatBoost/LSTM 训练的贡献几乎为零。作为避雷器的 top-4 影响因子之一，该特征无法为模型提供有效信息。
同时发现
- 负荷增长量(分) 与 雷击(分) 的相关系数高达 0.823，存在严重多重共线性
...
enhancement
What why
All persistence today is raw sqlite3 with hand-written SQL and a per-module _conn() context manager (no shared base). We
re standardizing on SQLModel (SQLAlchemy + Pydantic) so DB access is ...
area:storage
in-progress
needs-local-verification
ready
himanshu22500
问题描述
对 data.xlsx 的审查发现多条数据质量问题：
1. 需求量零值过于规律：固定月份必为零（如电缆的 1/2/12月），三年完全相同
2. 工程投资量零值严格周期：每 4 个月一次零值，人为痕迹明显
3. 10KV电缆 2022年需求异常下降：3月→11月从 22 单调递减至 18，但 2023/2024 年同期维持在 25
4. 负荷增长量跨 3 个 Sheet 完全相同：现实中不同物资的负荷增长速率应不同 ...
enhancement
Learn how you can use GitHub Issues to plan and track your work.
Save views for sprints, backlogs, teams, or releases. Rank, sort, and filter issues to suit the occasion. The possibilities are endless.Learn more about GitHub IssuesProTip! Restrict your search to the title by using the in:title qualifier.
Learn how you can use GitHub Issues to plan and track your work.
Save views for sprints, backlogs, teams, or releases. Rank, sort, and filter issues to suit the occasion. The possibilities are endless.Learn more about GitHub IssuesProTip! Restrict your search to the title by using the in:title qualifier.