dosubot[bot]

You don’t need to fine-tune the gwen3:8b model to use RAGFlow as a legal assistant—RAGFlow is designed to work by retrieving relevant content from your uploaded PDFs (using mxbai-embed-large for...

For the most accurate results as a legal assistant in RAGFlow, focus on prompt engineering and document chunking. Use the mxbai-embed-large embedding model and set your chunk size to 256...

Automatic keyword and question generation in RagFlow is handled in the chunk processing workflow, mainly in the build_chunks function in `rag/svr/task_executor.py`. When enabled via the `auto_keywords` or `auto_questions` options in...

RagFlow automatically uses generated keywords and questions (metadata) in its search and ranking logic—you do not need to manually select or specify metadata for them to be included. These fields...

Automatic keywords and questions are used directly and automatically in RagFlow's search process. When you search, these fields—stored as important_kwd (keywords) and question_kwd (questions), along with their tokenized forms—are included...

Yes, RagFlow uses automatic keywords, automatic questions, and tag sets for weighted sorting during search. Automatic keywords and questions are given strong weights in both the initial query matching and...

The retrieval batch size being "typically a multiple of page size" is set in the search logic in rag/nlp/search.py. Specifically, the reranking batch size (RERANK_LIMIT) is calculated as math.ceil(64 /...

Hi @ZhangTianweo! I'm [Dosu](https://go.dosu.dev/dosubot) and I’m helping the TikTokDownloader team. 你的分析和调试过程非常细致，社区会非常受益！ 目前 TikTok 评论采集的最大技术障碍，正如你总结的，是 API 现在强制要求 X-Bogus 和 X-Gnarly 两个签名参数。项目已有 X-Bogus 的生成，但 X-Gnarly 的算法尚未被公开或逆向出来，项目维护者也明确表示 X-Gnarly 暂未开源[[相关讨论1]](https://github.com/JoeanAmier/TikTokDownloader/issues/591#issuecomment-3422256991) [[相关讨论2]](https://github.com/JoeanAmier/TikTokDownloader/issues/598#issuecomment-3478720630)。即使你模拟了所有 headers、参数、cookie、msToken，只要缺少 X-Gnarly，TikTok 服务器就会“静默拒绝”请求（返回...

Hi, @balazshasprai. I'm [Dosu](https://dosu.dev), and I'm helping the kyverno team manage their backlog and am marking this issue as stale. **Issue Summary** - You requested support for using existingImagePullSecrets in...

@realshuting, the user is still encountering the UNAUTHORIZED error with Kyverno v1.16.1 and chart 3.6.1 when applying the inject-misc-label policy due to image pull secret authentication issues. Could you please...