知识库向量文件报错

采用火山的embedding-version模型，配置endpoint节点后报错err: [Store] upsert failed, params column dense_text_content vector dim 2048 not match collection definition, which has dim of 256，这个模型不是支持2048吗，配置后也重启了
如果将ep-20251112001520-xfclr改为doubao-embedding-version-250615报错找不到该模型，请各位大佬帮忙指正

Start with the reported endpoint/model configuration and the vector collection definition, then reproduce the upsert error using the shown embedding settings. Done means the configured model is recognized and its vector dimension matches the collection definition without an upsert failure.

Written by the indexing model from the issue text.