知识库上传文档失败提示num_rows与预期不一致

[kulou123]

上传文档没成功过，自定义也没成功过
pre_sale.txt
code=105000004 message=SearchStore operation failed: store search store failed, err: [Store] upsert failed, the num_rows (1) of field (dense_text_content) is not equal to passed num_rows (2): invalid parameter[expected=2][actual=1]
这个是在哪里配置

[kulou123]

export EMBEDDING_TYPE="ollama"
export EMBEDDING_MAX_BATCH_SIZE=100

ollama embedding

export OLLAMA_EMBEDDING_BASE_URL="http://10.1.24.15:11434" # (string, required) Ollama embedding base_url
export OLLAMA_EMBEDDING_MODEL="bge-m3:latest" # (string, required) Ollama embedding model
export OLLAMA_EMBEDDING_DIMS=2048 # (int, required) Ollama embedding dimensions

[kulou123]

pre_sale.txt
code=105000004 message=SearchStore operation failed: store search store failed, err: [Store] upsert failed, the length(11264) of float data should divide the dim(2048)
还有这个问题

pre_sale.txt code=105000004 message=SearchStore operation failed: store search store failed, err: [Store] upsert failed, the length(11264) of float data should divide the dim(2048) 还有这个问题

我也有同样的问题，我用的ollama的模型“nomic-embed-text”，.env配置都全了没有问题，新建知识库后上传.txt出现同样的报错。依照之前的issue #838 #749 都没有解决。删除./docker/data里面的数据重启coze service也没用。

[helei945]

这种问题，大家解决了

[victorming]

ollama的嵌入模型换成mxbai-embed-large后ok