jiangsier-xyz

Solve the problem of occasional incomplete response content.

2

It is now possible for Qwen models to return messages with content when finishReason == STOP (but not always). compatible with this situation. When finishReason != STOP, if the returned...

Add a metadata field to the Response

2

Add a metadata field to the Response that can be populated by model providers. (#698) DashScope callings fill the requestId as the response metadata. ## Summary by CodeRabbit - **New...

Currently, the EmbeddingStore is designed with 'add/addAll' methods, making it very difficult to support idempotent operations, which are essential for tasks that can failover (many operations related to RAG might...

We might need to distinguish APIs for embedding queries and embedding documents/keys in EmbeddingModel.java. DashScope's embedding service has long provided an input parameter for two types of text_type (query, document):...

fix: #1147 ## Issue https://github.com/langchain4j/langchain4j/issues/1147 ## Change It seems that Qwen's online tokenizer service doesn't support strings made up of whitespace characters. Workaround: when the input is detected to be...

fix: #1140 ## Issue ## Change ## General checklist - [x] There are no breaking changes - [x] I have added unit and integration tests for my change - [x]...