Qwen3 Embedding 0.6B Playground

Tools

• qwen3_embedding_0_6b_demo_explore_mrl Analyze the 1024-d embedding and how Matryoshka (MRL) truncation affects similarity. Encodes both texts at 32, 64, ..., 1024 dimensions and plots the cosine similarity between the truncated vectors, plus full-vector statistics and a preview of the first dimensions.
• qwen3_embedding_0_6b_demo_semantic_search Search a multilingual corpus with a query and return ranked matches. Encodes the query (optionally with the retrieval instruction) and every corpus document, then ranks documents by cosine similarity. Returns the ranked table, a bar chart of scores, and a 2D PCA map of the query vs the corpus colored by language.
• qwen3_embedding_0_6b_demo_text_similarity Compute the cosine similarity between two texts. Optionally applies the retrieval instruction to both sides. Returns the similarity score (0–1) and a qualitative label.

No tool matches.

As the server's own listing gives them. We never call a tool.