Arize-ai/phoenix v.0.0.14

on GitHub

✨ Native support for LLM prompt / response pairs

Schema(
    tag_column_names=[
        "bleu_score",
        "rouge_score",
    ],
    prompt_column_names=EmbeddingColumnNames(
        vector_column_name="document_vector", raw_data_column_name="document"
    ),
    response_column_names=EmbeddingColumnNames(
        vector_column_name="summary_vector", raw_data_column_name="summary"
    ),
)

Tutorials coming soon!

What's Changed

• feat(embeddings): support rendering grid previews of llm prompts and responses by @mikeldking in #599
• feat(embeddings): show prompt and response for LLMs in event details by @mikeldking in #600
• fix: incorrect actual label shown in UI by @mikeldking in #601
• feat(embeddings): display prompt response pairs in selection table by @mikeldking in #602
• feat: Enable generative prompt/response pair by @fjcasti1 in #598
• chore: llm summarization fixture by @axiomofjoy in #606
• feat(embeddings): prompt/response pairs on inference events by @mikeldking in #605
• fix: exclude bool from numeric by @RogerHYang in #607
• feat(embeddings): plumb through prompt / response pairs to the UI by @mikeldking in #608
• fix: handle prompt response pairs in feature discovery by @axiomofjoy in #609

Full Changelog: v0.0.13...v.0.0.14