v.0.0.14

✨ 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

v0.0.13

What's Changed

• chore: bump to typescript 5 by @mikeldking in #585
• fix: ignore non-vectors for embeddings by @RogerHYang in #584
• chore: legal info by @axiomofjoy in #583
• chore: create SECURITY.md by @mikeldking in #586
• ci: Create .github/dependabot.yml by @mikeldking in #587
• chore: Use pre commit for prettier and eslint by @mikeldking in #588
• chore: change https to http for downloading fixtures and example datasets by @axiomofjoy in #589
• fix: Update pyproject.toml to remove Development Status by @RogerHYang in #595
• fix: don't raise error during dimension type inference by @axiomofjoy in #596
• docs: update quickstart notebook by @axiomofjoy in #564
• docs: credit card fraud tutorial notebook update by @axiomofjoy in #555
• fix: don't compile js/html if exists - unblock conda by @mikeldking in #597

Full Changelog: v0.0.12...v0.0.13

v0.0.12

What's Changed

• feat(embeddings): grid view improvements: sizes, multi-modal output by @mikeldking in #565

Full Changelog: v0.0.11...v0.0.12

v0.0.11

What's Changed

• chore: scrub output and metadata from jupyter notebooks by @axiomofjoy in #550
• chore: update fixture by @axiomofjoy in #551
• fix: minimum time resolution by @RogerHYang in #557
• fix(embeddings): properly attribute points as unknown correctness when predictions/actuals are missing by @mikeldking in #558
• fix: coerce timestamps in exported parquet files by @RogerHYang in #560
• chore: copyright notice by @axiomofjoy in #561
• chore: update pip install magic by @axiomofjoy in #562
• docs: update image classification notebook by @axiomofjoy in #563

Full Changelog: v0.0.10...v0.0.11

v0.0.10

Phoenix now by default runs in a thread rather than a separate process, making boot time an order of magnitude faster.

What's Changed

• fix: add default prediction id, exclude from dimensions by @fjcasti1 in #511
• fix: display point-cloud errors in the UI for troubleshooting by @mikeldking in #525
• chore: fix version retrieval by @fjcasti1 in #527
• feat: run uvicorn in thread by @RogerHYang in #515
• fix(embeddings): make embedding details scroll by @mikeldking in #530
• docs: fix missing field in quickstart schema by @axiomofjoy in #526
• refactor: persist dataset only when starting server in a Process by @RogerHYang in #548
• docs: Update README.md with Thanks section to UMAP and HDBSCAN by @mikeldking in #549

Full Changelog: v0.0.9...v0.0.10

v0.0.9

What's Changed

• feat: scrollbar styles via modernizr by @mikeldking in #488
• chore: fix links by @mikeldking in #490
• docs: image classification tutorial by @axiomofjoy in #491
• chore: improve sentiment classification tutorial by @axiomofjoy in #492
• feat: make selection table cells resizable by @mikeldking in #494
• chore: fix style on image classification notebook by @axiomofjoy in #501
• docs: example datasets tutorial by @axiomofjoy in #497
• docs: simplify notebooks by @axiomofjoy in #498
• fix: stable indexing of exported files by @RogerHYang in #499
• docs: update readme quickstart by @axiomofjoy in #503
• fix: exclude floats from cardinality by @RogerHYang in #493
• docs: update quickstart notebook by @axiomofjoy in #502
• feat: contextual help for hyperparameters by @mikeldking in #512
• fix(embeddings): lasso selects only visible points by @mikeldking in #520
• docs: sync 4-05-23 by @mikeldking in #521

Full Changelog: v0.0.8...v0.0.9

v0.0.8

What's Changed

• fix: terminate by @RogerHYang in #472
• fix(embeddings): use eventIds instead of point IDs for everything by @mikeldking in #473
• fix: percent empty cell format by @RogerHYang in #474
• refactor: make percent cell declarative by @mikeldking in #479
• chore: readme and pypi cleanup by @axiomofjoy in #481
• fix: allow dimension values to be nullable by @mikeldking in #486
• feat: add route level error handling by @mikeldking in #485

Full Changelog: v0.0.7...v0.0.8

v0.0.7

What's Changed

• feat: read exported Parquet files into pd.DataFrame by @RogerHYang in #462
• feat(embeddings): cluster export UI by @mikeldking in #466
• docs: sync 3-29-23 by @mikeldking in #467

Full Changelog: v0.0.6...v0.0.7

v0.0.6

What's Changed

• refactor: DatasetDict to ExampleDatasets dataclass by @axiomofjoy in #452
• chore: add ExampleDatasets as a top-level import by @axiomofjoy in #453
• feat: normalize timestamps by @axiomofjoy in #451
• chore: add Session to top-level imports by @axiomofjoy in #455
• feat: support arize schema if provided by @mikeldking in #454
• feat: export to parquet by @RogerHYang in #424
• feat(embeddings): single dataset timeseries graph and selection by @mikeldking in #457
• feat(model): column formatting and time range selectors by @mikeldking in #458
• refactor: rename excludes schema field by @axiomofjoy in #460
• feat: allow download of exported parquet files by @RogerHYang in #459
• refactor: rename dataset type to dataset role by @RogerHYang in #461

Full Changelog: v0.0.5...v0.0.6

v0.0.4

What's Changed

• docs: sync 3-16-2023 by @mikeldking in #390
• chore: convert DatasetDict to dict and add repr for Dataset by @axiomofjoy in #425
• fix: make selected timestamp more explicit by @mikeldking in #445
• refactor: improve readability by @RogerHYang in #419
• fix(embedding): provide a light theme, increase point-size for viz by @mikeldking in #448
• docs: sync 2-24-2023 by @mikeldking in #449

Full Changelog: v0.0.3...v0.0.4