From 3d9f45308c8b8897a5569bc27da110b4defa8734 Mon Sep 17 00:00:00 2001 From: Sebastian Niehus <165138846+SebastianNiehusAA@users.noreply.github.com> Date: Tue, 4 Jun 2024 14:49:33 +0200 Subject: [PATCH] Bump major version to 3.0.0 (#874) --- CHANGELOG.md | 34 ++++++++++++++++++++-------------- RELEASE.md | 2 ++ pyproject.toml | 2 +- 3 files changed, 23 insertions(+), 15 deletions(-) diff --git a/CHANGELOG.md b/CHANGELOG.md index 5261067f4..807f06b39 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -2,9 +2,20 @@ ## Unreleased +### Breaking Changes +... +### New Features +... +### Fixes +... +### Deprecations +... + +## 3.0.0 + ### Breaking Changes - We removed the `trace_id` as a concept from various tracing-related functions and moved them to a `context`. If you did not directly use the `trace_id` there is nothing to change. - - `Task.run` no longer takes a trace id. This was an largely unused feature and we revamped the trace ids for the traces. + - `Task.run` no longer takes a trace id. This was a largely unused feature, and we revamped the trace ids for the traces. - Creating `Span`, `TaskSpan` or logs no longer takes `trace_id`. This is handled by the spans themselves, who now have a `context` that identifies them. - `Span.id` is therefore also removed. This can be accessed by `span.context.trace_id`, but has a different type. - The `OpenTelemetryTracer` no longer logs a custom `trace_id` into the attributes. Use the existing ids from its context instead. @@ -18,28 +29,26 @@ - The `ArgillaClient` now has methods `create_dataset` for less fault-ignoring dataset creation and `add_records` for performant uploads. ### New Features + - Add support for Python 3.12 - Add `skip_example_on_any_failure` flag to `evaluate_runs` (defaults to True). This allows to configure if you want to keep an example for evaluation, even if it failed for some run. - Add `how_to_implement_incremental_evaluation`. - - Improve README.md - - Add `export_for_viewing` to tracers to be able to export traces in a unified format similar to opentelemetry. + - Add `export_for_viewing` to tracers to be able to export traces in a unified format similar to OpenTelemetry. - This is not supported for the `OpenTelemetryTracer` because of technical incompatibilities. - All exported spans now contain the status of the span. - - We now support python 3.12 - Add `description` parameter to `Evaluator.evaluate_runs` and `Runner.run_dataset` to allow individual descriptions without the need to create a new `Evaluator` or `Runner`. - All models raise an error during initialization if an incompatible `name` is passed, instead of only when they are used. - - Add `aggregation_overviews_to_pandas` function to allow for easier comparison of multiple aggregation overviews + - Add `aggregation_overviews_to_pandas` function to allow for easier comparison of multiple aggregation overviews. - Add `parameter_optimization.ipynb` notebook to demonstrate the optimization of tasks by comparing different parameter combinations. - Add `convert_file_for_viewing` in the `FileTracer` to convert the trace file format to the new (OpenTelemetry style) format and save as a new file. - All tracers can now call `submit_to_trace_viewer` to send the trace to the Trace Viewer. ### Fixes - The document index client now correctly URL-encodes document names in its queries. - - The `ArgillaEvaluator` not properly supports `dataset_name` - - Update broken README links to Read The Docs - - The `evaluation` tutorial contained a broken multi-label classify example. This was fixed. - -### Deprecations -... + - The `ArgillaEvaluator` not properly supports `dataset_name`. + - Update outdated `how_to_human_evaluation_via_argilla.ipynb`. + - Fix bug in `FileSystemBasedRepository` causing spurious mkdir failure if the file actually exists. + - Update broken README links to Read The Docs. + - Fix a broken multi-label classify example in the `evaluation` tutorial. ## 2.0.0 @@ -56,9 +65,6 @@ ### Fixes - `ExpandChunks`-task is now fast even for very large documents -### Deprecations -... - ## 1.2.0 We did a major revamp of the `ArgillaEvaluator` to separate an `AsyncEvaluator` from the normal evaluation scenario. diff --git a/RELEASE.md b/RELEASE.md index 8fa31699e..016fba488 100644 --- a/RELEASE.md +++ b/RELEASE.md @@ -11,3 +11,5 @@ - Copy the changelog into the release description. Also add a link to the commits since the last release at the bottom of the description. - Make sure the changes have been merged into the main branch. - Publish the release. +- Consider updating the changelog of the [docs](https://gitlab.aleph-alpha.de/engineering/docs). The repository for the docs can be found [here](https://gitlab.aleph-alpha.de/engineering/docs). + - Update it when we have big new features we want to communicate or in preparation of the sprint review. diff --git a/pyproject.toml b/pyproject.toml index 687addbbb..37fb329ce 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -4,7 +4,7 @@ build-backend = "poetry.core.masonry.api" [tool.poetry] name = "intelligence-layer" -version = "2.0.0" +version = "3.0.0" description = "" authors = ["Aleph Alpha Engineering "] readme = "README.md"