Skip to content

Releases: facebook/Ax

Ax 1.1.2

09 Sep 19:30
@mpolson64 mpolson64
1.1.2
6bdfd00
Compare
Choose a tag to compare
Ax 1.1.2 Latest
Latest

Bug fixes

  • Fixed rendering issue in ArmEffectsPlot when the number of arms displayed is greater than 20 (#4273)

Other changes

  • Enabled Winsorization transform by default, improving surrogate performance in the presence of outliers (#4277)
Assets 2
Loading

Ax 1.1.1

05 Sep 19:18
@mpolson64 mpolson64
1.1.1
03d1e4f
Compare
Choose a tag to compare
Ax 1.1.1

Bug fixes

  • Correctly filter out observations from Abandoned trials/arms during candidate
    generation (#4155)
  • Handle scalarized objectives in ResultsAnalysis (#4193)
  • Fix bug in polytope sampler when generating more than one candidate in a batch (#4244)

Other changes

  • Transition from setup.py to pyproject.toml for builds, modernizing Ax's build
    configuration and bringing it in compliance with PEP 518 and PEP 621 (#4100)
  • Add py.typed file, which allows typecheckers like Pyre, mypy, etc. to see Ax's types
    and avoid a TypeStub error when importing Ax (#4139)
  • Improve legibility of ArmEffectPlot by modifying legend and x-axis labels (#4220,
    #4243)
  • Address logspew in OneHotEncoder transform (#4232)
Loading

Ax 1.1.0

12 Aug 18:29
@mpolson64 mpolson64
1.1.0
0c0c892
Compare
Choose a tag to compare
Ax 1.1.0

New Features

  • New option for the method parameter in client.configure_generation_strategy:
    quality -- allows uers to indicate they would like Ax to generate the highest
    quality candidates it is able to at the expense of slower runtime (#4042)
  • New logic for deciding which analyses to produce by default in
    client.compute_analyses (#4013)
  • New parameters in client.summarize allow users to filter their summary by trial
    index and/or trial status (#4012, #4118)

Bug Fixes

  • Allow client.summarize to be called without a GenerationStrategy being set
    (i.e. before client.configure_generation_strategy or client.get_next_trails
    has been called.) (#3801)
  • Fixed incorrect grouping in TopSurfacesAnalysis (#4095)
  • Fixed ContourPlot failing to compute in certain search spaces with parameter
    constraints (#4124)
  • Misc. plotting fixes and improvements

Other changes

  • Bumped pinned botorch version to 0.15.1
  • Performance improvements in SensitivityAnalysis (#3891)
  • Improved optimization performance in constrained optimization settings (#3585)
  • Augmented logging in Client, early stopping module (#4044, #4108)
Loading

Ax 1.0.0

08 May 21:57
@mpolson64 mpolson64
1.0.0
cb16c55
Compare
Choose a tag to compare
Ax 1.0.0

We released Ax in 2019 along with Botorch, and both libraries have evolved significantly since their initial release. Today, we are excited to release Ax version 1.0.0.

Along with typical new features, improvements, and bug fixes this release introduces a new public API (found in ax.api) which allows users to access all fully supported Ax features via a singular entry point. Classes and functions from Ax entry points (“Loop API, “Service API”, and “Developer API”) remain present in the library, but are considered deprecated and may be changed backwards incompatibly (or removed) in subsequent major or minor releases of Ax.

This release also comes with a full rewrite of our documentation which can be found at ax.dev

Highlights

  • New unified API can be found in ax.api module. The classes and functions here constitute the Ax public API
    • The Client class manages state throughout the optimization process
    • Various Config classes allow the user to configure their search space, the optimization’s goals, etc.
  • Complete documentation overhaul
    • An entirely rewritten https://ax.dev/ introduces users to important concepts, demonstrates usage on synthetic and real-world optimization problems, and thoroughly documents the API

Breaking changes

Now that we have defined a singular Ax API, users should make efforts to use this as an entry point into the library. This will give users assurance that the method signatures they use will remain stable between major versions of Ax, and allow our team to provide better support via Github Issues.

This will be straightforward for those previously using the “Service API”, which closely resembles the current API – users of the “Loop API” should look towards the Quickstart Guide (https://ax.dev/docs/next/tutorials/quickstart/) for an example which demonstrates using the new API for the simple optimizations the Loop API previously served, and “Developer API” users should also use the full documentation to understand how the new API can serve their needs.

Loading

Maintenance Release, New Website, Improved Performance in Mixed Search Spaces

03 Feb 18:05
@saitcakmak saitcakmak
0.5.0
7b19a37
Compare
Choose a tag to compare
Maintenance Release, New Website, Improved Performance in Mixed Search Spaces

Compatibility

  • Require BoTorch 0.13.0

Highlights

  • AX website has been upgraded to utilize Docusaurus v3, with the API reference being hosted by ReadTheDocs. The tutorials now expose an option to open with Colab, for easy access to a runtime with modifiable tutorials. The old versions of the website can be found at archive.ax.dev.
  • Updated the default transforms and optimizers used in MBM-based models, to offer improved optimization performance in mixed discrete search spaces.
  • Implemented a preview version of Ax 1.0 API (ax/preview). The new API is available for use & testing, but remains subject to change until the 1.0 release.
  • Introduced new Analysis cards (ax/analysis) for analyzing experiment progress, results and model quality; which will replace various plotting and reporting utilities with a unified API.
  • Deprecated a number of legacy models, including GPEI, in favor of the modular BoTorch model (MBM) API.
  • Refactored Ax benchmark APIs to simplify problem setup and to support additional use cases such as noisy & asynchronous benchmarking.
  • Added EBAshr model, which is used for repeating arms in online experiments as well as regression analysis in the health checks (#3276).
  • New advanced capabilities:
    • MBM now supports per-metric model specification and model selection.
    • Support for new transition criteria and input constructors in GenerationNodes.
Loading

V0.4.3 - Maintenance Release

23 Sep 20:03
@mgarrard mgarrard
0.4.3
2e7291e
Compare
Choose a tag to compare
V0.4.3 - Maintenance Release

Key things:

  • Update to BoTorch 0.12.0 – which included some major updates to underlying modeling, please see BoTorch release for detailed description
  • Fix for a sql storage bug related to an import trying to import a deleted file

Compatibility

New Features

Bug Fixes

Other Changes

Read more
Loading

V0.4.2 - Maintenance Release

23 Sep 17:28
@mgarrard mgarrard
0.4.2
ac3a7ec
Compare
Choose a tag to compare
V0.4.2 - Maintenance Release

Key things:

  • Update to BoTorch 0.12.0 – which included some major updates to underlying modeling, please see BoTorch release for detailed description
  • Fix for a sql storage bug related to an import trying to import a deleted file

Compatibility

New Features

Bug Fixes

Other Changes

Read more
Loading

V0.4.1 - Maintenance Release

23 Jul 19:07
@mgarrard mgarrard
0.4.1
9cde572
Compare
Choose a tag to compare
V0.4.1 - Maintenance Release

Compatibility

  • Require BoTorch version v0.11.3 (#2594)

New Features

  • Add TimeAsFeature transform (#2438)
  • Add tutorial to mimic parameter constraint functionality for nonlinear constraints (#2500)
  • Add option for using posterior predictive in cross-validation (#2517)
  • Refactor BestModelSelctor to operate on ModelSpecs (#2557)
  • Templates for questions on github (#2575)

Bug Fixes

  • Update tests so that data contains metric_names that are on the corresponding experiments (#2422)
  • Use name of enum instead of casting to string (#2423)
  • Fill missing values with NaN in observation_data_to_array (#2430)
  • Prevent unused NaNs from marking constraints as infeasible in pareto_frontier_evaluator (#2431)
  • Add copyright comment (#2445)
  • Fix various spelling errors in GS (#2448)
  • Add warning when metric is skipped because it’s not on the experiment (#2449)
  • Stop ignoring unused keyword arguments (#2451)
  • Improve error clarity in TensorboardMetric (#2456)
  • Save status quo name and feature when multiple status quo present (#2457)
  • Remove unused import from MOO tutorial (#2462)
  • Update MBM tutorial with modern, non-deprecated functionality and accurate description of defaults (#2466)
  • Include CANDIDATE trials in pending_features (#2482)
  • Test that we’re not referencing same parameter dict in ObservationFeatures.from_arm (#2483)
  • Do not deduplicate against FAILED in GenerationStrategy/Node (#2488)
  • Use updated BoTorch HitAndRunPolytopeSampler for Sobol fallback (#2492)
  • Fix test_optimize_l0_homotopy test numerics (#2493)
  • Remove unused imports from ax/ (#2501)
  • Remove unused experiment arg from compute_model_fit_metrics_from_modelbridge (#2504)
  • Fix equality check for floats (#2507)
  • Fix MOO tutorial (#2509)
  • Remove unnecessary pyre-fixme (#2513)
  • Disallow modeling metrics not attached to the experiment (#2518)
  • Fix pending point bug in ExternalGenerationNode (#2521)
  • Fix num_bayesopt_trials logic to support ensembling (#2522)
  • Update np.Inf -> np.inf per numpy 2.0 release (#2527)
  • Reinforce testenv in Ax test case (#2530)
  • Avoid errors in telemetry due to node-based GenerationStrategy (#2554)
  • Use fast_botorch_optimize to fix test timeouts (#2556)
  • Make MBM acquisition pass the correct argos to each BoTorch optimizer (#2571)
  • Remove unnecessary special-casing for qEHVI and qMES from optimizer argparse (#2572)
  • Remove link to the Raytune Tutorial (#2578)
  • Expand trials_as_df to support multiple GenerationNodes in a single trial (#2577)
  • Fixing typo in docstring of get_sobol_botorch_modular_acquisition (#2583)
  • Fix incorrect status quo update warning (#2586)

Other Changes

  • Update PairwiseModelBridge to support non-preference outcomes (#2424)
  • Allow more flexible definition of which trial statuses to fit (#2432)
  • Support multiple node generation for single trial (#2428)
  • Add TimeAsFeature to transform registery (#2439)
  • Handle pending observations in TimeAsFeature (#2440)
  • Handle ObservationFeatures without trial_index in Realativize (#2441)
  • Pyre update (#2442)
  • Creating preference testing stub (#2443)
  • Add qExpectedUtilityOfBestOption to test_PairwiseModelBridge (#2444)
  • Add transition edges property to GenerationNode (https://github.com/facebook/Ax/pull/24460)
  • Update should_transition_to_next_node to handle multiple transition edges (#2447)
  • Rename TaskEncode transform to TaskChoiceToIntTaskChoice (#2450)
  • Explicit mention that AxService API is recommended throughout docs (#2453)
  • Add validation of transition edges to GS (#2454)
  • Another pyre update (#2463)
  • Always consider choice parameters with 2 values as ordered (#2464)
  • Fix forward (#2465 )
  • Change flag name from supress_generation_errors to raise_generation_errors (#2468)
  • Improve typing in SyntheticFunctions (#2470)
  • Clean up unnecessary todos in _get_model_state_from_last_generator_run (#2476)
  • Update docstring to signal in-place_modification (#2477)
  • Test against python 3.12 (#2478)
  • Have MBM only pass acqf input constructor kwargs to BoTorch when they are not None (#2480)
  • Enable evaluating Branin test experiments with None status quo (#2489)
  • Remove unused suppress_all_errors kwarg (#2491)
  • Adding analysis_batch class to ax.analysis (#2498)
  • Adding schema for AnalysisBatch and Analysis (#2499)
  • Passing current_node_name when gs is node based (#2505)
  • Put model fit data in gen_metadata (#2511)
  • Save full[er] GeneratorRuns ([https://github....
Read more
Loading

V0.4.0 - Maintenance Release

02 May 21:44
@mgarrard mgarrard
0.4.0
9ba941c
Compare
Choose a tag to compare
V0.4.0 - Maintenance Release

V0.4.0 - Maintenance Release

Compatibility

  • Require BoTorch version v0.11.0 (#2420).
  • Require Python version >= 3.10 (#2349)

FYI on future deprecations:

  • ALEBO AND REMBO will be deprecated and removed in the next release

New Features

  • New, simplified TensorboardMetric (#2236)
  • Add support for noisy problems to the Ax Benchmarks (#2255)
  • Add ExternalGenerationNode (#2266)
  • Add ExternalGenerationNode tutorial (#2281)
  • Enable batched benchmarks (with BatchTrial) (#2331)
  • Entropy of observations metric (#2340)
  • Global Sensitivity Analysis for Categorical Features (#2357)
  • Enable Sobol sensitivity analysis for SAAS models (#2363)
  • Support fixed features in Service API (#2372)
  • Support X_observed=None in Acquisition (#2393)
  • Added PredictedOutcomesDotPlot to ax.analysis (#2303)

Bug Fixes

  • Address misc lint+pyre failures impacting OSS linter (#2244)
  • Improve flakey test reliability (#2246)
  • Ensure minimize is passed to make_experiment (#2251)
  • Fix GenerationNode transition logic (#2253)
  • Handle errors in score trace computations (#2263)
  • Fix Sphinx build errors (#2267)
  • Fix flaky test for sensitivity analysis (#2268)
  • Do not re-attach the same data in get_test_map_data_experiment (#2273)
  • Fic issue with SQA storage never removing Data objects + upgrade testing for scheduler with Map Data intermediate results (#2276)
  • Fix prediction on training data in transformed space when calculating model fit quality metrics (#2279)
  • Fix integration test (#2282)
  • do not group by time cols when creating observations (#2293)
  • Fix test_compare_to_baseline_equal (#2296)
  • Fix test_sqa_storage_map_metric_experiment (#2297)
  • Ax Trial: Bug fix for error message that references nonexistent function (#2304)
  • Cloning over _time_created when cloning Experiment (#2307)
  • Fix "cannot collect test class 'TestMetric' because it has a init constructor (from: ax/core/tests/test_experiment.py)" (#2308)
  • Address non-determinism in model.metric_names in tests (#2309)
  • fix skip_runners_and_metrics for metrics on generator runs with mutable multi-objective optimization config (#2312)
  • Properly handle generators in Data.from_multiple (#2318)
  • Update some test helpers in core_stubs (#2319)
  • Rework TransitionCriterion storage to remove circular dep (#2320)
  • Fix generation node tests (#2322)
  • Fix docstring in Ax SyntheticFunction._f, Pyre fix (#2329)
  • Fix missing DB update on AxClient.stop_trial_early (#2337)
  • Fix deprecation warning from pandas.read_json (#2346)
  • Remove unused import (#2348)
  • Add an error if "use_batch_trials" is passed to AxClient (#2355)
  • Add List[str] to TConfig definition (#2360)
  • HSS: only check for dependents if the root parameter is present when casting parameterization (#2361)
  • Only untransform objective thresholds in TorchModelbridge.gen if they are not None (#2374)
  • Make sure TestCase.setUp is executed in tests (#2384)
  • Do not transform search space in-place in Modelbridge._get_transformed_gen_args (#2386)
  • Fix warning filters in TestCase, introduce AxParameterWarning (#2389)
  • Fix broken test from D56359739 (#2400)
  • Fix sphinx build (#2419)

Other Changes

  • Add output_tasks to MTGP in MBM (#2241)
  • Pyre Configurationless migration (#2243, #2261, #2359, #2368)
  • Update docstring for GenerationNode.gen & fit (#2245)
  • Implement Data.eq (#2247)
  • copying "cross_validation_helper" code from ax.plot (#2249)
  • validate metrics after setting options (#2250)
  • Introduces AxGenerationException to facilitate exception handling (#2254)
  • Light cleanup of GenerationStrategyInterface (#2256)
  • Add Sobol benchmark method (#2257)
  • Light GenStrategy cleanup (#2258)
  • Extend docstring of extract_parameter_constraints (#2262)
  • Retain original data timestamp in experiment.clone_with (#2269)
  • Docstring clarification (#2270)
  • Clean up too-verbose logs (#2275)
  • Add fit_out_of_design to TorchOptConfig (#2277)
  • Update doc strings on TransitionCriterion to improve usability (#2278)
  • split common target into native/non-native parts (#2280)
  • Unblock node based GS in AxClient.get_optimization_trace (#2283)
  • Allow customizing num_init_trials in MBM benchmark method (#2286)
  • Update log message in WithDBSettingsBase._load_experiment_and_generation_strategy (#2287)
  • Abstract out attribute copy during clone_to (#2288)
  • updates trial status during clone (#2290)
  • Plot top n features in countours (#2291)
  • Expose fit_out_of_design (#2292)
  • Clean up too verbose logs pt. 2 (#2298)
  • Changed concatenation to f-strings (#2300)
  • Silence input normalization warnings in cross validation (#2310)
  • Clean up GenNode class 1/3 (#2313)
  • Return weights from Acquisition.optimize (#2314)
  • Upgrade actions/checkout & actions/setup-python (#2315)
  • Rename OrderedChoiceEncode => OrderedChoiceToIntegerRange (#2323)
  • Tests for sequential=True as default for models in ModelBridge factory (#2324)
  • add MergeRepeatedMeasurements to the transform registry (#2325)
  • Typing improvements to RangeParameter (#2327)
  • update run_metadata in BaseTrial.run instead of overwriting (#2328)
  • Allow for passing model_gen_kwargs in benchmarks (#2336)
  • only infer reference point in global stopping if there is data (#2338)
  • Allow bulk_fetch_trial_data to return mix of successes/failures (#2339)
  • Tests which store and load dataframe and figure (#2344)
  • Add e2e tests with HSS (#2345)
  • Switch legacy LCE-A to qLogNEI from legacy ei_or_nei (#2352)
  • Remove fall-back to list (#2353)
  • Rename global stopping tests_strategies -> test_strategies (#2354)
  • Update HSS dummy value logic & expose it in Cast ([#2362](https://git...
Read more
Loading

v0.3.7 - Maintenance Release

01 Mar 16:10
@mpolson64 mpolson64
0.3.7
a1f94b0
Compare
Choose a tag to compare
v0.3.7 - Maintenance Release
Loading