0.21.0

Release Notes

0.21.0

Changelog

• b0e47da chore: bump version: 0.21.0-rc4 -> 0.21.0
• 7b09cc9 docs: add release notes for 0.21.0 (#6365)
• 0073a16 chore: bump version: 0.21.0-rc3 -> 0.21.0-rc4
• 1837993 chore: remove old slots patch endpoint (#6330)
• b61169c fix: pytorch loading accepts trial_class kwarg (#6356)
• c45cf5c chore: bump version: 0.21.0-rc2 -> 0.21.0-rc3
• 38d7a88 fix: use new router store for login redirect (#6322)
• 47cf4c0 fix: workspace modal filter (#6341)
• 87c0c66 fix: fix patchSlotState util (#6345)
• bedc738 fix: Workspace members, isFiltered / reset tableOffset (#6335)
• e0d30bb fix: skip op assert in test mode (#6338)
• 0fdee1e chore: bump version: 0.21.0-rc1 -> 0.21.0-rc2
• 98b342e chore: prepare for RBACed agents/slots enable/disable [DET-9156] (#6310)
• d3b9c85 fix: remove extra markResourcesStarted call (#6331)
• 403967f chore: reduce GPU count for GCP deployments (#6337)
• 4d54710 fix: Indicate and clear user search filter [WEB-1074] (#6309)
• a70452f fix: Learning curve on validation-only experiment (#6334)
• b5a1e88 chore: bump version: 0.21.0-rc0 -> 0.21.0-rc1
• 69dbc40 fix: Fix user store (#6327)
• aad5947 feat: show log viewer agent filter options even when there is one option (#6324)
• f85ab20 fix: workspace card issue (#6320)
• eef0be0 fix: expand torch + distutils workaround (#6313)
• 003bd74 chore: avoid usage of k8s NamespaceAll easing k8s permission requirements [DET-9123] (#6248)
• 93d9e81 chore: bump version: 0.21.0-dev0 -> 0.21.0-rc0
• eef6069 chore: bump version: 0.20.2-dev0 -> 0.21.0-dev0
• f0dd082 fix: tb_callback not found (#6311)
• 9e7a636 fix: fix disable slot unexpected msg [DET-9155] (#6308)
• 6b5eabf feat: cli expect bindings.APIHttpError (#6278)
• 0841794 ci: retry large pip installs (#6260)
• a3ab995 chore: lock api state for backward compatibility check
• fe966c1 feat: remove static agents from gcp deployments [MLG-366] (#6281)
• fb2f515 chore: deprecated sdk methods (#6268)
• fc885e7 docs: PyTorch Trainer (#6177)
• fec6cf2 feat: load PyTorchTrials with kwargs (#6286)
• 9d62826 feat: checkpoint delete in python SDK [MLG-377] (#6230)
• 640af40 feat: log gpu information in task logs [MLG-234] [DET-5274] (#6303)
• f006a4a fix: report_training_metrics on parallel (#6304)
• 09df35c docs: remove autodoc test snippet (#6301)
• b7b07f8 fix: allocation rekill shouldn't use time.UTC in fine time checks (#6298) [DET-9149]
• 1016033 fix: do not restore stopping_killed experiments (#6300) [DET-9153]
• fcdb8e0 feat: Make Torchwriter behave like Pytorch SummaryWriter (#5916)
• 0e024ca feat: show TrialWorkloads table when f_chart is on [WEB-1025] (#6269)
• 8504b2e ci: bump gke version to 1.23 (#6292)
• b331c58 fix: fix err status code for ResourceAllocationRaw (#6294)
• 4473ef3 fix: update webui static gzip path regex (#6285)
• ed6e0d0 fix: Run environment bump scripts after environments fix (#6287)
• 6e8d238 docs: add instructions on how to use glasbey colors (#6266)
• 6aa04b9 chore: document dropping K80 support [MLG-385] (#6272)
• 51cb2a7 build(deps): bump actions/setup-python from 3 to 4 (#6084)
• cce1237 build(deps): bump actions/setup-go from 3 to 4 (#6244)
• 75662c2 chore(deps): bump mellium.im/sasl from 0.3.0 to 0.3.1 in /master (#5668)
• 2f2390f fix: login validation (#6284)
• 3d56934 fix: avoid explosion of prom series per actor with better labelling (#6280)
• 073e6ac chore: update code for new flake8 upgrade (#6276)
• 1acb9fd fix: reflect data immediately in model page and model version page (#6277)
• 447a00d fix: remove perpetual spinner (#6279)
• 87cf260 fix: Load multi-trial learning curve with only training metrics [WEB-1026] (#6220)
• 9bb9caa test: initialize certs properly in an auth utility [DET-9104] (#6271)
• 1888c33 chore: remove deprecated validation callbacks [MLG-21] (#6215)
• 3e5c863 chore: do not error on shared_fs delete for nonexistent paths [MLG-378] (#6247)
• 44dcfc7 fix: browser warnings (#6258)
• 51528d8 fix: handle results of polling updates in allocation actor (#6264) [DET-9126]
• e444f0e test: Small refactor of CLI experiment tests (#6237)
• 37794fa chore: take advantage of CheckpointContext handling sharded data [MLG-13] (#6200)
• ad02a93 chore: update requirement for Launcher version to 3.2.4 (#6261)
• c6b7dde docs: retitle api reference docs (#6257)
• 935ec57 chore: minor cleanup in useValueMemoizedObservable (#6256)
• 6227782 docs: Slurm details updates [FOUNDENG-512] (#6263)
• d89d98a fix: Restore proper SIGTERM handling for Slurm/PBS [DET-9124] (#6250)
• 36ad652 feat: use Loadable in ChartGrid and LineChart (#6239)
• cf4d47d fix: update User Management table on new user [WEB-1011] (#6231)
• fd770ff fix: Nameplate icon width bug (#6251)
• eb954b9 fix: correct cluster slots card size [WEB-1028] (#6249)
• 672223c chore: update npm libraries (#6241)
• 555f791 chore: add merge func for task container defaults (#6236)
• d808168 chore: move from K80 to T4 GPU type defaults (#6224)
• 08b0444 chore: bump version: 0.20.1-dev0 -> 0.20.2-dev0
• 0a69a8c docs: add release notes for 0.20.1 (#6235)
• 9950726 fix: Chart focus maintained as series changes (#6240)
• 352a08b fix: permission issue in member tab in workspace (#6208)
• 821994c docs: use theme templates better (#6226)
• ca330f5 fix: ensures docker container is cleaned up if a container is killed after start, but before run [DET-8988] (#6223)
• 7ddffdb ci: update ESLint to v8 (#6238)
• e3b6071 chore: preparing to integrate UserByExternalToken with SCIM support (#6201)
• 045d3e8 fix: CSS inside router (#6234)
• 2138c0c docs: update script for new core api user guide (#6228)
• 6f3463c docs: add HPC Launcher job exit code known issue (#6233)
• 0b9698a feat: remove TODOS from users loadable "loading" state (#6104)
• 6ec8f55 feat: hide non-global roles on user/group modal (#6189)
• e9c2259 chore: Update max_concurrent_trials to have a default value of 16 (#6221)
• a74ccd0 feat: disable user management if external sessions are enabled [DET-8791] (#6123)
• b6f7163 ci: update Stylelint to v15 (#6218)
• d5665e1 chore: Remove deprecated react-router [WEB-580] (#6157)
• b54eda8 docs: minor edits to commands and shells (#6227)
• 69c36b3 feat: show unique HPs for Custom Searcher trials on experiment table [MLG-343] (#6192)
• e7b04c7 feat: collect timing data for actor system internals (#6216)
• 7dd63d0 feat: Enable username and display name search on cluster "user" list [WEB-1009] (#6198)
• fea9aa4 fix: username case sensitive issue (#6196)
• 1c8c473 fix: log level filter width (#6217)
• 5a348d6 chore: create Nameplate component (#6153)
• 2b3da4d chore: remove type/interface member ordering rule from eslint (#5896)
• 10f3914 fix: det user change-password for user's own passwords. (#6214)
• 35056ef docs: add current rbac limitations to docs (#6212)
• b083ad1 fix: stop ws/prj failed notifications (#6210)
• a1c3d93 fix: flip undefined check logic for url params in ts bindings (#6213)
• b429b3f fix: project card issue (#6185)
• ac586f9 feat: update the regex to match the new build output format (#6163)
• e6813e1 fix: default tag in AWS stack updates (#6207)
• 11fbfb7 feat: PyTorch Trainer (#6059)
• efdba0b docs: clarified rbac and k8s docs (#6169)
• 87b3cfb chore: move assigned event to correct location (#6197) [DET-9012]
• 2269d78 perf: improve checkpoint size performance [DET-8960] (#6171)
• 9240dfc build: Eliminate CRACO, use Vitest for testing (#6021)
• c52e317 perf: remove Trialv1.latest_training field. (#6155)
• 5b39a94 fix: adding exit reason to tensorboard allocation signal (#6150)
• 1571089 fix: support mixed CPU/GPU k8s quotas (#6195)
• b4a1081 fix: modified historical task allocation endpoint to instead aggregate allocations (#6184)
• d13169c refactor: use generic Go set type in more places (#6179)
• 77f8ec4 chore: bumpenvs reflecting tf27 not having pytorch (#6182)
• 744bd1b fix: remove get-deps-bindings target from top level makefile (#6191)
• a3e7b89 fix: add archived param and simplify query (#6175)
• 9f43ad1 fix: exp move modal (#6183)
• 3bc5524 fix: don't print ':' when err msg is empty (#6190)
• 47c139d fix: pass workspace ID when creating tensor board from WebUI [WEB-1019] (#6186)
• 94d37ab fix: Continue Trial flow does not take the new max_length (#6168)
• 5c3d6ae fix: Close expiriment fork/ continue trial modal properly (#6174)
• e8d4dcb build: eliminate java dependency for typescript swagger bindings (#6139)
• ba87f94 revert: reflag new chart experience (#6181)
• dcf2586 feat: add labels to GCP instances created with det deploy gcp [MLG-170] (#6147)
• e8858a6 fix: hide Foldable menu options when button is visible (#6178)
• 0849631 refactor: replace user store with observables [WEB-799] (#6140)
• b3e7ecc chore: pass Labels/project/workspace to TaskSpec (#6172)
• 455f536 fix: useMemo does not depend on trial having been loaded (#6173)
• 7d8b711 fix: separate Router and authCheck (#6170)