Skip to content

Actions: google-deepmind/kfac-jax

Actions

All workflows

Actions

Loading...
Loading

Showing runs from all workflows
1,404 workflow runs
1,404 workflow runs

Filter by Event

Filter by Status

Filter by Branch

Filter by Actor

Scheduled CodeQL #775: by github-advanced-security bot
December 3, 2024 12:23 1m 12s main
main
December 3, 2024 12:23 1m 12s
Adding NaN/Inf guard on call to matrix inverses/solves since LU decomp on GPU can cause an infinite loop when the matrix has these values. ci #1531: Pull request #305 synchronize by copybara-service bot
November 30, 2024 22:40 2m 18s test_701379417
test_701379417
November 30, 2024 22:40 2m 18s
Adding NaN/Inf guard on call to matrix inverses/solves since LU decom… ci #1530: Commit bc000c6 pushed by copybara-service bot
November 30, 2024 22:40 2m 23s main
main
November 30, 2024 22:40 2m 23s
PR #305 CodeQL #774: by copybara-service bot
November 30, 2024 22:40 1m 12s refs/pull/305/head
refs/pull/305/head
November 30, 2024 22:40 1m 12s
Adding NaN/Inf guard on call to matrix inverses/solves since LU decomp on GPU can cause an infinite loop when the matrix has these values. ci #1529: Pull request #305 synchronize by copybara-service bot
November 30, 2024 22:30 2m 24s test_701379417
test_701379417
November 30, 2024 22:30 2m 24s
PR #305 CodeQL #773: by copybara-service bot
November 30, 2024 22:30 1m 5s refs/pull/305/head
refs/pull/305/head
November 30, 2024 22:30 1m 5s
Adding NaN/Inf guard on call to matrix inverses/solves since LU decomp on GPU can cause an infinite loop when the matrix has these values. ci #1528: Pull request #305 opened by copybara-service bot
November 30, 2024 01:43 2m 33s test_701379417
test_701379417
November 30, 2024 01:43 2m 33s
PR #305 CodeQL #772: by copybara-service bot
November 30, 2024 01:43 1m 5s refs/pull/305/head
refs/pull/305/head
November 30, 2024 01:43 1m 5s
Push on main CodeQL #771: by copybara-service bot
November 29, 2024 18:47 1m 8s main
main
November 29, 2024 18:47 1m 8s
Minor bug fix: changing train_step in examples code to take a mean of… ci #1527: Commit face046 pushed by copybara-service bot
November 29, 2024 18:47 2m 25s main
main
November 29, 2024 18:47 2m 25s
PR #304 CodeQL #770: by copybara-service bot
November 29, 2024 18:47 1m 5s refs/pull/304/head
refs/pull/304/head
November 29, 2024 18:47 1m 5s
Minor bug fix: changing train_step in examples code to take a mean of the stats instead of taking from the first device. Because the optimizer syncs its own stats (like loss), this didn't matter except for stats returned from the kfac_jax optimizer (or Optax optimizers using OptaxWrapper). However, the Polyak averaged loss wasn't actually synced across devices (as its not part of the optimizer anymore), so "loss_polyak" was being reported only for the first device. ci #1526: Pull request #304 synchronize by copybara-service bot
November 29, 2024 18:41 2m 24s test_700710272
test_700710272
November 29, 2024 18:41 2m 24s
PR #304 CodeQL #769: by copybara-service bot
November 29, 2024 18:41 1m 4s refs/pull/304/head
refs/pull/304/head
November 29, 2024 18:41 1m 4s
Minor bug fix: changing train_step in examples code to take a mean of the stats instead of taking from the first device. Because the optimizer syncs its own stats (like loss), this didn't matter except for stats returned from the kfac_jax optimizer (or Optax optimizers using OptaxWrapper). However, the Polyak averaged loss wasn't actually synced across devices (as its not part of the optimizer anymore), so "loss_polyak" was being reported only for the first device. ci #1525: Pull request #304 synchronize by copybara-service bot
November 27, 2024 21:00 2m 32s test_700710272
test_700710272
November 27, 2024 21:00 2m 32s
PR #304 CodeQL #768: by copybara-service bot
November 27, 2024 21:00 1m 17s refs/pull/304/head
refs/pull/304/head
November 27, 2024 21:00 1m 17s
Minor bug fix: changing train_step in examples code to take a mean of the stats instead of taking from the first device. Because the optimizer syncs its own stats (like loss), this didn't matter except for stats returned from the kfac_jax optimizer (or Optax optimizers using OptaxWrapper). However, the Polyak averaged loss wasn't actually synced across devices (as its not part of the optimizer anymore), so "loss_polyak" was being reported only for the first device. ci #1524: Pull request #304 opened by copybara-service bot
November 27, 2024 19:45 2m 20s test_700710272
test_700710272
November 27, 2024 19:45 2m 20s
PR #304 CodeQL #767: by copybara-service bot
November 27, 2024 19:45 1m 9s refs/pull/304/head
refs/pull/304/head
November 27, 2024 19:45 1m 9s
Making pmap axis names consistent in examples code to support things like cross-replica batch norm layers. ci #1523: Pull request #301 synchronize by copybara-service bot
November 27, 2024 14:56 2m 21s test_699200113
test_699200113
November 27, 2024 14:56 2m 21s
Making pmap axis names consistent in examples code to support things … ci #1522: Commit 4de99f5 pushed by copybara-service bot
November 27, 2024 14:56 2m 27s main
main
November 27, 2024 14:56 2m 27s
PR #301 CodeQL #766: by copybara-service bot
November 27, 2024 14:56 1m 13s refs/pull/301/head
refs/pull/301/head
November 27, 2024 14:56 1m 13s
Making pmap axis names consistent in examples code to support things like cross-replica batch norm layers. ci #1521: Pull request #301 synchronize by copybara-service bot
November 27, 2024 14:49 2m 26s test_699200113
test_699200113
November 27, 2024 14:49 2m 26s
PR #301 CodeQL #765: by copybara-service bot
November 27, 2024 14:49 1m 9s refs/pull/301/head
refs/pull/301/head
November 27, 2024 14:49 1m 9s
PR #303 CodeQL #764: by copybara-service bot
November 26, 2024 22:07 1m 6s refs/pull/303/head
refs/pull/303/head
November 26, 2024 22:07 1m 6s
- Fixing bug with step rejection where reject_damping_increase_factor was applied when step was *not* rejected. ci #1520: Pull request #303 synchronize by copybara-service bot
November 26, 2024 22:07 2m 20s test_699297543
test_699297543
November 26, 2024 22:07 2m 20s
PR #303 CodeQL #763: by copybara-service bot
November 26, 2024 21:54 1m 10s refs/pull/303/head
refs/pull/303/head
November 26, 2024 21:54 1m 10s