Skip to content

Actions: fattorib/ZeRO-transformer

Actions

All workflows

Actions

Loading...

Showing runs from all workflows
163 workflow runs
163 workflow runs
Event

Filter by event

Status

Filter by status

Branch
Actor

Filter by actor

ok but now tests are actually passing
Tests #1048: Commit b741d77 pushed by fattorib
May 30, 2023 21:32 3m 44s pjit-tensor-parallel
May 30, 2023 21:32 3m 44s
parallel residual fixes
Tests #1047: Commit 16081f4 pushed by fattorib
May 29, 2023 21:39 4m 33s pjit-tensor-parallel
May 29, 2023 21:39 4m 33s
clean up shard_map + add pmean for grads
Tests #1046: Commit c0285c0 pushed by fattorib
May 29, 2023 21:39 3m 39s pjit-tensor-parallel
May 29, 2023 21:39 3m 39s
parallel residual
Tests #1045: Commit 196a405 pushed by fattorib
May 29, 2023 10:12 4m 58s pjit-tensor-parallel
May 29, 2023 10:12 4m 58s
add optimizer to mp=1 case
Tests #1044: Commit 885d57a pushed by fattorib
May 29, 2023 09:24 4m 19s pjit-tensor-parallel
May 29, 2023 09:24 4m 19s
repo cleanup
Tests #1042: Commit 15e8a04 pushed by fattorib
May 29, 2023 09:12 3m 46s pjit-tensor-parallel
May 29, 2023 09:12 3m 46s
fix failing tests by adding tp_comms flag
Tests #1041: Commit 871b017 pushed by fattorib
May 28, 2023 19:19 3m 29s pjit-tensor-parallel
May 28, 2023 19:19 3m 29s
optimizer sharding + updates
Tests #1040: Commit 5c76fc6 pushed by fattorib
May 28, 2023 19:10 3m 35s pjit-tensor-parallel
May 28, 2023 19:10 3m 35s
bump jax version on TPU
Tests #1039: Commit 03bee97 pushed by fattorib
May 28, 2023 18:23 5m 5s pjit-tensor-parallel
May 28, 2023 18:23 5m 5s
add explicit comms to model fwd pass
Tests #1038: Commit 088f3fc pushed by fattorib
May 28, 2023 18:22 3m 41s pjit-tensor-parallel
May 28, 2023 18:22 3m 41s
working TP impl with shard_map
Tests #1037: Commit d023e90 pushed by fattorib
May 28, 2023 15:23 3m 35s pjit-tensor-parallel
May 28, 2023 15:23 3m 35s
branch cleanup
Tests #1036: Commit ced6393 pushed by fattorib
May 28, 2023 13:12 3m 34s pjit-tensor-parallel
May 28, 2023 13:12 3m 34s
flexibility for mp axis name in partitioning.py
Tests #1035: Commit f89c2ad pushed by fattorib
May 27, 2023 09:39 4m 19s pjit-tensor-parallel
May 27, 2023 09:39 4m 19s
"None" -> None
Tests #1034: Commit d482e38 pushed by fattorib
May 27, 2023 09:33 4m 2s pjit-tensor-parallel
May 27, 2023 09:33 4m 2s
merge another einops call
Tests #1033: Commit a4c7853 pushed by fattorib
May 27, 2023 09:25 3m 52s pjit-tensor-parallel
May 27, 2023 09:25 3m 52s
drop jit for loss and remove bias sharding for tp
Tests #1032: Commit 168278e pushed by fattorib
May 27, 2023 09:20 5m 18s pjit-tensor-parallel
May 27, 2023 09:20 5m 18s
clean up vars and lint
Tests #1031: Commit a5886d4 pushed by fattorib
May 26, 2023 17:32 4m 46s pjit-tensor-parallel
May 26, 2023 17:32 4m 46s
self-contained dp example for debugging
Tests #1030: Commit 5635b79 pushed by fattorib
May 26, 2023 17:30 3m 43s pjit-tensor-parallel
May 26, 2023 17:30 3m 43s
drop unused args
Tests #1029: Commit 41f32f8 pushed by fattorib
May 25, 2023 18:27 4m 12s pjit-tensor-parallel
May 25, 2023 18:27 4m 12s
reorder loss operations to reduce all-gathers
Tests #1027: Commit d2239f3 pushed by fattorib
May 24, 2023 12:25 6m 25s pjit-tensor-parallel
May 24, 2023 12:25 6m 25s
remove logit/label reshape
Tests #1026: Commit 2615cce pushed by fattorib
May 24, 2023 12:11 4m 40s pjit-tensor-parallel
May 24, 2023 12:11 4m 40s
reorder tp profile
Tests #1025: Commit 0ec93ba pushed by fattorib
May 24, 2023 10:52 4m 20s pjit-tensor-parallel
May 24, 2023 10:52 4m 20s
perfetto trace for xmap
Tests #1024: Commit 5b032f0 pushed by fattorib
May 24, 2023 10:43 4m 32s pjit-tensor-parallel
May 24, 2023 10:43 4m 32s