-
Notifications
You must be signed in to change notification settings - Fork 281
Pull requests: google/tunix
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
Relax batch size constraints when compute ref_logps to consumer
#1431
opened Apr 24, 2026 by
copybara-service
Bot
Loading…
Fix deleting destination buffers during reshard.
#1430
opened Apr 24, 2026 by
NicoGrande
Collaborator
Loading…
6 tasks done
Update model name in Gemma2 config for SFT and RL.
#1429
opened Apr 24, 2026 by
copybara-service
Bot
Loading…
Prevent adding environment tokens for terminal steps in trajectory collection.
#1427
opened Apr 23, 2026 by
copybara-service
Bot
Loading…
Update training dataset to R2E-Gym-Subset.
#1426
opened Apr 23, 2026 by
copybara-service
Bot
Loading…
Rename build_and_test_tunix.yml to build_andtest_tunix.yml
#1417
opened Apr 17, 2026 by
Tsukimarf
Loading…
3 of 6 tasks
Update Checkpointing Options construction through CLI configuration to move from Orbax V0 CheckpointManagerOptions to Tunix's CheckpointOptions.
#1410
opened Apr 14, 2026 by
copybara-service
Bot
Loading…
Migrate Tunix's usage of Orbax V0 Checkpoint Manager to V1 Checkpointer.
#1409
opened Apr 14, 2026 by
copybara-service
Bot
Loading…
Introduce checkpoint_options to manage options for V1 Orbax Checkpointing.
#1408
opened Apr 14, 2026 by
copybara-service
Bot
Loading…
Migrate Tunix notebook examples to Orbax Checkpoint v1 API.
#1407
opened Apr 14, 2026 by
copybara-service
Bot
Loading…
Update DeepScaler and DeepSWE examples to use Orbax v1 checkpointing APIs.
#1404
opened Apr 13, 2026 by
copybara-service
Bot
Loading…
Migrate automodel class to use Orbax Checkpoint V1 API.
#1403
opened Apr 13, 2026 by
copybara-service
Bot
Loading…
Migrate distillation and qlora examples to use Orbax v1 checkpointing APIs.
#1402
opened Apr 13, 2026 by
copybara-service
Bot
Loading…
Migrate Tunix gemma params checkpoint loading to Orbax V1
#1401
opened Apr 13, 2026 by
copybara-service
Bot
Loading…
Update Tunix tests that leverage ocp.CheckpointManagerOptions to instead use tunix.sft.checkpoint_options.
#1396
opened Apr 10, 2026 by
copybara-service
Bot
Loading…
fix: use jnp.exp(log_probs) instead of softmax(log_probs) in compute_entropy_from_logits
#1387
opened Apr 10, 2026 by
kuishou68
Loading…
Add RLOO (REINFORCE Leave-One-Out) learner for lower-variance policy …
#1377
opened Apr 9, 2026 by
kbhujbal
Loading…
6 tasks done
Previous Next
ProTip!
What’s not been updated in a month: updated:<2026-03-25.