Codeseys
/

composer-replication-framework

Reinforcement Learning

Model card Files Files and versions

composer-replication-framework / composer_replication /diloco /serverless

194 kB

Ctrl+K

Ctrl+K

4 contributors

History: 9 commits

Baladithya Balamurugan

Wave 20: close F4 s3:// gap — live-S3 DiLoCo allreduce smoke (AWS_SMOKE-gated)

bd37412 about 1 month ago

tests
Wave 20: close F4 s3:// gap — live-S3 DiLoCo allreduce smoke (AWS_SMOKE-gated) about 1 month ago
__init__.py

3.13 kB
Wave 2: 4 new modules (kill-switch, EKS/SageMaker executors, DockerSandbox) + B4/B7 completion about 2 months ago
allreduce.py

13.4 kB
Wave 14: close every Wave 13 review finding + 4 documentation files; Wave 14b: real PRIME-RL parity + multi-process DiLoCo convergence about 2 months ago
eks.py

28.4 kB
Wave 3 cleanup: close deferred-LOW review items R5/R6/R11 about 2 months ago
executor.py

11.8 kB
Wave 2: 4 new modules (kill-switch, EKS/SageMaker executors, DockerSandbox) + B4/B7 completion about 2 months ago
hf_jobs.py

3.2 kB
Wave 13: serverless DiLoCo + replaysim normalization + 3 distillation losses + PRIME-RL + Monarch about 2 months ago
modal.py

3.61 kB
Wave 13: serverless DiLoCo + replaysim normalization + 3 distillation losses + PRIME-RL + Monarch about 2 months ago
modal_spawn.py

15.2 kB
Wave 20: ModalSpawnExecutor — finish the Modal-backed serverless executor about 2 months ago
replica_entrypoint.py

5.2 kB
Wave 3: close the HIGH review findings (kill-switch wiring, HeldoutSplit, EKS entrypoint bug) about 2 months ago
sagemaker.py

28.5 kB
Wave 3 cleanup: close deferred-LOW review items R5/R6/R11 about 2 months ago