-
Notifications
You must be signed in to change notification settings - Fork 3.8k
All issues
Issue creation is restricted in this repository
Issues
is:issue state:open
is:issue state:open
Search results
Optimize remote checkpoint loading with parallel multiprocess downloads and zero-copy mmap
featureIs an improvement or enhancementIs an improvement or enhancementneeds triageWaiting to be triaged by maintainersWaiting to be triaged by maintainersStatus: Open.#21868 In Lightning-AI/pytorch-lightning;compatibility with rich 15.0.0
bugSomething isn't workingSomething isn't workingneeds triageWaiting to be triaged by maintainersWaiting to be triaged by maintainersStatus: Open.#21865 In Lightning-AI/pytorch-lightning;sync_dist mean-reduced epoch metrics are inflated when ranks log unequal batch counts (integer batch-size accumulator floors during mean-sync)
bugSomething isn't workingSomething isn't workingloggerRelated to the LoggersRelated to the LoggersStatus: Open.#21859 In Lightning-AI/pytorch-lightning;- Status: Open.#21842 In Lightning-AI/pytorch-lightning;
Mixed precision autocast is disabled when running FSDP on CPU
bugSomething isn't workingSomething isn't workingStatus: Open.#21815 In Lightning-AI/pytorch-lightning;Device mesh initialization fails when running FSDP on CPU
bugSomething isn't workingSomething isn't workingStatus: Open.#21814 In Lightning-AI/pytorch-lightning;ModelCheckpointdeletes *previous run's* checkpoint when remote filesystembugSomething isn't workingSomething isn't workingStatus: Open.#21813 In Lightning-AI/pytorch-lightning;Enable FSDP strategy on CPU (accelerator="cpu")
featureIs an improvement or enhancementIs an improvement or enhancementstrategy: fsdpFully Sharded Data ParallelFully Sharded Data ParallelStatus: Open.#21812 In Lightning-AI/pytorch-lightning;Add an end-to-end anomaly detection tutorial using PyTorch Lightning
docsDocumentation relatedDocumentation relatedStatus: Open.#21809 In Lightning-AI/pytorch-lightning;accumulate_grad_batchesis silently ignored when epoch length <accumulate_grad_batches, because Lightning always steps on the last batchbugSomething isn't workingSomething isn't workingStatus: Open.#21808 In Lightning-AI/pytorch-lightning;Checkpoints silently fail to save due to swallowed PermissionError in
_atomic_savebugSomething isn't workingSomething isn't workingneeds triageWaiting to be triaged by maintainersWaiting to be triaged by maintainersStatus: Open.#21800 In Lightning-AI/pytorch-lightning;Graceful stop on an extrenal file marker
featureIs an improvement or enhancementIs an improvement or enhancementStatus: Open.#21795 In Lightning-AI/pytorch-lightning;