-
Notifications
You must be signed in to change notification settings - Fork 4.5k
All issues
Issue creation is restricted in this repository
Issues
is:issue state:open
is:issue state:open
Search results
[ENHANCEMENT] Reduce torch_dist checkpoint preparation overhead for prepended-axis tensors
enhancementNew feature or requestNew feature or requestStatus: Open.#7275 In NVIDIA/Megatron-LM;Move PEFT model components into Megatron Core
enhancementNew feature or requestNew feature or requestStatus: Open.Apply PEFT adapters in the Megatron-LM training loop
enhancementNew feature or requestNew feature or requestStatus: Open.Add adapter-only checkpointing to Megatron-LM
enhancementNew feature or requestNew feature or requestStatus: Open.Consolidate PEFT support into Megatron-LM
enhancementNew feature or requestNew feature or requestStatus: Open.#7267 In NVIDIA/Megatron-LM;Introduce a callback system for the Megatron-LM training loop
enhancementNew feature or requestNew feature or requestStatus: Open.- Status: Open.#7264 In NVIDIA/Megatron-LM;
- Status: Open.#7232 In NVIDIA/Megatron-LM;
- Status: Open.#7223 In NVIDIA/Megatron-LM;
[Determinism] Add deterministic TE fused loss support
enhancementNew feature or requestNew feature or requestStatus: Open.#7221 In NVIDIA/Megatron-LM;🐛 CI failure: JET launcher killed with exit 137 after GPT-583M inference tests pass
bugSomething isn't workingSomething isn't workingStatus: Open.#7215 In NVIDIA/Megatron-LM;[Bug] global_aux_loss uses the current microbatch token count to normalize accumulated expert counts
bugSomething isn't workingSomething isn't workingwaiting-on-maintainersWaiting on maintainers to respondWaiting on maintainers to respondStatus: Open.#7213 In NVIDIA/Megatron-LM;