-
Notifications
You must be signed in to change notification settings - Fork 4.2k
Issues
is:issue state:open
is:issue state:open
Issue creation is restricted in this repository
Search results
CI perf regression: hybrid_2b_perf batch_32 on GB200
bugSomething isn't workingSomething isn't workingStatus: Open.#5693 In NVIDIA/Megatron-LM;CI perf regression: gpt_583m_perf_gb200_4gpu on GB200
bugSomething isn't workingSomething isn't workingStatus: Open.#5692 In NVIDIA/Megatron-LM;Make lm_head optional for BertModel
enhancementNew feature or requestNew feature or requestStatus: Open.#5689 In NVIDIA/Megatron-LM;qk_layernorm option for BertModel
enhancementNew feature or requestNew feature or requestStatus: Open.#5686 In NVIDIA/Megatron-LM;- Status: Open.#5685 In NVIDIA/Megatron-LM;
Support multimodal payloads in Dynamic-CP sequence packing scheduler
enhancementNew feature or requestNew feature or requestStatus: Open.#5683 In NVIDIA/Megatron-LM;- Status: Open.#5676 In NVIDIA/Megatron-LM;
Automatic Checkpoint Converter to HybridModel
enhancementNew feature or requestNew feature or requestStatus: Open.#5675 In NVIDIA/Megatron-LM;refactor hyper connection contraction to a module
enhancementNew feature or requestNew feature or requestquestionFurther information is requestedFurther information is requestedStatus: Open.#5670 In NVIDIA/Megatron-LM;[Proposal] HeartFlow - Cognitive Engine for Megatron
questionFurther information is requestedFurther information is requestedStatus: Open.#5663 In NVIDIA/Megatron-LM;[MCore][TE][NVFP4] Persist activation global input_scale/input_amax for offline FP4 inference
enhancementNew feature or requestNew feature or requestStatus: Open.#5660 In NVIDIA/Megatron-LM;- Status: Open.#5656 In NVIDIA/Megatron-LM;