-
Notifications
You must be signed in to change notification settings - Fork 443
All issues
Issue creation is restricted in this repository
- #3754 · anwithk opened
on May 8, 2026
Issues
is:issue state:open
is:issue state:open
Search results
[bug] gpt_sft force-enables TOKENIZERS_PARALLELISM at import, disabling the tokenizers fork-deadlock guard for the whole SFT path
area:dataDataset builders, preprocessing, and samplersDataset builders, preprocessing, and samplersbugSomething isn't workingSomething isn't workingneeds-triageNew item needs classification and ownershipNew item needs classification and ownershipStatus: Open.#5328 In NVIDIA-NeMo/Megatron-Bridge;[support] Qwen3.5-VL-35B-A3B MFU is less than 10% on B300.
area:perfPerformance optimizations and benchmarkingPerformance optimizations and benchmarkingneeds-triageNew item needs classification and ownershipNew item needs classification and ownershipStatus: Open.#5318 In NVIDIA-NeMo/Megatron-Bridge;[bug] DeepSeek V4 Pro FP8-MX pretraining gives gradient norm 'nan' on GB300
area:recipeTraining recipes and launch configsTraining recipes and launch configsbugSomething isn't workingSomething isn't workingneeds-triageNew item needs classification and ownershipNew item needs classification and ownershipStatus: Open.#5306 In NVIDIA-NeMo/Megatron-Bridge;[bug] DeepSeek MLA with q_lora_rank=null builds a trainable Q-norm the HF architecture does not define
area:modelModel implementations and HF bridge logicModel implementations and HF bridge logicbugSomething isn't workingSomething isn't workingneeds-triageNew item needs classification and ownershipNew item needs classification and ownershipStatus: Open.#5261 In NVIDIA-NeMo/Megatron-Bridge;[bug] Qwen35VLBridge MTP mappings use wrong Megatron prefix (language_model.mtp.* instead of mtp.*), crashes weight import
area:modelModel implementations and HF bridge logicModel implementations and HF bridge logicbugSomething isn't workingSomething isn't workingwaiting-on-maintainersWaiting on maintainers to respondWaiting on maintainers to respondStatus: Open.#5252 In NVIDIA-NeMo/Megatron-Bridge;[feature] Support DSpark speculative decoding draft training
area:modelModel implementations and HF bridge logicModel implementations and HF bridge logicfeatureNew capabilities, enhancements, or enablement workNew capabilities, enhancements, or enablement workwaiting-on-maintainersWaiting on maintainers to respondWaiting on maintainers to respondStatus: Open.#5230 In NVIDIA-NeMo/Megatron-Bridge;[support] How to determine if the DeepSeek V4 model conversion is correct?
area:ckptCheckpoint conversion, loading, export, and save pathsCheckpoint conversion, loading, export, and save pathswaiting-on-customerWaiting on the original author to respondWaiting on the original author to respondStatus: Open.#5206 In NVIDIA-NeMo/Megatron-Bridge;Unify checkpoint-derived HF config across CPU and GPU export backends
area:ckptCheckpoint conversion, loading, export, and save pathsCheckpoint conversion, loading, export, and save pathsbugSomething isn't workingSomething isn't workingtrackingTracking issue for an ongoing project with smaller stepsTracking issue for an ongoing project with smaller stepsStatus: Open.#5153 In NVIDIA-NeMo/Megatron-Bridge;[Tracking] Support pip-installed Bridge conversion across the previous MCore minor release (N-1) to the current pin
area:buildDependencies, packaging, images, and environment setupDependencies, packaging, images, and environment setuparea:ckptCheckpoint conversion, loading, export, and save pathsCheckpoint conversion, loading, export, and save pathsfeatureNew capabilities, enhancements, or enablement workNew capabilities, enhancements, or enablement workPoRPlan of record item for roadmap and release trackingPlan of record item for roadmap and release trackingtrackingTracking issue for an ongoing project with smaller stepsTracking issue for an ongoing project with smaller stepsStatus: Open.#5148 In NVIDIA-NeMo/Megatron-Bridge;[feature] Support SequentialMLP (non-fused MoE) AutoBridge Mappings for P0 models
area:modelModel implementations and HF bridge logicModel implementations and HF bridge logicfeatureNew capabilities, enhancements, or enablement workNew capabilities, enhancements, or enablement workneeds-triageNew item needs classification and ownershipNew item needs classification and ownershipStatus: Open.#5145 In NVIDIA-NeMo/Megatron-Bridge;[NeMo FW 26.08 Release] MBridge 0.6.0 Roadmap
area:miscCross-cutting utilities, logging, helpers, and other changesCross-cutting utilities, logging, helpers, and other changesfeatureNew capabilities, enhancements, or enablement workNew capabilities, enhancements, or enablement worktrackingTracking issue for an ongoing project with smaller stepsTracking issue for an ongoing project with smaller stepsStatus: Open.#5144 In NVIDIA-NeMo/Megatron-Bridge;[bug] Packed SFT data prep ignores CP padding options, blend weights, and consecutive turns
area:dataDataset builders, preprocessing, and samplersDataset builders, preprocessing, and samplersbugSomething isn't workingSomething isn't workingtrackingTracking issue for an ongoing project with smaller stepsTracking issue for an ongoing project with smaller stepsStatus: Open.#5080 In NVIDIA-NeMo/Megatron-Bridge;
