Megatron Bridge
Table of Contents
Performance
Guides
Bridge with 🤗 Hugging Face
Supported Models
Training and Customization
Model Optimization with ModelOpt
Development
bridge
bridge.recipes
bridge.recipes.qwen2_audio
bridge.recipes.stepfun
bridge.recipes.qwen_omni
bridge.recipes.wan
bridge.recipes.kimi_vl
bridge.recipes.moonlight
bridge.recipes.glm
bridge.recipes.deepseek
bridge.recipes.qwen
bridge.recipes.nemotron_omni
bridge.recipes.gemma3_vl
bridge.recipes.gemma4_vl
bridge.recipes.olmoe
bridge.recipes.utils
bridge.recipes.gemma
bridge.recipes.gpt
bridge.recipes.llama
bridge.recipes.ministral3
bridge.recipes.qwen_vl
bridge.recipes.nemotron_vl
bridge.recipes.glm_vl
bridge.recipes.flux
bridge.recipes.nemotronh
bridge.recipes.gpt_oss
bridge.recipes.common
bridge.recipes.run_plugins
bridge.diffusion
bridge.diffusion.recipes
bridge.diffusion.conversion
bridge.diffusion.common
bridge.diffusion.base
bridge.diffusion.data
bridge.diffusion.models
bridge.utils
bridge.utils.slurm_utils
bridge.utils.import_utils
bridge.utils.cuda_graph
bridge.utils.safe_pickle
bridge.utils.decorators
bridge.utils.common_utils
bridge.utils.safe_url
bridge.utils.yaml_utils
bridge.utils.instantiate_utils
bridge.utils.activation_map
bridge.utils.vocab_utils
bridge.utils.fusions
bridge.peft
bridge.peft.adapter_wrapper
bridge.peft.dora_layers
bridge.peft.recompute
bridge.peft.lora
bridge.peft.lora_layers
bridge.peft.dora
bridge.peft.module_matcher
bridge.peft.base
bridge.peft.canonical_lora
bridge.peft.utils
bridge.peft.walk_utils
bridge.data
bridge.data.builders
bridge.data.hf_processors
bridge.data.datasets
bridge.data.energon
bridge.data.megatron_mimo
bridge.data.vlm_datasets
bridge.data.iterator_utils
bridge.data.loaders
bridge.data.finetuning
bridge.data.utils
bridge.data.samplers
bridge.training
bridge.training.post_training
bridge.training.tokenizers
bridge.training.utils
bridge.training.mlm_compat
bridge.training.gpt_step
bridge.training.optim
bridge.training.pretrain_megatron_mimo
bridge.training.inprocess_restart
bridge.training.eval_context_parallel_rebinding
bridge.training.llava_step
bridge.training.state
bridge.training.fault_tolerance
bridge.training.nemotron_omni_step
bridge.training.megatron_mimo_step
bridge.training.pretrain
bridge.training.forward_step_func_types
bridge.training.losses
bridge.training.setup
bridge.training.mixed_precision
bridge.training.audio_lm_step
bridge.training.tensor_inspect
bridge.training.checkpointing
bridge.training.vlm_step
bridge.training.callbacks
bridge.training.flex_dispatcher_backend
bridge.training.model_load_save
bridge.training.train_megatron_mimo
bridge.training.setup_megatron_mimo
bridge.training.initialize
bridge.training.distill
bridge.training.finetune
bridge.training.config
bridge.training.nvrx_straggler
bridge.training.megatron_mimo_parallel_utils
bridge.training.train
bridge.training.comm_overlap
bridge.training.eval
bridge.training.profiling
bridge.models
bridge.models.glm_moe_dsa
bridge.models.qwen3_asr
bridge.models.gemma_vl
bridge.models.bailing
bridge.models.sarvam
bridge.models.stepfun
bridge.models.qwen_omni
bridge.models.kimi_vl
bridge.models.glm
bridge.models.minimax_m2
bridge.models.deepseek
bridge.models.mimo
bridge.models.qwen
bridge.models.nemotron_omni
bridge.models.llama_nemotron
bridge.models.qwen_audio
bridge.models.olmoe
bridge.models.mamba
bridge.models.nemotron
bridge.models.conversion
bridge.models.kimi
bridge.models.hf_pretrained
bridge.models.gemma
bridge.models.falcon_h1
bridge.models.decorators
bridge.models.mistral
bridge.models.gpt
bridge.models.llama
bridge.models.common
bridge.models.ministral3
bridge.models.megatron_mimo
bridge.models.qwen_vl
bridge.models.nemotron_vl
bridge.models.glm_vl
bridge.models.nemotronh
bridge.models.gpt_oss
bridge.models.gpt_provider
bridge.models.t5_provider
bridge.models.gpt_full_te_layer_autocast_spec
bridge.models.transformer_config
bridge.models.mla_provider
bridge.models.model_provider
bridge.models.config
bridge.models.distillation_provider
bridge.package_info
Releases
Agent Skills
Directory Readme Files
GPT-OSS model documentation is organized by model variant.
Variant
Guide
GPT OSS
gpt-oss.md
