Python Module Index

c


	c
	`core`
	`core._rank_utils`
	`core.activations`
	`core.config`
	`core.config_logger`
	`core.datasets`
	`core.datasets.bert_dataset`
	`core.datasets.blended_dataset`
	`core.datasets.blended_megatron_dataset_builder`
	`core.datasets.blended_megatron_dataset_config`
	`core.datasets.data_schedule`
	`core.datasets.gpt_dataset`
	`core.datasets.helpers`
	`core.datasets.indexed_dataset`
	`core.datasets.masked_dataset`
	`core.datasets.megatron_dataset`
	`core.datasets.multimodal_dataset`
	`core.datasets.object_storage_utils`
	`core.datasets.t5_dataset`
	`core.datasets.utils`
	`core.datasets.utils_s3`
	`core.dist_checkpointing`
	`core.dist_checkpointing.core`
	`core.dist_checkpointing.dict_utils`
	`core.dist_checkpointing.exchange_utils`
	`core.dist_checkpointing.mapping`
	`core.dist_checkpointing.optimizer`
	`core.dist_checkpointing.serialization`
	`core.dist_checkpointing.state_dict_utils`
	`core.dist_checkpointing.strategies`
	`core.dist_checkpointing.strategies.async_utils`
	`core.dist_checkpointing.strategies.base`
	`core.dist_checkpointing.strategies.cached_metadata_filesystem_reader`
	`core.dist_checkpointing.strategies.checkpointable`
	`core.dist_checkpointing.strategies.common`
	`core.dist_checkpointing.strategies.filesystem_async`
	`core.dist_checkpointing.strategies.fully_parallel`
	`core.dist_checkpointing.strategies.state_dict_saver`
	`core.dist_checkpointing.strategies.torch`
	`core.dist_checkpointing.tensor_aware_state_dict`
	`core.dist_checkpointing.utils`
	`core.dist_checkpointing.validation`
	`core.distributed`
	`core.distributed.data_parallel_base`
	`core.distributed.distributed_data_parallel`
	`core.distributed.distributed_data_parallel_config`
	`core.distributed.finalize_model_grads`
	`core.distributed.fsdp`
	`core.distributed.fsdp.mcore_fsdp_adapter`
	`core.distributed.fsdp.src`
	`core.distributed.fsdp.src.megatron_fsdp`
	`core.distributed.fsdp.src.megatron_fsdp.distributed_data_parallel_config`
	`core.distributed.fsdp.src.megatron_fsdp.fully_shard`
	`core.distributed.fsdp.src.megatron_fsdp.megatron_fsdp`
	`core.distributed.fsdp.src.megatron_fsdp.mixed_precision`
	`core.distributed.fsdp.src.megatron_fsdp.package_info`
	`core.distributed.fsdp.src.megatron_fsdp.param_and_grad_buffer`
	`core.distributed.fsdp.src.megatron_fsdp.uneven_dtensor`
	`core.distributed.fsdp.src.megatron_fsdp.utils`
	`core.distributed.param_and_grad_buffer`
	`core.distributed.reduce_scatter_with_fp32_accumulation`
	`core.distributed.torch_fully_sharded_data_parallel`
	`core.distributed.torch_fully_sharded_data_parallel_config`
	`core.energy_monitor`
	`core.enums`
	`core.export`
	`core.export.data_type`
	`core.export.export_config`
	`core.export.model_type`
	`core.export.trtllm`
	`core.export.trtllm.engine_builder`
	`core.export.trtllm.engine_builder.trtllm_engine_builder`
	`core.export.trtllm.model_to_trllm_mapping`
	`core.export.trtllm.model_to_trllm_mapping.default_conversion_dict`
	`core.export.trtllm.trt_model_config`
	`core.export.trtllm.trt_model_type`
	`core.export.trtllm.trtllm_helper`
	`core.export.trtllm.trtllm_layers`
	`core.export.trtllm.trtllm_weights_converter`
	`core.export.trtllm.trtllm_weights_converter.distributed_trtllm_model_weights_converter`
	`core.export.trtllm.trtllm_weights_converter.single_device_trtllm_model_weights_converter`
	`core.export.trtllm.trtllm_weights_converter.utils`
	`core.extensions`
	`core.extensions.kitchen`
	`core.extensions.transformer_engine`
	`core.extensions.transformer_engine_spec_provider`
	`core.fp4_utils`
	`core.fp8_utils`
	`core.full_cuda_graph`
	`core.fusions`
	`core.fusions.fused_bias_dropout`
	`core.fusions.fused_bias_geglu`
	`core.fusions.fused_bias_gelu`
	`core.fusions.fused_bias_swiglu`
	`core.fusions.fused_cross_entropy`
	`core.fusions.fused_indices_converter`
	`core.fusions.fused_layer_norm`
	`core.fusions.fused_mla_yarn_rope_apply`
	`core.fusions.fused_pad_routing_map`
	`core.fusions.fused_softmax`
	`core.fusions.fused_weighted_squared_relu`
	`core.hyper_comm_grid`
	`core.inference`
	`core.inference.async_stream`
	`core.inference.batch_dimensions_utils`
	`core.inference.common_inference_params`
	`core.inference.communication_utils`
	`core.inference.config`
	`core.inference.contexts`
	`core.inference.contexts.base_context`
	`core.inference.contexts.dynamic_block_allocator`
	`core.inference.contexts.dynamic_context`
	`core.inference.contexts.fused_kv_append_kernel`
	`core.inference.contexts.routing_metadata`
	`core.inference.contexts.static_context`
	`core.inference.data_parallel_inference_coordinator`
	`core.inference.engines`
	`core.inference.engines.abstract_engine`
	`core.inference.engines.async_zmq_communicator`
	`core.inference.engines.dynamic_engine`
	`core.inference.engines.mcore_engine`
	`core.inference.engines.static_engine`
	`core.inference.headers`
	`core.inference.inference_client`
	`core.inference.inference_request`
	`core.inference.model_inference_wrappers`
	`core.inference.model_inference_wrappers.abstract_model_inference_wrapper`
	`core.inference.model_inference_wrappers.gpt`
	`core.inference.model_inference_wrappers.gpt.gpt_inference_wrapper`
	`core.inference.model_inference_wrappers.t5`
	`core.inference.model_inference_wrappers.t5.t5_inference_wrapper`
	`core.inference.sampling_params`
	`core.inference.scheduler`
	`core.inference.text_generation_controllers`
	`core.inference.text_generation_controllers.encoder_decoder_text_generation_controller`
	`core.inference.text_generation_controllers.text_generation_controller`
	`core.inference.text_generation_controllers.vlm_text_generation_controller`
	`core.inference.text_generation_server`
	`core.inference.text_generation_server.dynamic_text_gen_server`
	`core.inference.text_generation_server.dynamic_text_gen_server.endpoints`
	`core.inference.text_generation_server.dynamic_text_gen_server.endpoints.chat_completions`
	`core.inference.text_generation_server.dynamic_text_gen_server.endpoints.common`
	`core.inference.text_generation_server.dynamic_text_gen_server.endpoints.completions`
	`core.inference.text_generation_server.dynamic_text_gen_server.endpoints.health`
	`core.inference.text_generation_server.dynamic_text_gen_server.flask_server`
	`core.inference.text_generation_server.dynamic_text_gen_server.tokenization`
	`core.inference.text_generation_server.run_mcore_engine`
	`core.inference.text_generation_server.text_generation_server`
	`core.inference.text_generation_server.tokenization`
	`core.inference.unified_memory`
	`core.inference.utils`
	`core.inference_params`
	`core.jit`
	`core.model_parallel_config`
	`core.models`
	`core.models.backends`
	`core.models.bert`
	`core.models.bert.bert_layer_specs`
	`core.models.bert.bert_lm_head`
	`core.models.bert.bert_model`
	`core.models.bert.pooler`
	`core.models.common`
	`core.models.common.embeddings`
	`core.models.common.embeddings.language_model_embedding`
	`core.models.common.embeddings.relative_pos_embedding`
	`core.models.common.embeddings.rope_utils`
	`core.models.common.embeddings.rotary_pos_embedding`
	`core.models.common.embeddings.yarn_rotary_pos_embedding`
	`core.models.common.language_module`
	`core.models.common.language_module.language_module`
	`core.models.common.model_chunk_schedule_plan`
	`core.models.common.vision_module`
	`core.models.common.vision_module.vision_module`
	`core.models.gpt`
	`core.models.gpt.experimental_attention_variant_module_specs`
	`core.models.gpt.fine_grained_callables`
	`core.models.gpt.gpt_layer_specs`
	`core.models.gpt.gpt_model`
	`core.models.gpt.moe_module_specs`
	`core.models.huggingface`
	`core.models.huggingface.clip_model`
	`core.models.huggingface.module`
	`core.models.huggingface.qwen_model`
	`core.models.mamba`
	`core.models.mamba.mamba_layer_specs`
	`core.models.mamba.mamba_model`
	`core.models.mimo`
	`core.models.mimo.config`
	`core.models.mimo.config.base_configs`
	`core.models.mimo.model`
	`core.models.mimo.model.base`
	`core.models.multimodal`
	`core.models.multimodal.context_parallel`
	`core.models.multimodal.llava_model`
	`core.models.multimodal.llava_spec`
	`core.models.T5`
	`core.models.T5.t5_model`
	`core.models.T5.t5_spec`
	`core.models.vision`
	`core.models.vision.clip_vit_model`
	`core.models.vision.multimodal_projector`
	`core.models.vision.radio`
	`core.models.vision.vit_layer_specs`
	`core.msc_utils`
	`core.nccl_allocator`
	`core.num_microbatches_calculator`
	`core.optimizer`
	`core.optimizer.clip_grads`
	`core.optimizer.cpu_offloading`
	`core.optimizer.cpu_offloading.hybrid_optimizer`
	`core.optimizer.distrib_optimizer`
	`core.optimizer.grad_scaler`
	`core.optimizer.layer_wise_optimizer`
	`core.optimizer.muon`
	`core.optimizer.optimizer`
	`core.optimizer.optimizer_config`
	`core.optimizer.qk_clip`
	`core.optimizer_param_scheduler`
	`core.package_info`
	`core.packed_seq_params`
	`core.parallel_state`
	`core.pipeline_parallel`
	`core.pipeline_parallel.bridge_communicator`
	`core.pipeline_parallel.combined_1f1b`
	`core.pipeline_parallel.fine_grained_activation_offload`
	`core.pipeline_parallel.hybrid_cp_schedule`
	`core.pipeline_parallel.multimodule_communicator`
	`core.pipeline_parallel.p2p_communication`
	`core.pipeline_parallel.schedules`
	`core.pipeline_parallel.utils`
	`core.post_training`
	`core.post_training.modelopt`
	`core.post_training.modelopt.gpt`
	`core.post_training.modelopt.gpt.model_specs`
	`core.post_training.modelopt.gpt.state_dict_hooks`
	`core.post_training.modelopt.layers`
	`core.post_training.modelopt.mamba`
	`core.post_training.modelopt.mamba.model_specs`
	`core.process_groups_config`
	`core.quantization`
	`core.quantization.quant_config`
	`core.quantization.utils`
	`core.rerun_state_machine`
	`core.resharding`
	`core.resharding.copy_services`
	`core.resharding.copy_services.base`
	`core.resharding.copy_services.gloo_copy_service`
	`core.resharding.copy_services.nccl_copy_service`
	`core.resharding.copy_services.nvshmem_copy_service`
	`core.resharding.execution`
	`core.resharding.nvshmem_copy_service`
	`core.resharding.nvshmem_copy_service.core`
	`core.resharding.nvshmem_copy_service.core.gpu_resource_manager`
	`core.resharding.nvshmem_copy_service.core.kernel_launcher`
	`core.resharding.nvshmem_copy_service.core.pipeline_executor`
	`core.resharding.nvshmem_copy_service.logger`
	`core.resharding.nvshmem_copy_service.memory`
	`core.resharding.nvshmem_copy_service.memory.double_buffer_manager`
	`core.resharding.nvshmem_copy_service.memory.tensor_pointer_utils`
	`core.resharding.nvshmem_copy_service.nvshmem_types`
	`core.resharding.nvshmem_copy_service.planning`
	`core.resharding.nvshmem_copy_service.planning.communication_scheduler`
	`core.resharding.nvshmem_copy_service.planning.gpu_execution_planner`
	`core.resharding.nvshmem_copy_service.planning.task_segmenter`
	`core.resharding.nvshmem_copy_service.planning.workload_packer`
	`core.resharding.nvshmem_copy_service.service`
	`core.resharding.nvshmem_copy_service.validation`
	`core.resharding.planner`
	`core.resharding.refit`
	`core.resharding.utils`
	`core.safe_globals`
	`core.ssm`
	`core.ssm.gated_delta_net`
	`core.ssm.mamba_block`
	`core.ssm.mamba_context_parallel`
	`core.ssm.mamba_hybrid_layer_allocation`
	`core.ssm.mamba_layer`
	`core.ssm.mamba_mixer`
	`core.ssm.mlp_layer`
	`core.ssm.triton_cache_manager`
	`core.tensor_parallel`
	`core.tensor_parallel.cross_entropy`
	`core.tensor_parallel.data`
	`core.tensor_parallel.inference_layers`
	`core.tensor_parallel.layers`
	`core.tensor_parallel.mappings`
	`core.tensor_parallel.random`
	`core.tensor_parallel.utils`
	`core.timers`
	`core.tokenizers`
	`core.tokenizers.base_tokenizer`
	`core.tokenizers.megatron_tokenizer`
	`core.tokenizers.text`
	`core.tokenizers.text.libraries`
	`core.tokenizers.text.libraries.abstract_tokenizer`
	`core.tokenizers.text.libraries.bytelevel_tokenizer`
	`core.tokenizers.text.libraries.chat_template`
	`core.tokenizers.text.libraries.huggingface_tokenizer`
	`core.tokenizers.text.libraries.megatron_hf_tokenizer`
	`core.tokenizers.text.libraries.null_tokenizer`
	`core.tokenizers.text.libraries.sentencepiece_tokenizer`
	`core.tokenizers.text.libraries.sft_tokenizer`
	`core.tokenizers.text.libraries.tiktoken_tokenizer`
	`core.tokenizers.text.models`
	`core.tokenizers.text.models.bert_tokenizer`
	`core.tokenizers.text.models.default_tokenizer`
	`core.tokenizers.text.models.gpt_tokenizer`
	`core.tokenizers.text.models.mamba_tokenizer`
	`core.tokenizers.text.models.t5_tokenizer`
	`core.tokenizers.text.text_tokenizer`
	`core.tokenizers.vision`
	`core.tokenizers.vision.libraries`
	`core.tokenizers.vision.libraries.multimodal_tokenizer`
	`core.tokenizers.vision.libraries.null_multimodal_tokenizer`
	`core.tokenizers.vision.models`
	`core.tokenizers.vision.models.default_tokenizer`
	`core.tokenizers.vision.vision_tokenizer`
	`core.transformer`
	`core.transformer.attention`
	`core.transformer.cuda_graphs`
	`core.transformer.custom_layers`
	`core.transformer.custom_layers.batch_invariant_kernels`
	`core.transformer.custom_layers.transformer_engine`
	`core.transformer.dot_product_attention`
	`core.transformer.enums`
	`core.transformer.fsdp_dtensor_checkpoint`
	`core.transformer.identity_op`
	`core.transformer.mlp`
	`core.transformer.module`
	`core.transformer.moe`
	`core.transformer.moe.experts`
	`core.transformer.moe.fused_a2a`
	`core.transformer.moe.grouped_gemm_util`
	`core.transformer.moe.moe_layer`
	`core.transformer.moe.moe_utils`
	`core.transformer.moe.router`
	`core.transformer.moe.router_replay`
	`core.transformer.moe.shared_experts`
	`core.transformer.moe.token_dispatcher`
	`core.transformer.moe.upcycling_utils`
	`core.transformer.multi_latent_attention`
	`core.transformer.multi_token_prediction`
	`core.transformer.pipeline_parallel_layer_layout`
	`core.transformer.spec_utils`
	`core.transformer.torch_layer_norm`
	`core.transformer.torch_norm`
	`core.transformer.transformer_block`
	`core.transformer.transformer_config`
	`core.transformer.transformer_layer`
	`core.transformer.utils`
	`core.typed_torch`
	`core.utils`