LICENSE
README.md
pyproject.toml
assets/version.txt
torchtitan/__init__.py
torchtitan/train.py
torchtitan/trainer.py
torchtitan.egg-info/PKG-INFO
torchtitan.egg-info/SOURCES.txt
torchtitan.egg-info/dependency_links.txt
torchtitan.egg-info/requires.txt
torchtitan.egg-info/top_level.txt
torchtitan/components/checkpoint.py
torchtitan/components/checkpoint_utils.py
torchtitan/components/dataloader.py
torchtitan/components/lora.py
torchtitan/components/loss.py
torchtitan/components/metrics.py
torchtitan/components/tokenizer.py
torchtitan/components/validate.py
torchtitan/components/checkpointer/__init__.py
torchtitan/components/checkpointer/base.py
torchtitan/components/checkpointer/dcp.py
torchtitan/components/checkpointer/torch_checkpointing.py
torchtitan/components/checkpointer/utils.py
torchtitan/components/optimizer/__init__.py
torchtitan/components/optimizer/lr_scheduler.py
torchtitan/components/optimizer/optimizer.py
torchtitan/components/optimizer/utils.py
torchtitan/components/quantization/__init__.py
torchtitan/components/quantization/float8.py
torchtitan/components/quantization/mx.py
torchtitan/components/quantization/nvfp4.py
torchtitan/components/quantization/utils.py
torchtitan/config/__init__.py
torchtitan/config/configs.py
torchtitan/config/configurable.py
torchtitan/config/function.py
torchtitan/config/manager.py
torchtitan/config/override.py
torchtitan/distributed/__init__.py
torchtitan/distributed/activation_checkpoint.py
torchtitan/distributed/compile.py
torchtitan/distributed/cudagraph.py
torchtitan/distributed/fsdp.py
torchtitan/distributed/full_dtensor.py
torchtitan/distributed/linear.py
torchtitan/distributed/parallel_dims.py
torchtitan/distributed/pipeline_parallel.py
torchtitan/distributed/spmd_types.py
torchtitan/distributed/tensor_parallel.py
torchtitan/distributed/utils.py
torchtitan/distributed/context_parallel/__init__.py
torchtitan/distributed/context_parallel/api.py
torchtitan/distributed/deepep/deepep.py
torchtitan/distributed/deepep/hybridep.py
torchtitan/distributed/flex_shard/__init__.py
torchtitan/distributed/flex_shard/_optimizer_reshard_runtime.py
torchtitan/distributed/flex_shard/_optimizer_reshard_schedule.py
torchtitan/distributed/flex_shard/dist_muon.py
torchtitan/distributed/flex_shard/optimizer_reshard.py
torchtitan/distributed/minimal_async_ep/__init__.py
torchtitan/distributed/minimal_async_ep/api.py
torchtitan/distributed/minimal_async_ep/kernels.py
torchtitan/experiments/__init__.py
torchtitan/experiments/forge/__init__.py
torchtitan/experiments/forge/engine.py
torchtitan/experiments/forge/example_train.py
torchtitan/experiments/graph_trainer/__init__.py
torchtitan/experiments/graph_trainer/autoparallel_api.py
torchtitan/experiments/graph_trainer/chunked_loss.py
torchtitan/experiments/graph_trainer/common_utils.py
torchtitan/experiments/graph_trainer/compile.py
torchtitan/experiments/graph_trainer/configs.py
torchtitan/experiments/graph_trainer/cpu_offload.py
torchtitan/experiments/graph_trainer/cudagraph.py
torchtitan/experiments/graph_trainer/custom_codegen.py
torchtitan/experiments/graph_trainer/debug_utils.py
torchtitan/experiments/graph_trainer/dynamic_shapes.py
torchtitan/experiments/graph_trainer/ep_chunk_pass.py
torchtitan/experiments/graph_trainer/ep_eager_chunk.py
torchtitan/experiments/graph_trainer/ep_overlap_pass.py
torchtitan/experiments/graph_trainer/ep_pass_utils.py
torchtitan/experiments/graph_trainer/ep_process_group_pass.py
torchtitan/experiments/graph_trainer/fsdp_passes.py
torchtitan/experiments/graph_trainer/fsdp_patterns.py
torchtitan/experiments/graph_trainer/graph_utils.py
torchtitan/experiments/graph_trainer/inductor_passes.py
torchtitan/experiments/graph_trainer/jit_backend.py
torchtitan/experiments/graph_trainer/log_activation_memory_policy.py
torchtitan/experiments/graph_trainer/make_fx_tracer.py
torchtitan/experiments/graph_trainer/memory_policy.py
torchtitan/experiments/graph_trainer/passes.py
torchtitan/experiments/graph_trainer/performance_passes.py
torchtitan/experiments/graph_trainer/precompile.py
torchtitan/experiments/graph_trainer/precompile_main.py
torchtitan/experiments/graph_trainer/registry.py
torchtitan/experiments/graph_trainer/remove_noop_passes.py
torchtitan/experiments/graph_trainer/selective_activation_remat.py
torchtitan/experiments/graph_trainer/simple_fsdp.py
torchtitan/experiments/graph_trainer/storage.py
torchtitan/experiments/graph_trainer/trainer.py
torchtitan/experiments/graph_trainer/deepseek_v3/__init__.py
torchtitan/experiments/graph_trainer/deepseek_v3/config_registry.py
torchtitan/experiments/graph_trainer/deepseek_v3/model.py
torchtitan/experiments/graph_trainer/deepseek_v3/parallelize.py
torchtitan/experiments/graph_trainer/deepseek_v3/parallelize_autoparallel.py
torchtitan/experiments/graph_trainer/graph_pp/__init__.py
torchtitan/experiments/graph_trainer/graph_pp/graph_builder.py
torchtitan/experiments/graph_trainer/graph_pp/graph_multiplex.py
torchtitan/experiments/graph_trainer/graph_pp/partition.py
torchtitan/experiments/graph_trainer/graph_pp/pipeline.py
torchtitan/experiments/graph_trainer/graph_pp/runner.py
torchtitan/experiments/graph_trainer/graph_pp/split_di_dw.py
torchtitan/experiments/graph_trainer/graph_pp/split_fsdp_collectives.py
torchtitan/experiments/graph_trainer/graph_pp/stage.py
torchtitan/experiments/graph_trainer/graph_pp/utils.py
torchtitan/experiments/graph_trainer/llama3/__init__.py
torchtitan/experiments/graph_trainer/llama3/config_registry.py
torchtitan/experiments/graph_trainer/llama3/model.py
torchtitan/experiments/graph_trainer/llama3/parallelize.py
torchtitan/experiments/graph_trainer/llama3/parallelize_autoparallel.py
torchtitan/experiments/graph_trainer/muse_glimmer/__init__.py
torchtitan/experiments/graph_trainer/muse_glimmer/config_registry.py
torchtitan/experiments/graph_trainer/muse_glimmer/model.py
torchtitan/experiments/graph_trainer/muse_glimmer/parallelize.py
torchtitan/experiments/graph_trainer/qwen3/__init__.py
torchtitan/experiments/graph_trainer/qwen3/config_registry.py
torchtitan/experiments/graph_trainer/qwen3/model.py
torchtitan/experiments/graph_trainer/qwen3/parallelize.py
torchtitan/experiments/graph_trainer/tests/__init__.py
torchtitan/experiments/graph_trainer/tests/_trainer_test_utils.py
torchtitan/experiments/graph_trainer/tests/integration_tests.py
torchtitan/experiments/graph_trainer/tests/run_precompile_tests.py
torchtitan/experiments/graph_trainer/tests/test_autoparallel.py
torchtitan/experiments/graph_trainer/tests/test_bitwise_deterministic.py
torchtitan/experiments/graph_trainer/tests/test_chunked_loss.py
torchtitan/experiments/graph_trainer/tests/test_cpu_offload.py
torchtitan/experiments/graph_trainer/tests/test_custom_codegen.py
torchtitan/experiments/graph_trainer/tests/test_graph_pp_passes.py
torchtitan/experiments/graph_trainer/tests/test_graph_pp_runner.py
torchtitan/experiments/graph_trainer/tests/test_numerics.py
torchtitan/experiments/graph_trainer/tests/test_passes.py
torchtitan/experiments/graph_trainer/tests/test_performance_passes.py
torchtitan/experiments/graph_trainer/tests/test_precompile.py
torchtitan/experiments/graph_trainer/tests/test_profiler.py
torchtitan/experiments/graph_trainer/tests/test_sac_peak_memory.py
torchtitan/experiments/graph_trainer/tests/test_simple_fsdp.py
torchtitan/experiments/graph_trainer/tests/test_trace_module.py
torchtitan/experiments/rl/__init__.py
torchtitan/experiments/rl/batch_invariance.py
torchtitan/experiments/rl/controller.py
torchtitan/experiments/rl/controller_metrics.py
torchtitan/experiments/rl/generate.py
torchtitan/experiments/rl/renderer.py
torchtitan/experiments/rl/rollout_recorder.py
torchtitan/experiments/rl/train.py
torchtitan/experiments/rl/types.py
torchtitan/experiments/rl/actors/generator.py
torchtitan/experiments/rl/actors/trainer.py
torchtitan/experiments/rl/components/__init__.py
torchtitan/experiments/rl/components/batcher.py
torchtitan/experiments/rl/components/training_sample_builder.py
torchtitan/experiments/rl/components/weight_sync.py
torchtitan/experiments/rl/components/work_buffer.py
torchtitan/experiments/rl/environment/__init__.py
torchtitan/experiments/rl/environment/message.py
torchtitan/experiments/rl/environment/token.py
torchtitan/experiments/rl/examples/__init__.py
torchtitan/experiments/rl/examples/alphabet_sort/__init__.py
torchtitan/experiments/rl/examples/alphabet_sort/config_registry.py
torchtitan/experiments/rl/examples/alphabet_sort/data.py
torchtitan/experiments/rl/examples/alphabet_sort/env.py
torchtitan/experiments/rl/examples/alphabet_sort/rollouter.py
torchtitan/experiments/rl/examples/alphabet_sort/rubric.py
torchtitan/experiments/rl/examples/dapo_math/__init__.py
torchtitan/experiments/rl/examples/dapo_math/config_registry.py
torchtitan/experiments/rl/examples/dapo_math/data.py
torchtitan/experiments/rl/examples/dapo_math/env.py
torchtitan/experiments/rl/examples/dapo_math/rollouter.py
torchtitan/experiments/rl/examples/dapo_math/rubric.py
torchtitan/experiments/rl/examples/search_r1/__init__.py
torchtitan/experiments/rl/examples/search_r1/config_registry.py
torchtitan/experiments/rl/examples/search_r1/data.py
torchtitan/experiments/rl/examples/search_r1/env.py
torchtitan/experiments/rl/examples/search_r1/rollouter.py
torchtitan/experiments/rl/examples/search_r1/rubric.py
torchtitan/experiments/rl/losses/__init__.py
torchtitan/experiments/rl/losses/dapo.py
torchtitan/experiments/rl/losses/grpo.py
torchtitan/experiments/rl/models/attention.py
torchtitan/experiments/rl/models/cast_linear.py
torchtitan/experiments/rl/models/vllm_registry.py
torchtitan/experiments/rl/models/vllm_worker.py
torchtitan/experiments/rl/models/vllm_wrapper.py
torchtitan/experiments/rl/observability/__init__.py
torchtitan/experiments/rl/observability/metrics/__init__.py
torchtitan/experiments/rl/observability/metrics/console.py
torchtitan/experiments/rl/observability/metrics/processor.py
torchtitan/experiments/rl/observability/metrics/types.py
torchtitan/experiments/rl/rollout/__init__.py
torchtitan/experiments/rl/rollout/advantage.py
torchtitan/experiments/rl/rollout/rollouter.py
torchtitan/experiments/rl/rollout/types.py
torchtitan/experiments/rl/routing/__init__.py
torchtitan/experiments/rl/routing/inter_generator_router.py
torchtitan/experiments/rl/routing/intra_generator_router.py
torchtitan/experiments/rl/routing/strategies.py
torchtitan/experiments/rl/routing/types.py
torchtitan/experiments/rl/rubrics/__init__.py
torchtitan/experiments/rl/rubrics/rubric.py
torchtitan/experiments/rl/scripts/__init__.py
torchtitan/experiments/rl/scripts/loss_compare.py
torchtitan/experiments/rl/tests/__init__.py
torchtitan/experiments/rl/tests/integration_tests.py
torchtitan/experiments/rl/tests/test_alphabet_sort.py
torchtitan/experiments/rl/tests/test_async_controller.py
torchtitan/experiments/rl/tests/test_bitwise_parity.py
torchtitan/experiments/rl/tests/test_cast_linear.py
torchtitan/experiments/rl/tests/test_dapo_math.py
torchtitan/experiments/rl/tests/test_engine_loop.py
torchtitan/experiments/rl/tests/test_generator.py
torchtitan/experiments/rl/tests/test_inter_generator_router.py
torchtitan/experiments/rl/tests/test_intra_generator_router.py
torchtitan/experiments/rl/tests/test_metrics.py
torchtitan/experiments/rl/tests/test_rollout_recorder.py
torchtitan/experiments/rl/tests/test_rollout_utils.py
torchtitan/experiments/rl/tests/test_search_r1.py
torchtitan/experiments/rl/tests/test_shutdown.py
torchtitan/experiments/rl/tests/test_train_loop.py
torchtitan/experiments/rl/tests/test_weight_sync.py
torchtitan/experiments/torchft/__init__.py
torchtitan/experiments/torchft/checkpoint.py
torchtitan/experiments/torchft/manager.py
torchtitan/experiments/torchft/optimizer.py
torchtitan/experiments/torchft/trainer.py
torchtitan/experiments/torchft/config/__init__.py
torchtitan/experiments/torchft/config/job_config.py
torchtitan/experiments/torchft/diloco/__init__.py
torchtitan/experiments/torchft/diloco/utils.py
torchtitan/experiments/torchft/llama3/__init__.py
torchtitan/experiments/torchft/llama3/config_registry.py
torchtitan/experiments/torchft/tests/__init__.py
torchtitan/experiments/torchft/tests/integration_tests.py
torchtitan/experiments/torchft/tests/test_torchft_checkpoint.py
torchtitan/experiments/transformers_modeling_backend/__init__.py
torchtitan/experiments/transformers_modeling_backend/config_registry.py
torchtitan/experiments/transformers_modeling_backend/configs.py
torchtitan/experiments/transformers_modeling_backend/hf_sharding.py
torchtitan/experiments/transformers_modeling_backend/model.py
torchtitan/experiments/transformers_modeling_backend/module_conversion.py
torchtitan/experiments/transformers_modeling_backend/moe_replacement.py
torchtitan/experiments/transformers_modeling_backend/parallelize.py
torchtitan/experiments/transformers_modeling_backend/pipeline.py
torchtitan/experiments/transformers_modeling_backend/state_dict_adapter.py
torchtitan/experiments/transformers_modeling_backend/tokenizer.py
torchtitan/experiments/transformers_modeling_backend/trainer.py
torchtitan/experiments/transformers_modeling_backend/tests/__init__.py
torchtitan/experiments/transformers_modeling_backend/tests/cp_pp_numerical.py
torchtitan/experiments/transformers_modeling_backend/tests/integration_tests.py
torchtitan/experiments/transformers_modeling_backend/tests/test_flex_attention.py
torchtitan/experiments/transformers_modeling_backend/tests/test_flex_cp_numerical.py
torchtitan/experiments/transformers_modeling_backend/tests/test_moe_parallelism.py
torchtitan/hf_datasets/__init__.py
torchtitan/hf_datasets/interleaved.py
torchtitan/hf_datasets/text_datasets.py
torchtitan/hf_datasets/multimodal/__init__.py
torchtitan/hf_datasets/multimodal/mm_collator.py
torchtitan/hf_datasets/multimodal/mm_datasets.py
torchtitan/hf_datasets/multimodal/utils/image.py
torchtitan/hf_datasets/multimodal/utils/packing.py
torchtitan/hf_datasets/multimodal/utils/text.py
torchtitan/hf_datasets/multimodal/utils/video.py
torchtitan/models/__init__.py
torchtitan/models/utils.py
torchtitan/models/common/__init__.py
torchtitan/models/common/attention.py
torchtitan/models/common/config_utils.py
torchtitan/models/common/decoder.py
torchtitan/models/common/decoder_sharding.py
torchtitan/models/common/dist_gemm.py
torchtitan/models/common/embedding.py
torchtitan/models/common/feed_forward.py
torchtitan/models/common/linear.py
torchtitan/models/common/moe.py
torchtitan/models/common/moe_sharding.py
torchtitan/models/common/multimodal.py
torchtitan/models/common/nn_modules.py
torchtitan/models/common/param_init.py
torchtitan/models/common/rope.py
torchtitan/models/common/token_dispatcher.py
torchtitan/models/common/vision_encoder.py
torchtitan/models/common/vision_encoder_sharding.py
torchtitan/models/deepseek_v3/__init__.py
torchtitan/models/deepseek_v3/config_registry.py
torchtitan/models/deepseek_v3/model.py
torchtitan/models/deepseek_v3/mtp.py
torchtitan/models/deepseek_v3/parallelize.py
torchtitan/models/deepseek_v3/sharding.py
torchtitan/models/deepseek_v3/state_dict_adapter.py
torchtitan/models/flux/__init__.py
torchtitan/models/flux/config_registry.py
torchtitan/models/flux/configs.py
torchtitan/models/flux/flux_datasets.py
torchtitan/models/flux/parallelize.py
torchtitan/models/flux/sharding.py
torchtitan/models/flux/tokenizer.py
torchtitan/models/flux/trainer.py
torchtitan/models/flux/utils.py
torchtitan/models/flux/validate.py
torchtitan/models/flux/inference/infer.py
torchtitan/models/flux/inference/sampling.py
torchtitan/models/flux/model/autoencoder.py
torchtitan/models/flux/model/hf_embedder.py
torchtitan/models/flux/model/layers.py
torchtitan/models/flux/model/model.py
torchtitan/models/flux/model/state_dict_adapter.py
torchtitan/models/gpt_oss/__init__.py
torchtitan/models/gpt_oss/config_registry.py
torchtitan/models/gpt_oss/model.py
torchtitan/models/gpt_oss/moe.py
torchtitan/models/gpt_oss/parallelize.py
torchtitan/models/gpt_oss/sharding.py
torchtitan/models/gpt_oss/state_dict_adapter.py
torchtitan/models/kimi_k2_7/__init__.py
torchtitan/models/kimi_k2_7/config_registry.py
torchtitan/models/kimi_k2_7/model.py
torchtitan/models/kimi_k2_7/parallelize.py
torchtitan/models/kimi_k2_7/sharding.py
torchtitan/models/kimi_k2_7/state_dict_adapter.py
torchtitan/models/kimi_k2_7/vision_encoder.py
torchtitan/models/llama3/__init__.py
torchtitan/models/llama3/config_registry.py
torchtitan/models/llama3/model.py
torchtitan/models/llama3/parallelize.py
torchtitan/models/llama3/sharding.py
torchtitan/models/llama3/state_dict_adapter.py
torchtitan/models/muse_glimmer/__init__.py
torchtitan/models/muse_glimmer/config_registry.py
torchtitan/models/muse_glimmer/model.py
torchtitan/models/muse_glimmer/parallelize.py
torchtitan/models/muse_glimmer/sharding.py
torchtitan/models/muse_glimmer/state_dict_adapter.py
torchtitan/models/muse_glimmer/vision_encoder.py
torchtitan/models/qwen3/__init__.py
torchtitan/models/qwen3/config_registry.py
torchtitan/models/qwen3/model.py
torchtitan/models/qwen3/parallelize.py
torchtitan/models/qwen3/sharding.py
torchtitan/models/qwen3/state_dict_adapter.py
torchtitan/models/qwen3_5/__init__.py
torchtitan/models/qwen3_5/config_registry.py
torchtitan/models/qwen3_5/model.py
torchtitan/models/qwen3_5/parallelize.py
torchtitan/models/qwen3_5/rope.py
torchtitan/models/qwen3_5/sharding.py
torchtitan/models/qwen3_5/state_dict_adapter.py
torchtitan/models/qwen3_5/vision_encoder.py
torchtitan/observability/__init__.py
torchtitan/observability/structured_logger/__init__.py
torchtitan/observability/structured_logger/gantt_generator.py
torchtitan/observability/structured_logger/jsonl_handler.py
torchtitan/observability/structured_logger/step_state.py
torchtitan/observability/structured_logger/structured_logging.py
torchtitan/ops/__init__.py
torchtitan/ops/scatter_add.py
torchtitan/overrides/__init__.py
torchtitan/overrides/fused_mla.py
torchtitan/overrides/fused_swiglu.py
torchtitan/overrides/helion_rope.py
torchtitan/overrides/moe_token_dispatcher.py
torchtitan/protocols/__init__.py
torchtitan/protocols/model.py
torchtitan/protocols/model_spec.py
torchtitan/protocols/module.py
torchtitan/protocols/sharding.py
torchtitan/protocols/state_dict_adapter.py
torchtitan/tools/filesystem.py
torchtitan/tools/logging.py
torchtitan/tools/profiler.py
torchtitan/tools/utils.py
torchtitan_recipes/__init__.py
torchtitan_recipes/tests/__init__.py
torchtitan_recipes/tests/features.py
torchtitan_recipes/tests/flux.py
torchtitan_recipes/tests/h100.py
torchtitan_recipes/tests/models.py
torchtitan_recipes/tests/transformers_modeling_backend.py